The Fuse AI MCP connector routes your AI agent's tool calls to Fuse AI's own MCP server through Scalekit. Each user signs in to Fuse AI once, and Scalekit stores and refreshes their tokens, so your agent never handles credentials. It comes with 305 tools.
- Tools
- 305
- What they doRead · write · destructive
- 114 · 12 · 179114 read12 write179 destructive
- Users sign in with
- OAuth
- OAuth app
- Your own Fuse AI MCP server app
Setup
Install the SDK
Terminal window npm install @scalekit-sdk/node dotenvTerminal window pip install scalekit-sdk-python python-dotenvSet your credentials
Add your Scalekit credentials to your
.envfile. Find values in app.scalekit.com > Developers > API Credentials..env SCALEKIT_ENVIRONMENT_URL=<your-environment-url>SCALEKIT_CLIENT_ID=<your-client-id>SCALEKIT_CLIENT_SECRET=<your-client-secret>Create the Fuse AI MCP connection
In AgentKit > Connections, create a Fuse AI MCP connection and copy its redirect URI. The name you give it is the
connection_nameyour code passes. See Configure connections.Register an OAuth app
Fuse AI MCP server connections use your own OAuth app. Register one with Fuse AI MCP server and add the redirect URI you copied.
Then enter the app's Client ID and Client Secret on the Fuse AI MCP connection.
Authorize a user and make your first call
quickstart.mts import { ScalekitClient } from '@scalekit-sdk/node'import 'dotenv/config'import { createInterface } from 'node:readline/promises'const scalekit = new ScalekitClient(process.env.SCALEKIT_ENVIRONMENT_URL,process.env.SCALEKIT_CLIENT_ID,process.env.SCALEKIT_CLIENT_SECRET,)const actions = scalekit.actionsconst connector = 'fuseaimcp'const identifier = 'user_123'// Generate an authorization link for the userconst { link } = await actions.getAuthorizationLink({ connectionName: connector, identifier })console.log('Authorize Fuse AI MCP:', link)const rl = createInterface({ input: process.stdin, output: process.stdout })await rl.question('Press Enter after authorizing...')rl.close()// Make your first callconst result = await actions.executeTool({connector,identifier,toolName: 'fuseaimcp_contact_call_engagement_summary',toolInput: {},})console.log(result)Terminal window npx tsx quickstart.mtsquickstart.py import osfrom scalekit import ScalekitClientfrom dotenv import load_dotenvload_dotenv()scalekit_client = ScalekitClient(env_url=os.getenv("SCALEKIT_ENVIRONMENT_URL"),client_id=os.getenv("SCALEKIT_CLIENT_ID"),client_secret=os.getenv("SCALEKIT_CLIENT_SECRET"),)actions = scalekit_client.actionsconnection_name = "fuseaimcp"identifier = "user_123"# Generate an authorization link for the userlink_response = actions.get_authorization_link(connection_name=connection_name,identifier=identifier,)print("Authorize Fuse AI MCP:", link_response.link)input("Press Enter after authorizing...")# Make your first callresult = actions.execute_tool(tool_input={},tool_name="fuseaimcp_contact_call_engagement_summary",connection_name=connection_name,identifier=identifier,)print(result)Terminal window python quickstart.pyEach user signs in once. See Authorize a user for the full flow and statuses.
Tools
Pass the exact name toexecute_toolfuseaimcp_add_spintax_to_text_toolRewrites a piece of email/message text with spintax {variation|variation} alternatives to improve deliverability.Read-onlyAdd Spintax To Text
Rewrites a piece of email/message text with spintax {variation|variation} alternatives to improve deliverability. Returns the spintax-enriched text without saving it; if the AI output is unusable or would alter merge fields, placeholders or existing spintax, the original text is returned unchanged.
Inputs
bodyTextstringrequired- The email/message text to enrich with spintax
fuseaimcp_build_prospect_search_filtersUse it first when the user asks FuseAI to find people or companies — find/see ('find me...', 'show me...'), 'create a list of...', 'add to my list...', and compound discovery+campaign requests.Read-onlyBuild Prospect Search Filters
Use it first when the user asks FuseAI to find people or companies — find/see ('find me...', 'show me...'), 'create a list of...', 'add to my list...', and compound discovery+campaign requests. Not for specific people the user names or that were already found ('add Steve, the CRO at Datadog, to my list', 'save these 3'): save those with save_people_to_list, which finds each one's real profile itself. Converts the request into Prospect Search filters, looking up the company names, job titles, places and similar values in the request with Fuse's external people and company data providers so the filters use values the search recognizes. Whenever any filter can be built, you must then run the search with them and present the first page of results. Criteria that cannot be expressed as filters are left out of the filters and reported back, with what to tell the user and what to offer. Creates NOTHING — no list, no saved contacts, no campaign — and charges no credits.
Inputs
entityTypestringrequired- Which Prospect Search tab the filters are for. "people" when the target is humans (job titles, roles, seniority). "companies" when the target is organizations (industries, headcount, funding, locations).one of
peoplecompanies querystringrequired- Natural language description of the people or companies to find. Include EVERY targeting criterion the user gave: job titles, seniority, departments, industries, company size, revenue, funding, location, skills, named companies, etc. Do NOT include a requested result count — counts are ignored for prepared searches.
fuseaimcp_check_company_savedCheck which companies the user has saved in their CRM, by website domain.Read-onlyCheck Saved Companies
Check which companies the user has saved in their CRM, by website domain. Pass `domains` (1 to 500 company domains or website URLs; each is normalized to its registrable domain). Returns { data } with one row per input domain: { domain, isSaved, companyId? } where isSaved is true only when this user has saved the company, and companyId is the CRM company id whenever a company record exists for that domain (it can exist without being saved by this user). Use it before creating a company or when a flow needs to know whether a company is already in the user's CRM.
Inputs
domainsarrayrequired- No description.
fuseaimcp_check_contact_has_emailCheck whether the user holds a usable email for a contact saved in FuseAI, based on the email statuses already stored (no new verification or other outside call is made).Read-onlyCheck Contact Has Email
Check whether the user holds a usable email for a contact saved in FuseAI, based on the email statuses already stored (no new verification or other outside call is made). Pass contactId (24-hex) and optional verifiedOnly (default true: only verified-valid emails count; false: any deliverable email counts, including unverified user-supplied ones). Returns { hasEmail, email } where email is the primary email when it qualifies, else the first qualifying email, else null. Use it as a gate before an email step.
Inputs
contactIdstringrequired- No description.
verifiedOnlyboolean- No description.default
true
fuseaimcp_check_linkedin_relationCheck whether any LinkedIn account the user owns, or one a teammate shared with them, is connected to a LinkedIn profile (by its slug/public id), and list pending invitations sent to or received from that person.Read-onlyCheck LinkedIn Relation
Check whether any LinkedIn account the user owns, or one a teammate shared with them, is connected to a LinkedIn profile (by its slug/public id), and list pending invitations sent to or received from that person. Returns isRelation, the connected accountIds and those invitations. It reads FuseAI's synced copy and does not contact LinkedIn, so a very recent change may not show yet.
Inputs
relationIdstringrequired- The LinkedIn profile slug / public id (the part after /in/ in the profile URL, e.g. 'john-doe-123')
fuseaimcp_check_sequence_list_compatibility_toolPre-flight check before attaching a list to a sequenced campaign: verifies that at least one contact in the list has a valid email and/or a LinkedIn URL, as the sequence's channels require (when both are required, one contact must have both; at least one of isEmailRequired / isLinkedInRequired must be true).Read-onlyCheck Sequence/List Compatibility
Pre-flight check before attaching a list to a sequenced campaign: verifies that at least one contact in the list has a valid email and/or a LinkedIn URL, as the sequence's channels require (when both are required, one contact must have both; at least one of isEmailRequired / isLinkedInRequired must be true). Returns whether it is compatible (and why not). Use it when unsure a list can actually be sent the planned sequence.
Inputs
isEmailRequiredbooleanrequired- Whether the sequence contains email steps
isLinkedInRequiredbooleanrequired- Whether the sequence contains LinkedIn steps
listIdstringrequired- The prospect list _id (24-character Mongo id) to check
fuseaimcp_contact_call_engagement_summaryPer-contact power-dialer engagement digest over a time range (default: the 30 days up to timeRangeTo, or up to now; read-only): total call attempts, connected conversations, per-disposition counts, latest outcome with note and a short transcript snippet with explicit speaker labels.Read-onlyContact Call Engagement Summary
Per-contact power-dialer engagement digest over a time range (default: the 30 days up to timeRangeTo, or up to now; read-only): total call attempts, connected conversations, per-disposition counts, latest outcome with note and a short transcript snippet with explicit speaker labels. Use to decide which leads to double down on or drop — e.g. 'connected twice but no next step' or 'tried 4+ times, never answered'. Optional: timeRangeFrom/timeRangeTo (ISO dates), sortBy (most-connected default | most-attempted | most-recent), minimumTotalCalls, maximumConnectedCalls (set to 0 for never-connected leads), latestDispositions, limit (default 20, max 50), page. For a specific contact's transcripts, follow up with get_contact_call_transcripts.
Inputs
latestDispositionsarray- No description.
limitinteger- No description.default
20 maximumConnectedCallsinteger- No description.
minimumTotalCallsinteger- No description.
pageinteger- No description.default
1 sortBystring- No description.one of
most-connectedmost-attemptedmost-recentdefaultmost-connected timeRangeFromstring- No description.
timeRangeTostring- No description.
fuseaimcp_count_all_enriched_contacts_toolReturns how many contacts are in the dynamic 'All Enriched Contacts' list.Read-onlyCount All Enriched Contacts
Returns how many contacts are in the dynamic 'All Enriched Contacts' list.
Inputs
This tool takes no inputs.
fuseaimcp_deep_research_statusChecks a Fuse Deep Research run started with deep_research_start.Read-onlyDeep Research Status
Checks a Fuse Deep Research run started with deep_research_start. Returns the run status (queued/planning/running/generating_csv/uploading/completed/failed/cancelled) and, once completed, the full markdown research report plus the id of any produced prospect list. Use it whenever the user asks whether their research is done or wants the results.
Inputs
runIdstringrequired- The runId returned by deep_research_start.
fuseaimcp_dialer_prospect_list_insightsGet power-dialer call stats for each of the team's contact lists: contacts on the list (numbersDialable), contacts dialed and answered, contacts with each connected outcome (positive, neutral, negative, follow-up) and when the list was last dialed.Read-onlyDialer Prospect List Insights
Get power-dialer call stats for each of the team's contact lists: contacts on the list (numbersDialable), contacts dialed and answered, contacts with each connected outcome (positive, neutral, negative, follow-up) and when the list was last dialed. Each contact is counted once, and only outbound calls placed outside campaigns by users whose call history you can see are included. search matches list names. Paginated via page/limit (default 20, max 100), most recently dialed or created lists first. Read-only.
Inputs
limitinteger- No description.default
20 pageinteger- No description.default
1 searchstring- No description.
fuseaimcp_extract_list_filter_values_toolReturns, as `companies`, the distinct values of a field ('name' or 'domain') over a window of a company list's rows (start/amount, in the list's display order), plus the list's row count as `total` — useful for building a filter before viewing rows.Read-onlyExtract List Filter Values
Returns, as `companies`, the distinct values of a field ('name' or 'domain') over a window of a company list's rows (start/amount, in the list's display order), plus the list's row count as `total` — useful for building a filter before viewing rows.
Inputs
fieldstringrequired- Which field to extract distinct values forone of
namedomain listIdstringrequired- No description.
amountinteger- Rows to read from start, 1–1000 (default 1000); the answer holds at most this many valuesdefault
1000 startinteger- 1-based row of the list to start from, in the list's display order (default 1)default
1
fuseaimcp_find_fuseai_pageFind the right FuseAI page/feature to navigate to from a natural-language query (e.g.Read-onlyFind FuseAI Page
Find the right FuseAI page/feature to navigate to from a natural-language query (e.g. 'where do I manage my domains?', 'where are campaign analytics?'). Returns up to 3 matching pages with their URLs. Pass the query as `q`.
Inputs
qstringrequired- A natural-language description of the FuseAI page or feature to navigate to (e.g. 'where do I manage my domains?').
fuseaimcp_generate_email_from_subjectAI-draft a full email from a subject line in the user's voice.Read-onlyGenerate email from subject
AI-draft a full email from a subject line in the user's voice. Returns the draft text only — it does NOT send anything.
Inputs
emailAddressstringrequired- The sending account's address (one of the user's connected email accounts).
subjectLinestringrequired- The subject line / topic to write the email about.
fuseaimcp_generate_email_replyAI-draft a reply to an email in the user's voice (uses their knowledge hub).Read-onlyGenerate email reply
AI-draft a reply to an email in the user's voice (uses their knowledge hub). Returns the draft text only — it does NOT send anything.
Inputs
emailAddressstringrequired- The receiving account's address (the email's receiverEmailAddress — one of the user's connected email accounts).
emailIdstringrequired- The id (_id) of the email to reply to, from the user's inbox.
fuseaimcp_generate_linkedin_replyAI-draft a LinkedIn reply in the user's voice for a chat.Read-onlyGenerate LinkedIn reply
AI-draft a LinkedIn reply in the user's voice for a chat. It works from FuseAI's synced copy of the chat, the team's knowledge hub and, when available, similar past chats on that LinkedIn account, and does not contact LinkedIn or mark the chat as read. Returns the draft text only — it does NOT send anything.
Inputs
accountIdstringrequired- The id of the user's connected LinkedIn account.
chatIdstringrequired- The id of the LinkedIn chat to reply to, from the user's LinkedIn inbox.
fuseaimcp_get_action_detailGet the stored details of a single action by its id: its type and config (for a send_message action: channel, drafted subject and body, sender and recipient).Read-onlyGet Smart Action Detail
Get the stored details of a single action by its id: its type and config (for a send_message action: channel, drafted subject and body, sender and recipient). The feed card's own fields (title, status, contact) come from list_action_feed. Takes the actionId from a feed row (not the actionFeedId).
Inputs
actionIdstringrequired- The action id (24-character Mongo id) — NOT the actionFeedId. Feed items surface it as `actionId`.
fuseaimcp_get_assistant_texting_settingsRead the user's settings for texting Dash, FuseAI's assistant, on iMessage (channel "imessage") or WhatsApp (channel "whatsapp"): the number they text Dash on, their linked phone and whether it is active, and (iMessage only) the texts Dash sends them first: urgent-task alerts on or off, and the weekday morning brief on or off with its local hour.Read-onlyGet Assistant Texting Settings
Read the user's settings for texting Dash, FuseAI's assistant, on iMessage (channel "imessage") or WhatsApp (channel "whatsapp"): the number they text Dash on, their linked phone and whether it is active, and (iMessage only) the texts Dash sends them first: urgent-task alerts on or off, and the weekday morning brief on or off with its local hour.
Inputs
channelstringrequired- The messaging channelone of
imessagewhatsapp
fuseaimcp_get_automationRead a single automation by automationId (full graph, status, health, schedule), with owner { _id, name } (a teammate's rule runs as its owner), permissions { canEditGraph, canManageStatus, canDelete, canSaveTemplate, canDuplicate } (what the user may do with it).Read-onlyGet Automation
Read a single automation by automationId (full graph, status, health, schedule), with owner { _id, name } (a teammate's rule runs as its owner), permissions { canEditGraph, canManageStatus, canDelete, canSaveTemplate, canDuplicate } (what the user may do with it).
Inputs
automationIdstringrequired- No description.
fuseaimcp_get_automation_analytics_summarySummarize the user's own automations, not teammates' (total, active, paused and auto-paused rule counts, rules by origin, successful executions today and this week, latest failure).Read-onlyAutomation Analytics Summary
Summarize the user's own automations, not teammates' (total, active, paused and auto-paused rule counts, rules by origin, successful executions today and this week, latest failure). Optional IANA timezone (e.g. "America/New_York") for the today/this-week windows; defaults to UTC.
Inputs
timezonestring- No description.
fuseaimcp_get_automation_executionRead one execution of an automation with every step's output and fan-out items.Read-onlyGet Automation Execution
Read one execution of an automation with every step's output and fan-out items.
Inputs
automationIdstringrequired- No description.
executionIdstringrequired- No description.
fuseaimcp_get_available_custom_propertiesLive catalog of custom contact properties available on the CRM — each entry carries the property's crmPropertyName.Read-onlyGet Available Custom Properties
Live catalog of custom contact properties available on the CRM — each entry carries the property's crmPropertyName.
Inputs
crmIdentifierstringrequired- The connected CRMone of
hubspotsalesforceattiozohopipedrivegohighlevelnotion
fuseaimcp_get_call_historyFetch the power-dialer call history visible to the user: their own calls, plus teammates' calls when the team's call-history visibility allows it (read-only).Read-onlyGet Call History
Fetch the power-dialer call history visible to the user: their own calls, plus teammates' calls when the team's call-history visibility allows it (read-only). Returns recent calls, each with callRecordingSignedUrl, a link to the call recording that anyone holding it can open for 1 hour (null when no recording is stored); supports paging (page, pageSize), an optional contact-name or phone-number search term, and optional filters: toNumber / fromNumber (free-text phone match), directions (inbound/outbound), dispositions (predefined call dispositions), a dialingTimeFrom/dialingTimeTo ISO date range, prospectListId / campaignId (mutually exclusive — a call belongs to either a campaign or a prospect list), companyId (the contact's company), dialedByUserId (a visible team member) and filter=voicemails (only inbound calls where the caller left a voicemail). Every call record counts toward pagination.total: inbound and outbound, whatever the outcome, including calls that failed or were never placed — narrow it with directions.
Inputs
campaignIdstring- No description.
companyIdstring- No description.
dialedByUserIdstring- No description.
dialingTimeFromstring- No description.
dialingTimeTostring- No description.
directionsarray- No description.
dispositionsarray- No description.
filterstring- No description.one of
voicemails fromNumberstring- No description.
pageinteger- No description.default
1 pageSizeinteger- No description.default
20 prospectListIdstring- No description.
searchstring- No description.
toNumberstring- No description.
fuseaimcp_get_call_transcripts_by_dispositionFetch the newest power-dialer calls that match the requested disposition(s), each with its speaker-labeled transcription (Agent / Contact / Machine; cut at 8,000 characters, with transcriptionTruncated set when cut; read-only; calls without a transcript are skipped).Read-onlyGet Call Transcripts By Disposition
Fetch the newest power-dialer calls that match the requested disposition(s), each with its speaker-labeled transcription (Agent / Contact / Machine; cut at 8,000 characters, with transcriptionTruncated set when cut; read-only; calls without a transcript are skipped). Use to analyze patterns across conversations — e.g. common objections in connected-negative calls, or what's working in connected-positive ones. dispositions (required, one or more of): no-answer, left-voicemail, connected-busy, bad-wrong-number, connected-positive, connected-neutral, connected-negative, connected-disqualified, connected-follow-up, not-specified. Optional: limit (default 10, max 20), dialingTimeFrom/dialingTimeTo (ISO date range), campaignId, listId, dialedByUserId (a visible team member). For 'who do I owe a callback' questions prefer get_follow_up_callbacks, which dedupes per contact.
Inputs
dispositionsarrayrequired- No description.
campaignIdstring- No description.
dialedByUserIdstring- No description.
dialingTimeFromstring- No description.
dialingTimeTostring- No description.
limitinteger- No description.default
10 listIdstring- No description.
fuseaimcp_get_calls_by_durationUse when the user asks for power-dialer calls that lasted longer than X seconds/minutes.Read-onlyGet Calls By Duration
Use when the user asks for power-dialer calls that lasted longer than X seconds/minutes. Requires callDurationGreaterThanSeconds. Omit contactId to search ALL calls visible to the user, or provide contactId to restrict results to one contact. Returns newest matching calls with exact durationSeconds, contact, disposition, note, campaign/list context, and speaker-labeled transcription when available. Duration is strictly greater than the threshold using the call recording's length; calls without a recording are excluded. Read-only and paginated via page and limit (default 20, max 50).
Inputs
callDurationGreaterThanSecondsintegerrequired- Return only calls whose recorded duration (the call recording's length) is strictly greater than this whole number of seconds. Calls without a recording are excluded.
contactIdstring- Optional contact id. Omit it to search all calls visible to the requesting user.
limitinteger- No description.default
20 pageinteger- No description.default
1
fuseaimcp_get_campaign_analytics_trend_toolGet campaign outreach performance over time — email, LinkedIn, or dialer (sent/opened/replied, invites/accepted, dialed/answered/connected/voicemails etc.).Read-onlyCampaign Analytics Trend
Get campaign outreach performance over time — email, LinkedIn, or dialer (sent/opened/replied, invites/accepted, dialed/answered/connected/voicemails etc.). Every channel returns exact whole-window totals plus a per-time-bucket series. The email and LinkedIn series carry every count in their totals; the dialer series carries only dialed, answered, connected, connectedPositive and connectedFollowUp, and its other figures are in totals only. Use for both 'how is my campaign performing over time' and 'how are my emails / LinkedIn / calls doing, is it up or down vs before' questions: compare earlier buckets to later ones for the direction of travel. Scope defaults to every campaign the caller is allowed to see — pass campaignId for a single campaign, or targetUserId for one teammate's campaigns. The window defaults to lifetime; pass `since` to look at a recent period only. Quote headline figures from `totals`, never by summing the series — the dialer channel's per-bucket counts deduplicate contacts within each bucket, so summing them over-counts anyone worked in more than one bucket. The dialer's call figures count outbound calls that were actually placed, never inbound calls or calls that failed or were never placed, and by default only calls made from a campaign's dialer steps: pass callType "all" to also count calls dialed straight from a prospect list or a custom number, or "non-campaign" for those alone. callsMade counts the calls; dialed and the other call totals count contacts, each once over the window. dialable is not a call figure: it counts the contacts enrolled in the campaigns' target lists, called or not, and is 0 for calls outside campaigns.
Inputs
channelstringrequired- Which outreach channel's trend to fetch. Every channel returns all of its metrics per bucket.one of
emaillinkedindialer callTypestring- Dialer only: which calls to count. "campaign" (the default) counts calls made from a campaign's dialer steps; "non-campaign" counts calls dialed straight from a prospect list or a custom number; "all" counts both. With campaignId, only that campaign's calls count; with targetUserId, the calls outside campaigns are the ones that member dialed.one of
campaignnon-campaignall campaignIdstring- Scope to a single campaign. Omit for numbers across campaigns. Cannot be combined with targetUserId.
sincestring- Start of the trend window as an ISO 8601 date-time instant (e.g. 2026-08-01T00:00:00.000Z). Omit for lifetime analytics — the window then starts at the scope's first campaign. Must not be in the future. Do not send a date-only YYYY-MM-DD.
targetUserIdstring- Scope to one team member's campaigns — the campaign OWNER's user id, not the person asking. Use only when the user asks about a specific teammate by name and you have resolved their id. Omit to cover every campaign the caller can see. Cannot be combined with campaignId.
timezonestring- IANA timezone (e.g. America/New_York) whose calendar days the series is bucketed on. Omit for UTC.default
UTC tostring- End of the trend window as an ISO 8601 date-time instant. Omit to run up to now. Do not send a date-only YYYY-MM-DD.
fuseaimcp_get_campaign_lane_funnel_toolFor each signal condition (conditionType 'invite_accepted' or 'clicked_link') of a campaign's branched sequence: contacts waiting on it, contacts it has routed, how many went down its yes and no lanes, and how many were routed early by their signal rather than at the end of the window.Read-onlyGet Campaign Lane Funnel
For each signal condition (conditionType 'invite_accepted' or 'clicked_link') of a campaign's branched sequence: contacts waiting on it, contacts it has routed, how many went down its yes and no lanes, and how many were routed early by their signal rather than at the end of the window. A/B splits are not listed here: read them with get_campaign_split_results_tool.
Inputs
campaignIdstringrequired- No description.
fuseaimcp_get_campaign_lead_activity_toolShow one lead's activity timeline within a campaign: planned steps, emails scheduled and sent, replies received, and LinkedIn invitations and messages (email opens and clicks are not included).Read-onlyGet Campaign Lead Activity
Show one lead's activity timeline within a campaign: planned steps, emails scheduled and sent, replies received, and LinkedIn invitations and messages (email opens and clicks are not included). Needs the campaignId and the lead's targetContactId.
Inputs
campaignIdstringrequired- The campaign the lead belongs to (its 24-hex MongoDB id).
targetContactIdstringrequired- The lead's targetContactId (a 24-hex MongoDB id).
fuseaimcp_get_campaign_sequence_steps_toolLists a campaign's sequence steps WITH their step ids (each step's _id), channels, wait days, and content.Read-onlyGet Campaign Sequence Steps
Lists a campaign's sequence steps WITH their step ids (each step's _id), channels, wait days, and content. The returned _id values are the campaign's real step ids, usable wherever a sequenceStepId is required. Steps form a tree: a step with channel 'condition' (conditionType 'invite_accepted' or 'clicked_link'; its daysBetween is the window in days) is always the last step of its list and splits the sequence into two lanes, the steps whose parentStepId is the condition's id with branchKey 'yes' or 'no'. Root-list steps have parentStepId null. sequenceNumber is the step's position in that tree, read depth-first (a condition, then its yes lane, then its no lane). An A/B split is a condition with conditionType 'ab_split' and daysBetween 0: its two lanes are its arms, branchKey 'a' and 'b' (read a then b), and its splitArms holds each arm's optional name (label) and whole-percent weight, the share of contacts routed to that arm (each contact's arm comes from a fixed hash of the contact and the split, so it keeps its arm).
Inputs
campaignIdstringrequired- No description.
fuseaimcp_get_campaign_split_results_toolPer-arm results of each A/B split (conditionType 'ab_split') in a campaign: contacts routed to each arm and their share, contacts the arm reached, LinkedIn invitations sent and accepted, human replies, positive replies and meetings booked (rates are per contact routed, except the acceptance rate, which is per invitation sent), bounces and unsubscribes; for all time and, once the split's weights changed on the running campaign, for the contacts the split routed since (contacts the change moved to the other arm count in all time only, in their new arm).Read-onlyGet Campaign Split Results
Per-arm results of each A/B split (conditionType 'ab_split') in a campaign: contacts routed to each arm and their share, contacts the arm reached, LinkedIn invitations sent and accepted, human replies, positive replies and meetings booked (rates are per contact routed, except the acceptance rate, which is per invitation sent), bounces and unsubscribes; for all time and, once the split's weights changed on the running campaign, for the contacts the split routed since (contacts the change moved to the other arm count in all time only, in their new arm). Each split also has replySeries: replies per arm over time, zero-filled. Optional since / to (ISO instants) keep only the contacts that entered an arm in that range; timezone (IANA) sets the days the series is bucketed by. A contact counts in the arm it is in, whatever step it replied to. There is no winner and no significance test: when isSmallSample is true (fewer than 20 replies across both arms), say that a gap between the arms is most likely chance. To shift traffic between arms, change the split's weights with update_sequence_step_tool.
Inputs
campaignIdstringrequired- No description.
sincestring- ISO 8601 instant: count only the contacts that entered an arm at or after it. Omitted: since launch.
timezonestring- IANA timezone whose calendar days the reply series is bucketed by. Default: UTC.default
UTC tostring- ISO 8601 instant: count only the contacts that entered an arm at or before it. Omitted: up to now.
fuseaimcp_get_campaign_toolFetch one campaign's full configuration by id — name, status, sequence settings, schedule, source lists, exclusions, unsubscribe/open/click tracking, and the resolved tracking-link origin (custom domain vs Fuse).Read-onlyGet Campaign
Fetch one campaign's full configuration by id — name, status, sequence settings, schedule, source lists, exclusions, unsubscribe/open/click tracking, and the resolved tracking-link origin (custom domain vs Fuse). trackingDomainId null still uses the team's default ACTIVE domain at send time when click tracking is on. senderName and jobTitle are only a fallback. Recipients see the mailbox or LinkedIn account that sends each message: an email's From name is the mailbox's name (an Outlook mailbox's follow-ups sent as a reply in the same thread carry the name set in Microsoft), and the sending account's name, title and signature fill the {Sender …} placeholders. The campaign's senderName and jobTitle fill those placeholders only for a sending account with no name or title of its own, and no screen in the app shows or edits them.
Inputs
campaignIdstringrequired- No description.
fuseaimcp_get_campaign_visibilityRead the team's campaign visibility setting (who on the team can see and act on each other's campaigns): PRIVATE, ADMINS_ONLY or ALL_MEMBERS, and ALL_MEMBERS if it was never set.Read-onlyGet Campaign Visibility
Read the team's campaign visibility setting (who on the team can see and act on each other's campaigns): PRIVATE, ADMINS_ONLY or ALL_MEMBERS, and ALL_MEMBERS if it was never set. Team owners and admins only.
Inputs
This tool takes no inputs.
fuseaimcp_get_company_filter_autocompleteResolve a company (or investor) name to its canonical name + domain, for use in company-search filters (nameLike / investors).Read-onlyCompany Filter Autocomplete
Resolve a company (or investor) name to its canonical name + domain, for use in company-search filters (nameLike / investors). Pass orgType=investor for investor lookups, or orgType=school for school lookups. The name is looked up live in the database of the external company data provider that search_companies searches; the lookup charges no credits and saves nothing.
Inputs
querystringrequired- No description.
orgTypestring- No description.one of
investorschool
fuseaimcp_get_company_search_filter_optionsGet the canonical reference lists for company-search filters: industries, company tags, NAICS codes, countries (apiCodes), accelerators, funding stages (`stage`), statuses and ranking lists (`fortuneRankings`) — the valid values for advanced company filter sets.Read-onlyCompany Search Filter Options
Get the canonical reference lists for company-search filters: industries, company tags, NAICS codes, countries (apiCodes), accelerators, funding stages (`stage`), statuses and ranking lists (`fortuneRankings`) — the valid values for advanced company filter sets. Read-only; most of these lists are fetched from FuseAI's external company-data provider when no copy from the last 24 hours is cached.
Inputs
This tool takes no inputs.
fuseaimcp_get_contact_call_summaryGet an all-history calling summary for ONE contact (read-only): attempts, connected calls, disposition counts, first/last call, total/average/longest recorded duration, meaningful-call count, and the latest meaningful call with speaker-labeled transcript.Read-onlyGet Contact Call Summary
Get an all-history calling summary for ONE contact (read-only): attempts, connected calls, disposition counts, first/last call, total/average/longest recorded duration, meaningful-call count, and the latest meaningful call with speaker-labeled transcript. Pass exactly one of contactId or contactName; ambiguous names return candidates for disambiguation. meaningfulCallDurationGreaterThanSeconds defaults to 30; a meaningful call must have a connected-* disposition AND a recorded duration strictly above that threshold.
Inputs
contactIdstring- No description.
contactNamestring- No description.
meaningfulCallDurationGreaterThanSecondsinteger- No description.default
30
fuseaimcp_get_contact_call_transcriptsFetch the most recent power-dialer calls for ONE contact, each with its speaker-labeled transcription (Agent / Contact / Machine; cut at 15,000 characters, with transcriptionTruncated set when cut), disposition, note, duration and campaign/list context (read-only).Read-onlyGet Contact Call Transcripts
Fetch the most recent power-dialer calls for ONE contact, each with its speaker-labeled transcription (Agent / Contact / Machine; cut at 15,000 characters, with transcriptionTruncated set when cut), disposition, note, duration and campaign/list context (read-only). Use for pre-call briefings, conversation recaps, objection review, or drafting a follow-up from what was said. Pass exactly one of contactId (24-hex id, preferred when known) or contactName (case-insensitive partial name match against contacts the team has called; several matches return up to 5 candidates). Optional callDurationGreaterThanSeconds filters to calls whose recorded duration is strictly greater than that many seconds. For generic 'calls longer than X seconds' requests, prefer get_calls_by_duration.
Inputs
callDurationGreaterThanSecondsinteger- Return only calls whose recorded duration (the call recording's length) is strictly greater than this whole number of seconds. Calls without a recording are excluded.
contactIdstring- Contact id. Provide exactly one of contactId or contactName.
contactNamestring- Fuzzy contact name. Provide exactly one of contactId or contactName.
limitinteger- Maximum matching calls to return, newest first.default
5
fuseaimcp_get_credit_pricesRead-only: what platform actions cost in credits, from the same rates the charges use: people and company search results, saving to lists, demographics and company matches, email/phone/deep-search enrichment (charged only when something is found), email validation, AI-written and AI-personalized campaign messages, and campaign email sends (with emailSendsIncludedInPlan saying whether this user's plan includes them).Read-onlyGet Credit Prices
Read-only: what platform actions cost in credits, from the same rates the charges use: people and company search results, saving to lists, demographics and company matches, email/phone/deep-search enrichment (charged only when something is found), email validation, AI-written and AI-personalized campaign messages, and campaign email sends (with emailSendsIncludedInPlan saying whether this user's plan includes them). Each price has action, credits, unit, chargedWhen and product (the product it is filed under in credit history). The notes give what a LinkedIn seat costs on this plan and what renting a phone number costs, how spend reports split enrichment, shared product labels and what is not listed. Signal monitors, smart columns, scrapes, deep research and automations price themselves and are not listed. Takes no arguments.
Inputs
This tool takes no inputs.
fuseaimcp_get_credits_balanceRead-only: returns the user's current FuseAI credit balance as { balance }.Read-onlyGet Credits Balance
Read-only: returns the user's current FuseAI credit balance as { balance }. Use it when the user asks how many credits they have, and BEFORE starting a credit-consuming bulk operation (e.g. bulk enrichment) to check the batch is affordable — bulk operations are priced up front as price-per-contact × contact count and fail with INSUFFICIENT_CREDITS when that total exceeds this balance. Takes no arguments.
Inputs
This tool takes no inputs.
fuseaimcp_get_credits_spend_by_productRead-only: credits spent per product over a date window, as { points: [{ date, isoDate, products: [{ product, value }] }] } — one point per time bucket (daily, weekly, monthly or quarterly by window length), zero-filled.Read-onlyGet Credits Spend by Product
Read-only: credits spent per product over a date window, as { points: [{ date, isoDate, products: [{ product, value }] }] } — one point per time bucket (daily, weekly, monthly or quarterly by window length), zero-filled. Use it for 'where did my credits go' and 'how much did enrichment cost this week'. `since` (ISO 8601 date-time, e.g. 2026-11-01T00:00:00Z; periodStartDate from get_credits_usage is in this form) is optional: omit it for all-time spend. `to` (ISO 8601 date-time) defaults to now; `timezone` (IANA, default UTC) sets the calendar days. `memberIds` (team members' 24-hex user ids: userId in list_team_members, or userInfo._id from get_user_info for the user's own) narrows to those people; omit it for the whole account, which for a team member is the whole team (pass their own user id for their own spend). An id that is not the user id of someone currently on the user's team fails with TEAM_MEMBER_NOT_IN_SCOPE. Product labels are credit history's, except that Contact Enrichment is reported as Email Enrichment and Phone Enrichment.
Inputs
memberIdsarray- No description.
sincestring- No description.
timezonestring- No description.default
UTC tostring- No description.
fuseaimcp_get_credits_usageRead-only: the user's credits for the current period as { used, available, total, periodStartDate, periodEndDate, oneTimeCreditsHoldback }.Read-onlyGet Credits Usage
Read-only: the user's credits for the current period as { used, available, total, periodStartDate, periodEndDate, oneTimeCreditsHoldback }. periodStartDate and periodEndDate bound the usage window: the billing period, or the last 30 days ending today when there is no active subscription. periodEndDate is not a renewal date: for when credits renew, or whether the plan ends instead, read the subscription (get_user_info: renewalDate, isPlannedForCancellation). For an account owner `used` is the period's grants minus the current balance, so credits left from earlier grants lower it; for exact spend use get_credits_spend_by_product with since = periodStartDate. For a team member `used` is their spend against their credit cap, and `total` is the cap (the owner's grants for the period when they have no cap). For the spendable balance right now use get_credits_balance; for what the credits were spent on use get_credits_spend_by_product. Takes no arguments.
Inputs
This tool takes no inputs.
fuseaimcp_get_crm_company_by_idFetch one CRM company by its companyId (the company's 24-hex id — what a company event carries as companyId, and what a contact record's companyId points at).Read-onlyGet CRM Company
Fetch one CRM company by its companyId (the company's 24-hex id — what a company event carries as companyId, and what a contact record's companyId points at). Returns name, domain, industry, address, headquarters country, LinkedIn headcount and revenue. Use it to read the stored details of a company whose id you already have; never page through a list to find one company.
Inputs
companyIdstringrequired- The company's 24-hex id (companies._id).
fuseaimcp_get_crm_contact_by_idFetch one contact saved in FuseAI (not a record in the connected CRM) by its contactId (the contact's 24-hex id), returning the full record (profile fields, company data, and the emails and phones the user holds for it).Read-onlyGet CRM Contact
Fetch one contact saved in FuseAI (not a record in the connected CRM) by its contactId (the contact's 24-hex id), returning the full record (profile fields, company data, and the emails and phones the user holds for it).
Inputs
contactIdstringrequired- The CRM contact id (24-character Mongo id)
fuseaimcp_get_crm_contactsList the contacts saved in FuseAI by the user or their teammates (FuseAI's own records, not records in the connected CRM; no outside service is called), with optional filters (name, firstName, lastName, country, isArchived, and company, which takes the company's 24-hex FuseAI id; a company name is ignored and returns unfiltered results) and pagination (page, pageSize, sortBy, sortOrder asc|desc).Read-onlyList CRM Contacts
List the contacts saved in FuseAI by the user or their teammates (FuseAI's own records, not records in the connected CRM; no outside service is called), with optional filters (name, firstName, lastName, country, isArchived, and company, which takes the company's 24-hex FuseAI id; a company name is ignored and returns unfiltered results) and pagination (page, pageSize, sortBy, sortOrder asc|desc). Returns contacts plus pagination info. Use this to browse or filter saved contacts.
Inputs
companystring- No description.
countrystring- No description.
firstNamestring- No description.
isArchivedboolean- No description.
lastNamestring- No description.
namestring- No description.
pageinteger- No description.default
1 pageSizeinteger- No description.default
20 sortBystring- No description.
sortOrderstring- No description.one of
ascdesc
fuseaimcp_get_crm_integration_dataGet the CRM the user has connected, as stored in FuseAI (HubSpot, Salesforce, Attio, Zoho, Pipedrive, GoHighLevel or Notion): crmIdentifier, the CRM account id (teamPortalId) and the email it was connected with; null when none is connected.Read-onlyGet CRM Integration Data
Get the CRM the user has connected, as stored in FuseAI (HubSpot, Salesforce, Attio, Zoho, Pipedrive, GoHighLevel or Notion): crmIdentifier, the CRM account id (teamPortalId) and the email it was connected with; null when none is connected. Returns no tokens and does not call the CRM.
Inputs
This tool takes no inputs.
fuseaimcp_get_crm_settingsGet the team's CRM sync settings as saved in FuseAI, with the field lists (availableContactFields, FuseAI's built-in list of syncable fields for this CRM; disabledContactFields; customContactProperties) and the contactPushConditions.Read-onlyGet CRM Settings
Get the team's CRM sync settings as saved in FuseAI, with the field lists (availableContactFields, FuseAI's built-in list of syncable fields for this CRM; disabledContactFields; customContactProperties) and the contactPushConditions. It reads FuseAI's stored settings only and does not call the CRM.
Inputs
crmIdentifierstringrequired- The connected CRMone of
hubspotsalesforceattiozohopipedrivegohighlevelnotion
fuseaimcp_get_crm_team_connectionGet the team-level CRM connection as stored in FuseAI: which CRM the team has connected and which members are connected (their userIds can be used as userForSyncId).Read-onlyGet CRM Team Connection
Get the team-level CRM connection as stored in FuseAI: which CRM the team has connected and which members are connected (their userIds can be used as userForSyncId). Returns no tokens and does not call the CRM.
Inputs
This tool takes no inputs.
fuseaimcp_get_domain_visibility_settingsRead who on the team can see and manage the purchased domains/mailboxes (PRIVATE | ADMINS_ONLY | ALL_MEMBERS).Read-onlyGet Domain Visibility Settings
Read who on the team can see and manage the purchased domains/mailboxes (PRIVATE | ADMINS_ONLY | ALL_MEMBERS). Team owners and admins only.
Inputs
This tool takes no inputs.
fuseaimcp_get_duplicate_contacts_jobRead a duplicate scan, duplicate review or merge job by its jobId.Read-onlyGet Duplicate Contacts Job
Read a duplicate scan, duplicate review or merge job by its jobId. Until it finishes it answers its status. A completed scan or review answers its totals and one page of duplicate groups (page, pageSize up to 25, default 10): each group's members with name, title, company, LinkedIn URL, primary email and how many emails and phones they hold, what the group matched on, the lists its members are in, and the contact a merge would keep. A completed merge answers how many groups it merged, how many contacts it archived and how many groups it skipped.
Inputs
jobIdstringrequired- The jobId a duplicate scan, duplicate review or merge answered with
pageinteger- Page of duplicate groups to read, starting at 1default
1 pageSizeinteger- Duplicate groups per page, up to 25default
10
fuseaimcp_get_follow_up_callbacksList contacts the user, or a teammate whose calls they can see, still owes a callback: contacts whose MOST RECENT call has a follow-up-worthy connected disposition and who have not been called since — each with that call's speaker-labeled transcription (Agent / Contact / Machine; cut at 8,000 characters, with transcriptionTruncated set when cut) and note so the promised next step can be read from the conversation (read-only).Read-onlyGet Follow-Up Callbacks
List contacts the user, or a teammate whose calls they can see, still owes a callback: contacts whose MOST RECENT call has a follow-up-worthy connected disposition and who have not been called since — each with that call's speaker-labeled transcription (Agent / Contact / Machine; cut at 8,000 characters, with transcriptionTruncated set when cut) and note so the promised next step can be read from the conversation (read-only). Use for 'who should I call back and why / did I promise anyone a follow-up'. Optional: dispositions (default connected-follow-up; also accepts connected-busy, connected-positive, connected-neutral, connected-negative, connected-disqualified), windowDays (default 30, max 90), limit (default 10, max 20), page. Sorted by most recently recorded disposition.
Inputs
dispositionsarray- No description.
limitinteger- No description.default
10 pageinteger- No description.default
1 windowDaysinteger- No description.default
30
fuseaimcp_get_fuse_property_mappingsGet the Fuse properties that can be written to the CRM (availableFuseProperties, e.g.Read-onlyGet Fuse Property Mappings
Get the Fuse properties that can be written to the CRM (availableFuseProperties, e.g. LinkedIn connection status), the team's current mappings onto CRM contact properties, and where the sync stands (lastSyncedAt, syncCounts, isSyncing, and syncProblem when something only the team can fix is stopping it). It reads the mappings and sync status stored in FuseAI only and does not call HubSpot. HubSpot only for now: any other crmIdentifier returns the error FUSE_PROPERTY_SYNC_NOT_SUPPORTED.
Inputs
crmIdentifierstringrequired- The connected CRMone of
hubspotsalesforceattiozohopipedrivegohighlevelnotion
fuseaimcp_get_global_campaign_config_toolGet the user's global campaign configuration.Read-onlyGet Global Campaign Config
Get the user's global campaign configuration. It currently returns only { ctaLink }: the calendar (booking) link remembered from the most recent campaign the user created with one, or null. It returns no sending defaults, limits or schedules, and new campaigns get no link unless ctaLink is passed to initialize_campaign_tool.
Inputs
This tool takes no inputs.
fuseaimcp_get_linkedin_accountRead one LinkedIn account the user owns by providerId (from list_linkedin_accounts): status, limits, signature, sharing.Read-onlyGet LinkedIn Account
Read one LinkedIn account the user owns by providerId (from list_linkedin_accounts): status, limits, signature, sharing. An account a teammate shared with the user returns not found; its details are in list_linkedin_accounts.
Inputs
providerIdstringrequired- No description.
fuseaimcp_get_linkedin_chatGet the full message thread of one LinkedIn chat by chatId, from FuseAI's synced copy of the user's LinkedIn inbox (it does not contact LinkedIn or mark the chat as read).Read-onlyGet LinkedIn Chat
Get the full message thread of one LinkedIn chat by chatId, from FuseAI's synced copy of the user's LinkedIn inbox (it does not contact LinkedIn or mark the chat as read).
Inputs
chatIdstringrequired- The LinkedIn conversation's chat id
fuseaimcp_get_linkedin_chat_by_urlFind the existing chat between a connected LinkedIn account (accountId) and a person, by the person's LinkedIn profile URL.Read-onlyFind LinkedIn Chat by Profile URL
Find the existing chat between a connected LinkedIn account (accountId) and a person, by the person's LinkedIn profile URL. Each call looks the profile up live on LinkedIn through that account (no profile-visit notification is sent) and counts against LinkedIn's rate limits. Returns FuseAI's stored copy of the chat with its chatId and messages, or chatId null and no messages when there is no chat yet. It sends no message or invitation and does not change the chat or mark it as read; if the account's LinkedIn session has expired, the call returns an error and does not mark the account disconnected, pause its campaigns or notify the owner.
Inputs
accountIdstringrequired- No description.
linkedinUrlstringrequired- No description.
fuseaimcp_get_list_prospect_ids_toolReturns the entity ids in a list: contactIds and companyIds for a stored list, or contactIds for the dynamic lists all_enriched_contacts, all_saved_contacts, job_changed_contacts, all_crm_records and my_crm_records.Read-onlyGet List Prospect Ids
Returns the entity ids in a list: contactIds and companyIds for a stored list, or contactIds for the dynamic lists all_enriched_contacts, all_saved_contacts, job_changed_contacts, all_crm_records and my_crm_records. Those five are the only symbolic ids it supports: the other symbolic ids the listId field accepts (all_relations, all_received_invitations, all_sent_invitations, all_crm_companies and my_crm_companies) always return empty arrays. That includes all_relations, although the listId field description gives it as an example, so an empty result for any of these five does not mean the user has no LinkedIn connections, LinkedIn invitations or CRM companies. The customContactIds and customCompanyIds arrays are always empty. It only reads FuseAI's stored lists and records, including its synced copy of CRM records, and does not contact LinkedIn, the CRM or any other outside service. Useful before removing rows, updating cells, or pushing contacts to a CRM with this plugin's other tools.
Inputs
listIdstringrequired- A list's Mongo id, or any symbolic dynamic list id (e.g. all_relations, all_enriched_contacts)
fuseaimcp_get_location_filter_autocompleteResolve a place name to the fully qualified place names the headquartersCity company-search filter takes.Read-onlyLocation Filter Autocomplete
Resolve a place name to the fully qualified place names the headquartersCity company-search filter takes. The place is looked up live with the external company data provider that search_companies searches; the lookup charges no credits and saves nothing. Send a chosen candidate's `address` (e.g. Dallas, Texas, United States) whole as the headquartersCity value, or its `name` when it has no address: the qualification is what tells same-named cities apart. Never send latitude and longitude: the filter geocodes the place name itself, and coordinates do not resolve to a city.
Inputs
querystringrequired- No description.
fuseaimcp_get_people_search_filter_optionsGet the accepted values for every people-search filter with a fixed vocabulary (job_title_levels, job_title_role, job_company_size, job_company_inferred_revenue, industry, job_company_industry, location_country, education.degrees, languages.name and the rest), keyed by the filter name search_people and search_and_save_people_to_list take, each with how its value must be shaped.Read-onlyPeople Search Filter Options
Get the accepted values for every people-search filter with a fixed vocabulary (job_title_levels, job_title_role, job_company_size, job_company_inferred_revenue, industry, job_company_industry, location_country, education.degrees, languages.name and the rest), keyed by the filter name search_people and search_and_save_people_to_list take, each with how its value must be shaped. Call it before setting one of those filters rather than guessing a value. Free-text filters (titles, company names, places, skills, schools) have no list here: the lookup_filter_values tool resolves titles, places and schools. Free: no credits, no provider call.
Inputs
This tool takes no inputs.
fuseaimcp_get_prospect_list_rows_toolFetches the rows (contacts or companies) inside a specific prospect list, returning each row (its `_id` plus its fields, including any enriched emails and phones, CRM data and custom-column values) under the top-level `data` array, with `pagination` ({ total, page, pageSize, totalPages }) alongside it.Read-onlyGet Prospect List Rows
Fetches the rows (contacts or companies) inside a specific prospect list, returning each row (its `_id` plus its fields, including any enriched emails and phones, CRM data and custom-column values) under the top-level `data` array, with `pagination` ({ total, page, pageSize, totalPages }) alongside it. `meta` ALSO carries the list's columns (`meta.customColumns`; each column's `_id` is its columnId) — those columnIds are usable wherever an EXISTING column's columnId is required. Supports searching/filtering via `filters`: per-field entries, each { include:[], exclude:[] } unless noted — common keys: icpMatch ({ include:["yes"] } for ICP-matched), emailValidity (valid|invalid|catch-all|unknown|do_not_mail|personal|unverified), enrichedEmail/enrichedPhone (enriched|not_enriched), crmStatus (include-only: in_crm|not_in_crm — Exists in CRM / Not in CRM), title, dept, level, industryName, country, state, companyName, campaignEngagement, createdBy, crmOwner; numberOfEmployees/revenue (numeric ranges); keywords (string, searched in the job title only: no filter searches a headline or bio); linkedinUrl (string); customColumns: { <columnId>: { dataType, value|include|exclude|rangeMin|rangeMax } }. Send only the fields you need. Plus sorting and pagination. Provide the list's listId — this also works on a system list's real _id (e.g. All Website Visitors).
Inputs
listIdstringrequired- No description.
filtersobject- No description.
includeCampaignStatusesboolean- No description.default
true limitinteger- No description.
pageinteger- No description.
pageNuminteger- No description.
sortBystring- No description.
sortOrderstring- No description.one of
ascdescdefaultdesc
fuseaimcp_get_provider_key_catalogThe Deep Research provider catalog as a flat list: each provider's id, category, keyMode (platform | byok | byok-required), whether the user already saved a key (hasUserKey), and docs link.Read-onlyGet Provider Key Catalog
The Deep Research provider catalog as a flat list: each provider's id, category, keyMode (platform | byok | byok-required), whether the user already saved a key (hasUserKey), and docs link. Use to answer which providers support bring-your-own-key.
Inputs
This tool takes no inputs.
fuseaimcp_get_push_condition_icp_optionsThe team's members who have ICPs, each with their ICPs' ids and names: what an icp_match push condition names in icpOwnerUserId and values (update_crm_settings).Read-onlyGet Push Condition ICP Options
The team's members who have ICPs, each with their ICPs' ids and names: what an icp_match push condition names in icpOwnerUserId and values (update_crm_settings). An ICP saved before ICPs had ids is not listed; it counts toward any_icp.
Inputs
crmIdentifierstringrequired- The connected CRMone of
hubspotsalesforceattiozohopipedrivegohighlevelnotion
fuseaimcp_get_signal_monitor_creation_configRead a monitor's creation-shaped config in ANY status: the stored snapshot for drafts, a best-effort reconstruction from FuseAI's current records of the monitor otherwise (unrecoverable fields listed in lossyFields).Read-onlyGet signal-monitor creation config
Read a monitor's creation-shaped config in ANY status: the stored snapshot for drafts, a best-effort reconstruction from FuseAI's current records of the monitor otherwise (unrecoverable fields listed in lossyFields). For discovery agents it also answers `pricing`, the source for cost questions: every delivered result costs minCreditsPerResult, up to maxCreditsPerResult when enrichment finds an email or phone, so N results cost N × minCreditsPerResult up to N × maxCreditsPerResult, and one run delivers at most maxResultsPerRun results under Max Credit Spend; emailEnrichmentMaxCredits and phoneEnrichmentMaxCredits are the most email or phone enrichment adds to one result, whether or not it is on, so you can price a change: turning one on adds it to maxCreditsPerResult and turning it off subtracts it. `pricing` is null for monitors priced another way. `frequencyHours` says how often the monitor runs (24 = daily), null when it runs once. `firstRunAt` and `nextRunAt` are ISO timestamps of its first and next run. For a monitor that is not a draft and runs on an external data provider (CrustData, Exa or Fiber), the tool reads them live from the monitor's watch at that provider; scrape monitors (LinkedIn, Sales Navigator, G2 and Google Maps scrapes) use the dates FuseAI stores. Each is null when it can't be read (both always for drafts, which make no provider call); `nextRunAt` is also null unless the monitor is active with a run due before its end date. The tool only reads: it saves nothing, costs no credits and changes nothing in FuseAI or at the provider.
Inputs
documentIdstringrequired- The signal monitor's documentId.
fuseaimcp_get_signal_monitor_draftRead a draft monitor's saved creationConfig (and type) so you can review or edit it before activating.Read-onlyGet signal-monitor draft
Read a draft monitor's saved creationConfig (and type) so you can review or edit it before activating.
Inputs
documentIdstringrequired- The signal monitor's documentId.
fuseaimcp_get_slack_settingsRead the current Slack integration settings (e.g.Read-onlyGet Slack Settings
Read the current Slack integration settings (e.g. the connected notification channel). A teammate without their own Slack sees the team owner's connection (isTeamConnection), which they post with; canConnect says whether this user can connect Slack themselves. It reads the Slack settings stored in FuseAI only and does not call Slack.
Inputs
This tool takes no inputs.
fuseaimcp_get_smart_column_job_status_toolChecks progress of a smart-column job: status (pending, processing, completed, partially_completed, failed or cancelled), row counts and credits used.Read-onlyGet Smart Column Job Status
Checks progress of a smart-column job: status (pending, processing, completed, partially_completed, failed or cancelled), row counts and credits used. Provide listId and jobId.
Inputs
jobIdstringrequired- No description.
listIdstringrequired- No description.
fuseaimcp_get_smart_column_provenance_toolExplains how one smart-column cell was produced (reasoning, research steps, source URLs).Read-onlyGet Smart Column Provenance
Explains how one smart-column cell was produced (reasoning, research steps, source URLs). Provide listId, entityId (the row's contact or company id) and columnId. After a manual edit of the cell, the result still describes the last AI run.
Inputs
columnIdstringrequired- No description.
entityIdstringrequired- No description.
listIdstringrequired- No description.
fuseaimcp_get_stale_meaningful_contactsFind contacts worth re-engaging because they had at least one meaningful connected call but have had no call of any outcome recently (read-only).Read-onlyGet Stale Meaningful Contacts
Find contacts worth re-engaging because they had at least one meaningful connected call but have had no call of any outcome recently (read-only). Returns each contact's last-call age/outcome and latest meaningful call with speaker-labeled transcript. inactiveForDays defaults to 14. minimumCallDurationSeconds defaults to 30 and is a strict duration threshold; meaningful also requires a connected-* disposition. Paginated via page and limit (default 20, max 50), stalest contacts first.
Inputs
inactiveForDaysinteger- No description.default
14 limitinteger- No description.default
20 minimumCallDurationSecondsinteger- No description.default
30 pageinteger- No description.default
1
fuseaimcp_get_website_allowed_domainsRead-only: the team's website-tracking config — the saved domains and tracked-page labels, plus the monthly website credit cap, credits used and each saved domain's verification status (the team owner's, for every teammate); null when the team has no website-tracking setup yet.Read-onlyGet Website Tracking Config
Read-only: the team's website-tracking config — the saved domains and tracked-page labels, plus the monthly website credit cap, credits used and each saved domain's verification status (the team owner's, for every teammate); null when the team has no website-tracking setup yet. Saved domains are only used for domain verification and the per-domain tag status; they do not limit which visits the tag records.
Inputs
This tool takes no inputs.
fuseaimcp_get_website_tag_scriptRead-only: the visitor-tracking tag/script snippet the user installs on their website.Read-onlyGet Tracking Tag Script
Read-only: the visitor-tracking tag/script snippet the user installs on their website.
Inputs
This tool takes no inputs.
fuseaimcp_get_website_visitor_analytics_by_periodRead-only: website visitor trend over time (visitorsOverTime buckets of distinct identified companies/people across all of the team's tracked sites; anonymous traffic is not counted) from a `since` start date up to an optional `to` end date (defaults to now).Read-onlyWebsite Visitor Trend
Read-only: website visitor trend over time (visitorsOverTime buckets of distinct identified companies/people across all of the team's tracked sites; anonymous traffic is not counted) from a `since` start date up to an optional `to` end date (defaults to now). Omit `since` for the lifetime trend, which starts at the first identified visit. Pass `timezone` (IANA name, e.g. Europe/Kyiv; defaults to UTC) so buckets start on the user's local calendar days. Bucket size is picked from the window length and cannot be set: daily when the window spans up to 13 calendar days, then weekly (starting Monday), monthly or quarterly, so a series has at most 13 points unless it spans more than 13 quarters. Each bucket is labelled by its start date. Returns time-bucketed counts, not overall totals.
Inputs
sincestring- Window start as an ISO 8601 date-time instant (e.g. 2026-08-01T00:00:00.000Z). Omit for lifetime analytics. Must not be in the future. Do not send a date-only YYYY-MM-DD.
timezonestring- IANA timezone name (e.g. Europe/Kyiv) whose calendar days the buckets are drawn on — the zone `since` / `to` day boundaries were computed in. Defaults to UTC. UTC offsets such as +03:00 are rejected.default
UTC tostring- Window end as an ISO 8601 date-time instant. Omit to run the window up to now. Must not be earlier than `since`. Do not send a date-only YYYY-MM-DD.
fuseaimcp_get_website_visitor_reportRead-only: the website visitor intent report — the top companies (mode TOP_COMPANIES), people (mode TOP_PEOPLE) or industries (mode TOP_INDUSTRIES) that visited the team's website, ranked by visit count unless `sortByField` says otherwise.Read-onlyWebsite Visitor Report
Read-only: the website visitor intent report — the top companies (mode TOP_COMPANIES), people (mode TOP_PEOPLE) or industries (mode TOP_INDUSTRIES) that visited the team's website, ranked by visit count unless `sortByField` says otherwise. It reads visitors FuseAI has already identified and stored for the team (no external calls, no credits spent); visits held back by the team's website credit cap are left out. Pass `since` (and optionally `to`) to scope the window; omit both for all time. Use for 'who visits our site' and 'which industries/companies visit most' summaries.
Inputs
modestringrequired- Which breakdown to return: TOP_COMPANIES (visiting companies), TOP_PEOPLE (individual visitors) or TOP_INDUSTRIES.one of
TOP_COMPANIESTOP_PEOPLETOP_INDUSTRIES icpFilterCompanyNamestring- Company-name filter: JSON array of { "status": "include" | "exclude", "value": "<substring>" } items, matched case-insensitively; includes are OR-ed, excludes AND-ed.
icpFilterDepartmentstring- Department filter: JSON array of { "status": "include" | "exclude", "value": "<substring>" } items, matched case-insensitively; includes are OR-ed, excludes AND-ed.
icpFilterIndustrystring- Industry filter: JSON array of { "status": "include" | "exclude", "value": "<substring>" } items, matched case-insensitively; includes are OR-ed, excludes AND-ed.
icpFilterLocationstring- Location filter: JSON array of { "status": "include" | "exclude", "value": "<substring>" } items, matched case-insensitively; includes are OR-ed, excludes AND-ed.
icpFilterPagePathstring- Page-path filter: JSON array of { "status": "include" | "exclude", "value": "<substring>" } items, matched case-insensitively; includes are OR-ed, excludes AND-ed.
icpFilterPersonNamestring- Person-name filter: JSON array of { "status": "include" | "exclude", "value": "<substring>" } items, matched case-insensitively; includes are OR-ed, excludes AND-ed.
icpFilterSizestring- Company-size filter: JSON array of { "status": "include" | "exclude", "value": "<substring>" } items, matched case-insensitively; includes are OR-ed, excludes AND-ed.
limitinteger- How many rows to return (default 10, max 500).default
10 pageinteger- 1-based page (default 1).default
1 searchQuerystring- Optional free-text filter on the company (TOP_COMPANIES) or person (TOP_PEOPLE) name; ignored for TOP_INDUSTRIES.
sincestring- Window start as an ISO 8601 date-time instant (e.g. 2026-08-01T00:00:00.000Z). Omit for the all-time report. Must not be in the future. Do not send a date-only YYYY-MM-DD.
sortByFieldstring- Field to sort by: count (visits, the default), lastVisitDate, companyName, name, companySize, companyIndustry, department, location, confidenceInterval; for TOP_INDUSTRIES also industry, companies, countries, locations, visitors. Unknown values fall back to visit count.
sortByOrderstring- Sort direction (default desc).one of
ascdescdefaultdesc tostring- Window end as an ISO 8601 date-time instant. Omit to run the window up to now. Must not be earlier than `since`. Do not send a date-only YYYY-MM-DD.
fuseaimcp_get_website_visitor_totalsOverall totals for a window across all of the team's tracked sites, counting only visits made visible in reports (charged credits or upgraded by background identification); other traffic is not counted: totalVisitors (those page views, not distinct people; a few may lack person and company details), totalUniqueCompanies, totalUniquePeople and totalHotVisits (page views whose visitor was identified with high confidence).Read-onlyWebsite Visitor Totals
Overall totals for a window across all of the team's tracked sites, counting only visits made visible in reports (charged credits or upgraded by background identification); other traffic is not counted: totalVisitors (those page views, not distinct people; a few may lack person and company details), totalUniqueCompanies, totalUniquePeople and totalHotVisits (page views whose visitor was identified with high confidence). Lifetime totals start at the first such visit. Pass a `since` start date and optionally a `to` end date (defaults to now); omit `since` for lifetime totals.
Inputs
sincestring- Window start as an ISO 8601 date-time instant (e.g. 2026-08-01T00:00:00.000Z). Omit for lifetime analytics. Must not be in the future. Do not send a date-only YYYY-MM-DD.
tostring- Window end as an ISO 8601 date-time instant. Omit to run the window up to now. Must not be earlier than `since`. Do not send a date-only YYYY-MM-DD.
fuseaimcp_google_jobs_search_toolSearches public Google Jobs listings for a keyword or phrase, in a country (two-letter code, default us) and optionally a city or state.Read-onlyGoogle Jobs Search
Searches public Google Jobs listings for a keyword or phrase, in a country (two-letter code, default us) and optionally a city or state. Returns one page of about 10 listings and saves nothing.
Inputs
keywordstringrequired- A google jobs search query.
countryCodestring- The two letter, lowercase country code to search google jobs for. For example, 'us' for United States. If not specified, it will search for jobs in the United States.default
us locationstring- The state or city to search google jobs for written in full. For example, 'San Francisco, California, United States' or 'California, United States'.
fuseaimcp_google_maps_search_toolSearches Google Maps for places matching a keyword or phrase; put the location in the query (e.g.Read-onlyGoogle Maps Search
Searches Google Maps for places matching a keyword or phrase; put the location in the query (e.g. 'coffee shops in Austin, TX'). Returns the matching places (name, address, phone, website, rating, opening hours) and saves nothing.
Inputs
keywordstringrequired- A google maps search query.
fuseaimcp_list_action_feedList items in the Smart Actions feed (the suggested next actions).Read-onlyList Smart Actions Feed
List items in the Smart Actions feed (the suggested next actions). Supports pagination, sort, free-text search, and filters (status, category, priority, actionType, sourceType, icpMatch, signalType). Each item carries its actionFeedId (as _id) and its actionId. Items with a null actionId (insights/alerts/reminders) cannot be used with the by-actionId tools.
Inputs
actionTypearray- No description.
assignedUserIdstring- No description.
categoryarray- No description.
createdFromstring- No description.
createdTostring- No description.
dueFromstring- No description.
dueTostring- No description.
icpIdarray- No description.
icpMatchboolean- No description.
isPinnedboolean- No description.
maxPointsinteger- No description.
minPointsinteger- No description.
pageinteger- No description.default
1 pageSizeinteger- No description.default
20 pinnedFirstboolean- No description.default
true priorityarray- No description.
searchstring- No description.
signalTypearray- No description.
sortBystring- No description.one of
createdAtupdatedAtdueAtprioritystatusdefaultcreatedAt sortOrderstring- No description.one of
ascdescdefaultdesc sourceTypearray- No description.
statusarray- No description.
userIdstring- No description.
fuseaimcp_list_active_campaignsList your active and initializing campaigns, with or without live list refresh — returns each one's campaignId, campaignName, listIds and listNames, activeStatus, isDynamic (true when live list refresh is on) and the channels its steps need (isEmailRequired, isLinkedInRequired, isDialerRequired).Read-onlyList Active Campaigns
List your active and initializing campaigns, with or without live list refresh — returns each one's campaignId, campaignName, listIds and listNames, activeStatus, isDynamic (true when live list refresh is on) and the channels its steps need (isEmailRequired, isLinkedInRequired, isDialerRequired). add_contacts_to_campaign_tool accepts only those whose activeStatus is active; a contact added to one of a campaign's lists is enrolled automatically only when isDynamic is true.
Inputs
This tool takes no inputs.
fuseaimcp_list_auto_reply_campaignsList the campaigns (campaignId + name) that have scheduled sends matching the filters, such as pending auto-replies; returns at most 500.Read-onlyList Auto-Reply Campaigns
List the campaigns (campaignId + name) that have scheduled sends matching the filters, such as pending auto-replies; returns at most 500. `statuses` is required: use ["approval_required"] (with sequenceStepTypes ["ai_reply"]) to mirror the pending-auto-replies view. Optional filters: campaignId, channels, sequenceStepTypes, targetContactId, fromEmail, scheduledDateFrom/To, sentDateFrom/To, includeDeleted.
Inputs
statusesarrayrequired- No description.
campaignIdstring- No description.
channelsarray- No description.
fromEmailstring- No description.
includeDeletedboolean- No description.
scheduledDateFromstring- No description.
scheduledDateTostring- No description.
sentDateFromstring- No description.
sentDateTostring- No description.
sequenceStepTypesarray- No description.
targetContactIdstring- No description.
fuseaimcp_list_automation_executionsList an automation's executions, newest first, with per-step status, summaries, timing, credits and fan-out counts.Read-onlyList Automation Executions
List an automation's executions, newest first, with per-step status, summaries, timing, credits and fan-out counts. Paginated (page, pageSize default 20, max 100).
Inputs
automationIdstringrequired- No description.
pageinteger- No description.default
1 pageSizeinteger- No description.default
20
fuseaimcp_list_automation_nodesList the automation node catalog this connector can save: each trigger, gate (condition) and action node with its nodeKey, entityType, the exact attributes it accepts, and its outputSchema.Read-onlyList Automation Nodes
List the automation node catalog this connector can save: each trigger, gate (condition) and action node with its nodeKey, entityType, the exact attributes it accepts, and its outputSchema. Read-only. Nodes that choose their operation only when the rule runs (calling a FuseAI tool by name, an action or trigger in a connected app, an AI agent step, running code) are not listed and cannot be saved through this connector. For an action node the outputSchema is what later steps bind with {{steps.<stepId>.output.<field>}}. For a trigger node it is the EVENT the rule receives, which is everything {{trigger.<field>}} can read: the entity id (contactId or companyId), occurredAt, attributes.<key> and meta.<key>; a contact's or company's profile (name, email, LinkedIn URL) is never on the event. Call this before creating or updating an automation: payloads are validated against this catalog. Optional step ("trigger"|"condition"|"action") and entityType ("contact"|"company"|"user") filters.
Inputs
entityTypestring- No description.one of
contactcompanyuser stepstring- No description.one of
triggerconditionaction
fuseaimcp_list_automationsList automations (id, name, summary, status, triggers, steps, health, last execution, executions in the last 7 days, userId of the owner).Read-onlyList Automations
List automations (id, name, summary, status, triggers, steps, health, last execution, executions in the last 7 days, userId of the owner). scope: 'mine' (the default) lists only the user's own rules: use it for 'my automations' and to find a rule the user names before pausing, resuming, editing or deleting it; 'team' also lists every teammate's rules. ownerId (one or more user ids) narrows to those owners, whatever the scope. Paginated via page (default 1) and pageSize (default 20, max 100); optional status filter (one or more of active|paused|auto_paused|draft|expired; auto_paused = paused automatically rather than by a user, e.g. after repeated failures, a credit shortfall, runs that found nothing new or a connected-app problem), type filter (scheduled = has a schedule trigger, trigger = fired by events only), lastRunResult filter (one or more latest-execution statuses, e.g. completed|failed, or not_run for rules that never ran) and free-text search. Each trigger and step carries attributeNames with the display names of its list, campaign and signal ids; other id-valued attributes (e.g. icpId) are returned raw — resolve those with the matching list tools when needed.
Inputs
lastRunResultarray- No description.
ownerIdarray- No description.
pageinteger- No description.default
1 pageSizeinteger- No description.default
20 scopestring- No description.one of
mineteamdefaultmine searchstring- No description.
statusarray- No description.
typestring- No description.one of
triggerscheduled
fuseaimcp_list_campaign_messagesList your campaigns' messages and steps by status and channel — most often the pending auto-replies awaiting approval (email and/or LinkedIn), but also scheduled, sent, skipped, stopped or failed sequence steps on any channel, including calls and to-dos.Read-onlyList Campaign Messages
List your campaigns' messages and steps by status and channel — most often the pending auto-replies awaiting approval (email and/or LinkedIn), but also scheduled, sent, skipped, stopped or failed sequence steps on any channel, including calls and to-dos. Returns each reply's scheduledEmailId (_id), subject, body, channel, and the email/chat it targets — needed to approve, modify, or skip them. `statuses` is required: use ["approval_required"] (with sequenceStepTypes ["ai_reply"]) for the pending-auto-replies view. Optional filters: campaignId, fromEmail, targetContactId, channels; paginated via page/limit (max 100) with sortBy/sortOrder.
Inputs
statusesarrayrequired- No description.
campaignIdstring- No description.
channelsarray- No description.
fromEmailstring- No description.
includeDeletedboolean- No description.
limitinteger- No description.
pageinteger- No description.
scheduledDateFromstring- No description.
scheduledDateTostring- No description.
sentDateFromstring- No description.
sentDateTostring- No description.
sequenceStepTypesarray- No description.
sortBystring- No description.one of
scheduledDateUTCsentDateUTCstatuschannelsequenceStepTypecreated_at sortOrderstring- No description.one of
ascdesc targetContactIdstring- No description.
fuseaimcp_list_campaign_target_contacts_toolLists a campaign's leads (target contacts) with full search, filtering (name, company, tags, statuses, steps sent, the arm an A/B split routed them to via splitArm), sorting and pagination.Read-onlyList Campaign Target Contacts
Lists a campaign's leads (target contacts) with full search, filtering (name, company, tags, statuses, steps sent, the arm an A/B split routed them to via splitArm), sorting and pagination. Returns each lead's targetContactId — use it for lead activity or activating/deactivating a lead.
Inputs
campaignIdstringrequired- No description.
clickCountinteger- No description.
clickCountMaxinteger- No description.
clickCountMininteger- No description.
companySearchstring- No description.
limitinteger- No description.
nameSearchstring- No description.
openCountinteger- No description.
openCountMaxinteger- No description.
openCountMininteger- No description.
outcomesarray- No description.
pageinteger- No description.
searchstring- No description.
sentBySearchstring- No description.
sentStepsinteger- No description.
sentStepsMaxinteger- No description.
sentStepsMininteger- No description.
sortBystring- No description.one of
fullNamecompanyNameschedulerStatusdisplayStatussentStepsfirstMessageReceivedAtopenCountclickCount sortOrderstring- No description.one of
ascdesc splitArmobject- Only the contacts an A/B split (splitStepId) routed to one arm ('a' or 'b').
statusesarray- No description.
tagsarray- No description.
templateTypesarray- No description.
fuseaimcp_list_campaign_templates_toolList the saved campaign templates the user can see (their own and, depending on the team's campaign visibility settings, teammates') WITH their ids (each template's _id, the value delete_campaign_templates_tool takes in campaignTemplateIds) — reusable campaign/sequence blueprints.Read-onlyList Campaign Templates
List the saved campaign templates the user can see (their own and, depending on the team's campaign visibility settings, teammates') WITH their ids (each template's _id, the value delete_campaign_templates_tool takes in campaignTemplateIds) — reusable campaign/sequence blueprints. Use to find a template id before deleting, or to show the user their templates.
Inputs
pageinteger- Page number (default 1)
pageSizeinteger- Templates per page (default 10, max 100)
searchstring- Optional template-name search text
fuseaimcp_list_campaigns_toolLists the user's outreach campaigns (name, status, metrics) with optional search and pagination.Read-onlyList Campaigns
Lists the user's outreach campaigns (name, status, metrics) with optional search and pagination. Use to find a campaign or report which campaigns exist. It does not rank campaigns by performance, and its metrics are all-time: since / to filter campaigns only by creation date.
Inputs
campaignStatusesarray- No description.
campaignTypearray- No description.
channelsarray- No description.
folderIdstring- No description.
isDynamicarray- No description.
listNamearray- No description.
pageinteger- No description.default
1 pageSizeinteger- No description.default
10 searchstring- No description.
sincestring- Only campaigns CREATED at or after this ISO 8601 instant. Not an activity window; metrics stay all-time.
tostring- Only campaigns CREATED at or before this ISO 8601 instant. Not an activity window; metrics stay all-time.
userIdarray- No description.
fuseaimcp_list_composio_connectionsList the apps the user has connected to FuseAI through Composio.Read-onlyList Composio Connections
List the apps the user has connected to FuseAI through Composio. Read-only. Returns connections, each with connectedAccountId, toolkit (the app's slug, e.g. googlecalendar, notion, zendesk), toolkitName, toolkitLogo and status (ACTIVE is usable; EXPIRED or FAILED needs a fresh connect link from create_composio_connect_link). Automations that use a connected app are built in the FuseAI app.
Inputs
This tool takes no inputs.
fuseaimcp_list_contact_notesList the notes the user's team wrote on one CRM contact (by contactId), newest first: each with noteId, body, author, createdAt, updatedAt and canEdit (true when the caller wrote it, so may update or delete it).Read-onlyList Contact Notes
List the notes the user's team wrote on one CRM contact (by contactId), newest first: each with noteId, body, author, createdAt, updatedAt and canEdit (true when the caller wrote it, so may update or delete it). Paginated via page and limit. Only notes written in Fuse are listed, not notes pulled in from a connected CRM such as HubSpot; both kinds appear on get_prospect_contact_activity_tool's timeline.
Inputs
contactIdstringrequired- CRM contact whose notes to list.
limitinteger- Notes per page.default
25 pageinteger- Page number.default
1
fuseaimcp_list_crm_ownersList the connected CRM's owners from FuseAI's stored copy, which sync_crm_owners refreshes (this tool does not call the CRM): each owner's _id (the crmOwnerId that set_crm_owner_mapping takes), crmUserId, name, email and the FuseAI user it is mapped to, or null.Read-onlyList CRM Owners
List the connected CRM's owners from FuseAI's stored copy, which sync_crm_owners refreshes (this tool does not call the CRM): each owner's _id (the crmOwnerId that set_crm_owner_mapping takes), crmUserId, name, email and the FuseAI user it is mapped to, or null. Needs the connected CRM integration's crmIdentifier and crmAccountId. Supports page/limit pagination.
Inputs
crmAccountIdstringrequired- No description.
crmIdentifierstringrequired- No description.one of
hubspotsalesforcezohoattiopipedrivegohighlevelnotion limitinteger- No description.default
20 pageinteger- No description.default
1
fuseaimcp_list_email_accountsList the email accounts the user owns or that teammates shared with them, in any state (including inactive, disconnected and needing re-authentication), with each account's status (isActive, isConnected, connectionMode, needsReauthAt, syncStatus), sender profile (name, title, signature, dailyCapacity), warmup status and its domain's SPF/DMARC/MX health, which is looked up live in public DNS when no result from the last 12 hours is cached.Read-onlyList Email Accounts
List the email accounts the user owns or that teammates shared with them, in any state (including inactive, disconnected and needing re-authentication), with each account's status (isActive, isConnected, connectionMode, needsReauthAt, syncStatus), sender profile (name, title, signature, dailyCapacity), warmup status and its domain's SPF/DMARC/MX health, which is looked up live in public DNS when no result from the last 12 hours is cached. Use a connected account's emailAddress (isConnected true) as the 'from' address when sending/replying, and as 'emailAddress' for AI email composition.
Inputs
This tool takes no inputs.
fuseaimcp_list_emailsList emails FuseAI has synced from the user's connected mailboxes (inbox or another label), read from FuseAI's stored copy without contacting the mail provider.Read-onlyList Inbox Emails
List emails FuseAI has synced from the user's connected mailboxes (inbox or another label), read from FuseAI's stored copy without contacting the mail provider. Mailboxes that are disconnected or need re-authentication are left out, and when the user has chosen to hide non-campaign emails, only emails with the user's campaign contacts are returned. Returns recent emails with ids, subjects, and senders. Each email's receiverEmailAddress is the connected mailbox it belongs to: for a reply or any other received email, the address that received it; for a sent, scheduled or draft email, the mailbox it is sent from, with the recipients in toRecipients, ccRecipients and bccRecipients. labelType picks the view (inbox, sent, drafts, important, unread, scheduled, snoozed, archived, deleted, spam, favorite, prospects). Optional filters: campaignId, isCampaignEmail (true: campaign emails only, false: non-campaign only; ignored when campaignId is set), contactName, mailBoxAccount, fromDate/toDate (ISO), searchVal (free text over names, subject and body); set filter="true" when scoping to a mailBoxAccount. tags and repliesOnly are accepted but have no effect on this list. Paginated via page (default 1) and limit (default 20, max 100).
Inputs
labelTypestringrequired- No description.one of
inboxsentdraftsimportantunreadscheduledsnoozedarchiveddeletedspamfavoriteprospects campaignIdstring- No description.
contactNamestring- No description.
filterstring- No description.one of
truefalse fromDatestring- No description.
isCampaignEmailboolean- No description.
limitinteger- No description.default
20 mailBoxAccountstring- No description.
pageinteger- No description.default
1 repliesOnlyboolean- No description.
searchValstring- No description.
tagsstring- No description.
toDatestring- No description.
fuseaimcp_list_icpsList the user's Ideal Customer Profiles (ICPs), each with its id, name and criteria.Read-onlyList ICPs
List the user's Ideal Customer Profiles (ICPs), each with its id, name and criteria. An ICP with a deactivatedAt is switched off: it is kept but applied nowhere until the user switches it back on.
Inputs
This tool takes no inputs.
fuseaimcp_list_knowledge_hub_profilesList the Knowledge Hub profiles (company knowledge / ICP context configured in settings) of the user and, unless includeTeam is false, of their teammates.Read-onlyList Knowledge Hub profiles
List the Knowledge Hub profiles (company knowledge / ICP context configured in settings) of the user and, unless includeTeam is false, of their teammates.
Inputs
defaultFirstboolean- No description.default
false includeTeamboolean- No description.default
true limitinteger- No description.default
20 pageinteger- No description.default
1 searchstring- No description.
sortBystring- No description.one of
createdAtprofileNameupdatedAt sortOrderstring- No description.one of
ascdesc
fuseaimcp_list_linkedin_accountsList the user's LinkedIn accounts and the accounts teammates shared with them (isOwner false).Read-onlyList Connected LinkedIn Accounts
List the user's LinkedIn accounts and the accounts teammates shared with them (isOwner false). Returns each account's accountId and providerId, with its status, limits and signature.
Inputs
This tool takes no inputs.
fuseaimcp_list_linkedin_chatsList the user's LinkedIn conversations from FuseAI's synced copy, without contacting LinkedIn: chats of the user's connected LinkedIn accounts and of accounts teammates have shared with them for inbox use.Read-onlyList LinkedIn Chats
List the user's LinkedIn conversations from FuseAI's synced copy, without contacting LinkedIn: chats of the user's connected LinkedIn accounts and of accounts teammates have shared with them for inbox use. When the user has chosen to hide non-campaign LinkedIn messages, only chats from campaigns are returned. Each chat includes its accountId and chatId. tab filters only when set to "unread" (chats with unread messages), sortBy only sorts when set to "lastMessage" (by last-message time; sortOrder asc or desc), and searchVal searches message text.
Inputs
attendeeProfileURLstring- No description.default
campaignIdstring- No description.
contactNamestring- No description.
endDatestring- No description.
limitinteger- No description.default
10 linkedinAccountIdstring- No description.
linkedinAccountIdsarray- No description.
pageinteger- No description.default
1 searchValstring- No description.
sortBystring- No description.
sortOrderstring- No description.one of
ascdesc startDatestring- No description.
tabstring- No description.
fuseaimcp_list_linkedin_connectionsList the first-degree LinkedIn connections that FuseAI has synced for the user's LinkedIn accounts, including accounts a teammate shared with them with inbox access.Read-onlyList LinkedIn Connections
List the first-degree LinkedIn connections that FuseAI has synced for the user's LinkedIn accounts, including accounts a teammate shared with them with inbox access. It reads FuseAI's stored copy and does not contact LinkedIn, so very recent connections can be missing until the next sync. filter matches part of a connection's name. Returns each connection's accountId, name, headline, profile and picture URLs, publicIdentifier and profileId.
Inputs
accountIdstring- No description.
filterstring- No description.
limitinteger- No description.default
10 pageinteger- No description.default
1
fuseaimcp_list_linkedin_invitationsList the pending LinkedIn connection invitations (received or sent) that FuseAI has synced for the user's LinkedIn accounts, including accounts a teammate shared with them with inbox access.Read-onlyList LinkedIn Invitations
List the pending LinkedIn connection invitations (received or sent) that FuseAI has synced for the user's LinkedIn accounts, including accounts a teammate shared with them with inbox access. It reads FuseAI's stored copy and does not contact LinkedIn, so it can lag LinkedIn until the next sync. contactName and publicId filter only when type is set. Each item's id is its invitationId; received invitations also include the sharedSecret and provider that respond_to_linkedin_invitation needs. On a sent invitation, inviterId, inviterName and inviterDescription describe the person invited.
Inputs
accountIdstring- No description.
contactNamestring- No description.
limitinteger- No description.default
10 pageinteger- No description.default
1 publicIdstring- No description.
typestring- No description.one of
sentreceived
fuseaimcp_list_prospect_lists_toolLists the prospect lists the user can see (their own plus teammates' public lists, contact and company) with their names and ids; supports search-by-name, sorting and pagination.Read-onlyList Prospect Lists
Lists the prospect lists the user can see (their own plus teammates' public lists, contact and company) with their names and ids; supports search-by-name, sorting and pagination. System lists are included by default; pass filters.includeSystemLists=false to leave out the synthetic ones (All Saved Contacts, All <CRM> Leads, All Enriched Contacts). Stored system lists such as All Website Visitors are returned either way. Stored lists return their 24-character id, usable wherever a listId is required; synthetic system lists return a symbolic id (e.g. all_enriched_contacts), which only tools that accept symbolic list ids, such as get_prospect_list_rows_tool, take.
Inputs
filtersobject- No description.
limitinteger- No description.default
20 pageNuminteger- No description.default
1 sortBystring- No description.default
createdAt sortOrderstring- No description.one of
ascdescdefaultdesc
fuseaimcp_list_provider_keysList the Deep Research providers the user has saved their own API key for (provider id + metadata; keys themselves are never returned).Read-onlyList Provider Keys
List the Deep Research providers the user has saved their own API key for (provider id + metadata; keys themselves are never returned).
Inputs
This tool takes no inputs.
fuseaimcp_list_rented_phone_numbersList the power-dialer phone numbers you can use or manage (Twilio — the active provider), both rented numbers and verified caller-IDs: _id, phoneNumber, name, numberType, assignee, health/status.Read-onlyList Rented Phone Numbers
List the power-dialer phone numbers you can use or manage (Twilio — the active provider), both rented numbers and verified caller-IDs: _id, phoneNumber, name, numberType, assignee, health/status. Team owners and admins see every team member's numbers; other members see only the numbers assigned to them. Health is the last saved snapshot; this reads FuseAI's stored records and does not contact Twilio. Each row's _id is the value to pass as phoneNumberId to deactivate_twilio_phone_number.
Inputs
This tool takes no inputs.
fuseaimcp_list_saved_searchesList the user's saved searches, including their ids (needed to update or delete a saved search).Read-onlyList Saved Searches
List the user's saved searches, including their ids (needed to update or delete a saved search). Without searchType, or with 'contacts', it lists people searches plus the user's active ICP filters as presets (source "icp"), which cannot be updated or deleted here; company searches are listed only with searchType 'companies'. People searches come back in the search tools' filter vocabulary, ready to pass to search_people; ignoredFilterKeys names any stored key that has no search filter.
Inputs
searchTypestring- Filter to people ('contacts') or company saved searches. Omit for all.one of
contactscompanies
fuseaimcp_list_sending_domainsList the FuseAI-purchased sending domains the user can see (their own, plus teammates' when the team's domain visibility setting allows), with _id, domain, status, forwardedTo, active mailboxCount, purchaseDate and expiryDate.Read-onlyList Sending Domains
List the FuseAI-purchased sending domains the user can see (their own, plus teammates' when the team's domain visibility setting allows), with _id, domain, status, forwardedTo, active mailboxCount, purchaseDate and expiryDate. Optional status (active|archived|deleted|processing|failed)/search filters; paginated via page/pageSize.
Inputs
pageinteger- No description.default
1 pageSizeinteger- No description.default
10 searchstring- No description.
statusstring- No description.one of
activearchiveddeletedprocessingfailed
fuseaimcp_list_signal_monitorsList the user's signal monitors (agents).Read-onlyList signal monitors
List the user's signal monitors (agents). Use this to find a monitor's id and status before activating or pausing it. For a people or companies watcher, errorMessage says why it failed, and progress.message says how its setup went, including how many companies or contacts the credit cap leaves out.
Inputs
accountTagsstring- Company Criteria terms, JSON-encoded like listName, matched against each of the agent's account tags.
contactsFoundMaxinteger- Most leads found.
contactsFoundMininteger- Fewest leads found.
createdAtFromstring- Created on or after this ISO instant.
createdAtTostring- Created on or before this ISO instant.
expirationDateFromstring- End date on or after this ISO instant; agents without one drop out.
expirationDateTostring- End date on or before this ISO instant; agents without one drop out.
folderIdstring- No description.
leadTagsstring- Target ICP terms, JSON-encoded like listName, matched against each of the agent's lead and keyword tags.
limitinteger- Max number of monitors to return.
listNamestring- Agent name terms, JSON-encoded as {"include":[],"exclude":[]}: names holding any include term and no exclude term.
listTypestring- Kind of CRM list the agents fill.one of
peoplecompany monitorTypestring- Comma-separated monitorType slugs.
pageinteger- No description.
searchstring- Optional text to filter monitors by name.
sortBystring- No description.one of
listNamemonitorTypestatusexpirationDatemaxContactscontactsFoundcreatedAtupdatedAtlistType sortOrderstring- No description.one of
ascdesc statusstring- Comma-separated statuses, as the agents table shows them: an agent past its end date reads as expired.
fuseaimcp_list_tasksList the Tasks hub rows the user may see (their own and their teammates' open work by default): to-dos, calls, emails, LinkedIn steps, call batches and AI-reply reviews, each with id, type, status, title, due date, priority, assignee, contact and source.Read-onlyList Tasks
List the Tasks hub rows the user may see (their own and their teammates' open work by default): to-dos, calls, emails, LinkedIn steps, call batches and AI-reply reviews, each with id, type, status, title, due date, priority, assignee, contact and source. Filter by assignedToUserIds (ids from list_team_members), statuses (default open), taskTypes, priorities, buckets, campaignIds or sourceKinds, or search by title / contact / company text; paginated via page and pageSize. Use it to find a task id before update_task or complete_task.
Inputs
assignedToUserIdsstring- Only tasks assigned to these teammate user ids (from list_team_members); omitted = everyone the caller may see.
blockedboolean- true narrows to tasks the assignee cannot execute right now, e.g. a missing channel.
bucketsstring- One or more due-date buckets to include.
campaignIdsstring- Only tasks that belong to these campaign ids.
pageinteger- Page number.default
1 pageSizeinteger- Tasks per page.default
50 prioritiesstring- One or more priorities to include.
searchstring- Case-insensitive text match on the task title, contact name or company name.
sortBystring- Column to sort by; omitted = the product default.one of
taskcontactcompanystatusdueprioritysourceowner sortOrderstring- Sort direction.one of
ascdescdefaultasc sourceKindsstring- One or more sources to include (where the task came from).
statusesstring- One or more task statuses to include; defaults to open.
taskTypesstring- One or more task types to include.
timezonestring- IANA time zone used to compute due-date buckets, e.g. America/New_York.
fuseaimcp_list_team_invitationsList the pending invitations the user has received to join another FuseAI team (matched on their email), with the inviting owner's name; at most 10, newest first.Read-onlyList Team Invitations
List the pending invitations the user has received to join another FuseAI team (matched on their email), with the inviting owner's name; at most 10, newest first. Each carries its invitation id (the `_id`), which respond_to_team_invitation takes. Invitations the user's own team has sent are not listed here; list_team_members shows them with status INVITED.
Inputs
This tool takes no inputs.
fuseaimcp_list_team_membersList the user's team / workspace: its members (role, credits cap, credits used) and its invitations that are pending or were declined (status INVITED or REJECTED; for these `_id` is the invitation id and userId is null).Read-onlyList Team Members
List the user's team / workspace: its members (role, credits cap, credits used) and its invitations that are pending or were declined (status INVITED or REJECTED; for these `_id` is the invitation id and userId is null). Returns at most 20 entries, newest first; pagination.totalRecords gives the full count.
Inputs
This tool takes no inputs.
fuseaimcp_list_tracking_domains_toolLists the team's custom email tracking domains (id, hostname, status, isDefault).Read-onlyList Tracking Domains
Lists the team's custom email tracking domains (id, hostname, status, isDefault). ACTIVE domains can be selected on a campaign via trackingDomainId. If a campaign's trackingDomainId is null, send-time click tracking still uses the team's default ACTIVE domain automatically. Click tracking must also be enabled or links are not rewritten onto the domain. Create/verify stays in Settings — this tool is read-only.
Inputs
activeOnlyboolean- When true, only ACTIVE domains that can carry a click-tracked link today. Default false (all non-retired).default
false
fuseaimcp_lookup_filter_valuesResolve free text to canonical region, industry, title, or school values, looked up live from Fuse's external people-data provider.Read-onlyLook up canonical filter values
Resolve free text to canonical region, industry, title, or school values, looked up live from Fuse's external people-data provider. Returns the top matching canonical values — use these exact strings rather than inventing them. Pass filterType (region|industry|title|school) and a query.
Inputs
filterTypestringrequired- Which canonical filter vocabulary to search: region, industry, title, or school.one of
regionindustrytitleschool querystringrequired- Free-text term to resolve to canonical values (e.g. 'software', 'united states', 'vp of sales').
countinteger- Maximum number of canonical values to return (default 10).default
10
fuseaimcp_prospecting_analyticsGet prospecting analytics over a date window: emails enriched, phones enriched, contacts saved and contacts searched, each as a total and a daily, weekly, monthly or quarterly series bucketed in timezone (IANA, default UTC).Read-onlyProspecting Analytics
Get prospecting analytics over a date window: emails enriched, phones enriched, contacts saved and contacts searched, each as a total and a daily, weekly, monthly or quarterly series bucketed in timezone (IANA, default UTC). Without userId it covers the user and the teammates whose campaigns they can see; userId narrows it to one such teammate. Read-only. `since` and `to` are ISO 8601 date-time instants (e.g. 2026-08-01T00:00:00.000Z) — do not send a date-only YYYY-MM-DD. `since` is optional (omit for lifetime). `to` is optional and defaults to now.
Inputs
sincestring- No description.
timezonestring- No description.default
UTC tostring- No description.
userIdstring- No description.
fuseaimcp_query_relationship_historyAsk the Fuse Brain — living relationship files for ENGAGED contacts/companies, built from the team's real emails, LinkedIn messages, calls and meetings.Read-onlyQuery Relationship History
Ask the Fuse Brain — living relationship files for ENGAGED contacts/companies, built from the team's real emails, LinkedIn messages, calls and meetings. Use it for questions about existing relationships: history, needs, objections, commitments, deal state ('what does Dana care about?', 'which contacts mentioned budget?', 'where did the Acme deal stall?', 'who is waiting on us?'). Answers cite the contacts/companies used. Not for cold prospects: no file exists until a contact engages. Works only when the team has Fuse Brain switched on (Settings -> AI Learning).
Inputs
questionstringrequired- The question about tracked contacts/companies, self-contained (the tool does not see the chat). General ('which contacts mentioned budget concerns?', 'who is closest to buying?') or specific ('what does Dana at Acme care about?', 'where did the Initech deal stall?').
focusNamestring- The contact or company name/domain the question is about, when it targets someone specific. Omit for general questions.
fuseaimcp_search_composio_toolkitsSearch Composio's catalog of apps the user can connect to FuseAI, by name (query: search, e.g.Read-onlySearch Composio Toolkits
Search Composio's catalog of apps the user can connect to FuseAI, by name (query: search, e.g. 'google calendar'; optional limit, default 20, max 50). Read-only. Returns toolkits, each with the app's slug, name, description, logo and website (appUrl). Only apps the user can connect through Composio's hosted sign-in (OAuth, or an API key they enter there) are returned, and never an app FuseAI integrates natively (Gmail, Outlook, Slack, LinkedIn). Use the slug as `toolkit` in create_composio_connect_link.
Inputs
searchstringrequired- No description.
limitinteger- No description.default
20
fuseaimcp_search_web_or_read_pageSearches the public web or reads one page, in one of two modes, and saves nothing.Read-onlySearch Web or Read Page
Searches the public web or reads one page, in one of two modes, and saves nothing. (1) SEARCH: pass `query` (a search string) to get matching pages' titles, links and snippets. (2) SCRAPE: pass `url` (a single page URL) to fetch that page's full content as markdown. Never pass both `query` and `url`.
Inputs
querystring- A general web search query for the public internet. Supports search operators like site:, intitle:, filetype:. Use EITHER `query` OR `url`, never both.
urlstring- A single page URL to fetch and read in full as markdown. Use when you already know the exact page that holds the answer (e.g. the user gave you a link, or a prior search surfaced the right page) — ideal for reading every person/item listed on a known page.
fuseaimcp_suggest_list_column_mapping_toolSuggests, with an AI model, how the columns of sample rows you hold (a file, CSV or scraped table) map to the row fields import_people_to_list and import_companies_to_list take (apiType upload-to-list with listType), and for each column with no field, whether to fill one of existingListId's own columns, add a new column, or ignore it.Read-onlySuggest List Column Mapping
Suggests, with an AI model, how the columns of sample rows you hold (a file, CSV or scraped table) map to the row fields import_people_to_list and import_companies_to_list take (apiType upload-to-list with listType), and for each column with no field, whether to fill one of existingListId's own columns, add a new column, or ignore it. With apiType bulk-enrichment it instead maps the columns to the person fields enrich_contact and single_enrichment_tool take inside input other than domain (name, firstName, lastName, url, company, location, contactId, customContactId), so a company-domain column is never mapped to input.domain, and it makes no column decisions. Returns mapping, customColumnDecisions and requiredNameMappings. It reads only the rows passed and existingListId's columns, and the AI model only works out the mapping from them: it searches no web source or data provider, looks no one up, contacts no one and calls no connected app. It saves nothing and costs no credits; use it before those imports when the source columns do not match their fields.
Inputs
apiTypestringrequired- Which API the mapping is forone of
upload-to-listbulk-enrichment inputsarrayrequired- Sample rows (records of input column name -> value) to analyse for the mapping
existingListIdstring- Optional existing list id whose columns should be considered
listTypestring- Type of the destination list — required when apiType is upload-to-listone of
contactListcompanyList
fuseaimcp_test_website_tagRead-only: fetch the given website's page once from FuseAI's servers and check whether the team's visitor-tracking tag appears in its HTML.Read-onlyTest Website Tracking Tag
Read-only: fetch the given website's page once from FuseAI's servers and check whether the team's visitor-tracking tag appears in its HTML. Only the initial HTML is read and no JavaScript runs, so a tag added by a tag manager or another script is not detected. An empty result means the tag was found; otherwise an error code (for example SCRIPT_NOT_LOADED or PAGE_FETCH_FAILED) says why.
Inputs
websitestringrequired- The website URL/domain to check the tracking tag on, e.g. 'acme.io'.
fuseaimcp_create_campaign_template_toolSave an existing campaign (its sequence steps, sending schedule, source lists and topic) as a new reusable template under a given name.WriteCreate Campaign Template
Save an existing campaign (its sequence steps, sending schedule, source lists and topic) as a new reusable template under a given name.
Inputs
campaignIdstringrequired- The campaign _id (24-character Mongo id) to save as a template
campaignTemplateNamestringrequired- Name for the new campaign template
fuseaimcp_create_composio_connect_linkCreate a sign-in link for connecting one app to FuseAI (body: toolkit, the app's slug from search_composio_toolkits).WriteCreate Composio Connect Link
Create a sign-in link for connecting one app to FuseAI (body: toolkit, the app's slug from search_composio_toolkits). Returns redirectUrl, valid for about ten minutes: show it to the user; they authorise the app there (or, for an app without OAuth, enter their API key, token or login there) and come back to FuseAI. Nothing is connected until they finish; from then on, automations the user builds in the FuseAI app can read and change data in that app as the user until they disconnect it in Settings > Integrations. 422 COMPOSIO_TOOLKIT_IN_HOUSE: the app is a native FuseAI integration (Gmail, Outlook, Slack, LinkedIn) that the user connects in the integrations tab. A CRM (HubSpot, Salesforce, Pipedrive, Zoho, Attio) is an ordinary Composio app: this link is a second, per-user sign-in, separate from the team's CRM Synchronization connection.
Inputs
toolkitstringrequired- No description.
fuseaimcp_create_contact_duplicates_reviewSave a finished review job (kept for 24 hours; one the user already saved for the same contact and nightly check is reused) holding the duplicate groups Fuse's nightly duplicate check found for one contact (contactId).WriteCreate Contact Duplicates Review
Save a finished review job (kept for 24 hours; one the user already saved for the same contact and nightly check is reused) holding the duplicate groups Fuse's nightly duplicate check found for one contact (contactId). Apart from saving that job it changes nothing: it does not merge, archive, edit or delete any contact or list. Answers { data: { jobId } }: read the groups with get_duplicate_contacts_job, and pass the jobId as scanJobId to merge_duplicate_contact_groups only for groups the user approves. Answers NO_DUPLICATES_FOUND when the check flagged none for the contact.
Inputs
contactIdstringrequired- The CRM contact id (24-character Mongo id)
fuseaimcp_create_email_tagCreate a new FuseAI inbox tag with a name and hex color (6-digit hex with leading '#', e.g.WriteCreate Email Tag
Create a new FuseAI inbox tag with a name and hex color (6-digit hex with leading '#', e.g. #FF0000). The tag is stored in FuseAI only; nothing is created in the connected Gmail, Outlook or SMTP mailbox.
Inputs
colorstringrequired- No description.
namestringrequired- No description.
fuseaimcp_download_list_csv_toolGenerates the CSV export of a list and returns the raw CSV text (large lists produce large outputs).WriteDownload List CSV
Generates the CSV export of a list and returns the raw CSV text (large lists produce large outputs). listId is a stored list's id or one of the symbolic lists all_enriched_contacts, all_saved_contacts, all_crm_records and my_crm_records (with a connected CRM: the team's contacts FuseAI has synced with it, or those the user owns there). The other symbolic ids the schema accepts are refused with LIST_NOT_FOUND: all_relations, all_received_invitations and all_sent_invitations (LinkedIn connection and invitation lists), job_changed_contacts, all_crm_companies and my_crm_companies. It reads only data FuseAI has already stored and does not contact LinkedIn, the CRM or any other outside service. A contact list over 20,000 rows is not returned: the call instead starts a background export that saves the CSV as a file on FuseAI's servers for 24 hours and returns only a job id and the row count, which no tool in this plugin can read.
Inputs
listIdstringrequired- No description.
fuseaimcp_duplicate_signal_monitorDuplicate an existing signal monitor into a new DRAFT.WriteDuplicate signal monitor
Duplicate an existing signal monitor into a new DRAFT. The draft does not run or cost credits until it is activated with activate_signal_monitor. Some advanced filters may need to be re-supplied before activating.
Inputs
documentIdstringrequired- The signal monitor's documentId.
listNamestring- Optional name for the duplicate's destination list. Defaults to '<source> (Copy)'.
fuseaimcp_generate_email_summaryAI-summarize an email thread, identified by its emailId (the email's id from the user's inbox).WriteSummarize email thread
AI-summarize an email thread, identified by its emailId (the email's id from the user's inbox). It summarizes that email and the earlier messages quoted in its body, so pass the newest email of the thread. The first summary is saved on the email record (list_emails then returns it as aiSummary), and later calls for that email, including a teammate's on a shared mailbox, return the saved summary instead of a new one.
Inputs
emailIdstringrequired- The id (_id) of the email/thread to summarize, from the user's inbox.
fuseaimcp_scrape_webScrape the web and return the extracted items as a structured list.WriteScrape Web
Scrape the web and return the extracted items as a structured list. source: {kind:'page', url} for one page, {kind:'pages', urls} for up to 10, or {kind:'search', query, includeDomains?, excludeDomains?, recencyMinutes?, domainType?:'web'|'news'} to search first. extract.outputSchema is a JSON Schema object with an `items` array whose elements have at least `url` (string); extract.prompt says what counts as an item and what to keep (topics, keywords, exclusions). Optional strategy (defaults to 'auto' when omitted, so a call without it can be charged: 'auto' fetches the pages and, if they give fewer than maxItems items, sends to the agent the pages that came back empty or behind a bot or login wall, or uses the agent in place of the fetch when the fetch provider rejects FuseAI's account; 'fetch' never uses the agent and is never charged; 'agent' sends the pages straight to the agent), filters.keywords (hard filter) and filters.cssScope. maxItems 1-25. Optional exclude: values (urls) already collected — the extractor and the agent are told to look past them and return the items after them, so a repeated scrape of the same list moves forward instead of returning the same head. Returns { items, strategyUsed, agentSteps, creditsCharged }. The agent is a third-party AI research agent, not a browser: it searches the web and reads page content without signing in. It reads at most 3 pages per call: with 'agent', or when 'auto' uses it in place of the fetch, only the first 3 pages or search hits can be read, and in 'auto' at most 3 of the empty or walled pages are sent to it; pages past that cap are skipped without an error and add no items. Fetch and search cost no credits; the agent tier costs 5 credits per page it reads (so at most 15 credits per call), runs only when the user's credit balance covers those pages, and its cost is debited from that balance after a successful run; each call is charged on its own, so a repeated call that uses the agent is charged again. Apart from that credit charge it saves or changes nothing: the items are not saved to a FuseAI list or record, and nothing is posted or submitted to the sites.
Inputs
extractobjectrequired- No description.
maxItemsintegerrequired- No description.
sourcestringrequired- No description.
excludearray- No description.
filtersobject- No description.
strategystring- No description.one of
autofetchagent
fuseaimcp_search_companiesSearch a global company database from an external data provider (not only the user's saved companies) and RETURN the matched companies (name, domain, industry, size, location, and whether the user already saved each) plus a cursor for paging.WriteSearch Companies
Search a global company database from an external data provider (not only the user's saved companies) and RETURN the matched companies (name, domain, industry, size, location, and whether the user already saved each) plus a cursor for paging. It charges the user 1 credit per company returned and saves or changes nothing else. Pass a structured `filters` object; industry/tag/country filter values must be valid catalog values, which get_company_search_filter_options lists, and the investors filter takes an investor's domain, not its name. Use get_company_filter_autocomplete to resolve company names for nameLike and, with orgType=investor, investor domains for investors; use get_location_filter_autocomplete for the place names headquartersCity takes. The filter descriptions in the input schema cite REST routes that are not tools here: GET /companies/filter-options stands for get_company_search_filter_options, which returns every list they cite except `technologies`; GET /companies/autocomplete?field=company, with or without orgType=investor, stands for get_company_filter_autocomplete (its matches give `preferred_name` and `domains`); and field=location stands for get_location_filter_autocomplete. No tool serves field=state or the `technologies` list, so send headquartersStateName values as full state or province names, as is (e.g. Florida), and technologies values as names from the provider's technology vocabulary in their exact spelling and casing (e.g. Salesforce, Next.js, AWS). Send at least one filter (exclusions alone are fine): an empty filter set is refused with 422 SEARCH_FILTERS_INVALID. A balance below 1 credit per requested result (`size`, default 25) is refused with 402 INSUFFICIENT_CREDITS.
Inputs
filtersobjectrequired- No description.
cursorstring- No description.
sizeinteger- No description.default
25
fuseaimcp_search_peopleSearch an external data provider's database of professional profiles (not the user's saved contacts) and RETURN the matched people (name, title, company, location, LinkedIn) plus a scroll_token for paging.WriteSearch People
Search an external data provider's database of professional profiles (not the user's saved contacts) and RETURN the matched people (name, title, company, location, LinkedIn) plus a scroll_token for paging. Use this to actually see results — it returns them without saving anything to a list. The first 200 results of a search cost no credits, up to a daily allowance set by the account's plan; each result after them costs 1 credit. Apart from that charge and the free results it counts against the daily allowance, it changes nothing: no list, contact or saved search is created or updated. To save people it found, pass their linkedinUrl to save_people_to_list.
Inputs
filtersobjectrequired- People-search filters as FLAT dotted keys (e.g. job_title_levels, location_region, education.school.name — never a nested education/experience object). Include/exclude fields take arrays of {status: include|exclude, value} objects; a bare string or an array of strings are includes, and an exclude is always an entry. location_name/skills/education.school.name take string arrays; keywords is a string array; numeric ranges go under rangeInputs. Unknown keys are rejected, inside company_filters too. These filters are no longer served and are ignored if sent, so they never narrow the search: sex, github_url, github_username, facebook_url, job_title_sub_role, experience.title.sub_role, experience.company.industry, experience.company.location.country, experience.company.location.name, experience.company.location.continent, experience.company.type, location_continent, job_company_location_continent.
pageSizeinteger- How many results to return, at most 100. Defaults to 10.
scrollTokenstring- The scroll_token from a previous result, to fetch the next page.
fuseaimcp_start_duplicate_contacts_scanStart a background scan of the team's saved contacts (every contact in the lists the user can see) for duplicates, matching on matchFields: any of linkedinUrl, email, phone, fullName, jobTitle, location, industry, companyName and companyDomain.WriteStart Duplicate Contacts Scan
Start a background scan of the team's saved contacts (every contact in the lists the user can see) for duplicates, matching on matchFields: any of linkedinUrl, email, phone, fullName, jobTitle, location, industry, companyName and companyDomain. Pick at least one of linkedinUrl, email or phone, or at least three of the others; weaker picks group unrelated people and are refused. Answers { data: { jobId } } at once; the duplicate groups and totals are stored on that job once the scan completes. Starting the same scan while one is still running answers the running job.
Inputs
matchFieldsarrayrequired- Fields two contacts must share to count as duplicates. Include at least one of linkedinUrl, email or phone, or at least three of the other fields.
fuseaimcp_validate_email_addressesValidate (verify) email addresses: whether mail to each address the user gives would be delivered or bounce, whether or not the address belongs to a contact.WriteValidate Email Addresses
Validate (verify) email addresses: whether mail to each address the user gives would be delivered or bounce, whether or not the address belongs to a contact. Each address gets a live mailbox check from ZeroBounce, an outside email verification service, against the address's mail server (no email is sent to it), and a verdict: valid (the mailbox exists and accepts mail), invalid (mail to it bounces), risky (it may be delivered but the mailbox cannot be confirmed or should not be mailed: a catch-all domain, a role, disposable or abuse address), unknown (the mail server gave no answer) or not_checked (the check failed), with the checks behind it (format, mail servers, mailbox found, disposable, role address, catch-all). Takes 1 to 25 addresses per call; a repeated address is checked once, and a malformed one refuses the whole call before any address is checked or charged, naming the entry. Costs 10 credits per address checked and nothing for one not checked; when the balance cannot cover every address, nothing is checked. Apart from that credit charge, nothing is saved or changed: the deliverability statuses stored on contacts stay as they were.
Inputs
emailsarrayrequired- The email addresses to check, 1 to 25, each a well-formed email address; a repeated address is checked once.
fuseaimcp_activate_campaign_lead_toolResume one stopped lead's sequence inside a campaign, so the campaign sends them its remaining steps again, and report the step that runs next.DestructiveResume Campaign Lead
Resume one stopped lead's sequence inside a campaign, so the campaign sends them its remaining steps again, and report the step that runs next. Their next step is written right away: an AI-written or AI-personalized step costs 10 credits, and if credits or a usable sender account are missing the whole campaign is paused instead, with an in-app notice (and a Slack message if Slack is connected). It resumes the lead whatever stopped them, an unsubscribe included, without removing them from Unsubscribed Contacts; use resubscribe_contact_tool to reverse an unsubscribe. Takes a targetContactId from the campaign's lead list.
Inputs
targetContactIdstringrequired- No description.
fuseaimcp_activate_signal_monitorActivate a DRAFT signal monitor (one just created, in 'initializing' status) by its documentId so it starts running.DestructiveActivate signal monitor
Activate a DRAFT signal monitor (one just created, in 'initializing' status) by its documentId so it starts running. Only a draft can be activated, so once activation has succeeded a repeated call is refused and does not start or charge the monitor again. Activation creates its destination list, starts its watches, searches or scrapes at external data providers (CrustData, Fiber, Exa or Apify, depending on the monitor type), and charges credits as it sets up and runs (it can be refused when the balance is too low). A people, companies or job-changes watcher sends the watched contacts' LinkedIn URLs, or the watched companies' websites or LinkedIn pages, to Fiber. If the draft has email or phone enrichment on, the people it adds to its list are also looked up through FuseAI's contact enrichment, which checks FuseAI's own data and a waterfall of outside email and phone data providers, and each email or phone number found is saved on the contact at extra credits. Each new match is added to that list and to the user's saved companies or contacts (one the user had removed is saved again), can update a contact that already exists (a job-changes watcher rewrites the contact's job title and company), and can trigger the user's automations, which may add people to campaigns or act in connected apps. For a saved company, a background lookup with Fiber can fill empty fields on its shared record and replace a name that is still just its domain.
Inputs
documentIdstringrequired- The signal monitor's documentId.
fuseaimcp_add_contacts_to_campaign_toolAdds specific saved contacts, by CRM contact id (1-60 per call), straight into an ACTIVE campaign, without adding them to any of its lists, and starts its sequence for the ones added: the campaign begins sending its emails and LinkedIn actions to them from the user's connected accounts, and AI writing in those messages is charged in credits as each message is written.DestructiveAdd Contacts To Campaign
Adds specific saved contacts, by CRM contact id (1-60 per call), straight into an ACTIVE campaign, without adding them to any of its lists, and starts its sequence for the ones added: the campaign begins sending its emails and LinkedIn actions to them from the user's connected accounts, and AI writing in those messages is charged in credits as each message is written. With enrichMissingContactData, contacts without a valid work email are first looked up, charged in credits for each email found. Answers one outcome per contact: added, already_in_campaign, missing_valid_email (the campaign sends email and the contact has no valid work email), missing_linkedin (the campaign has LinkedIn steps and the contact has no LinkedIn URL), contact_not_found or enrolment_failed. Refused with CAMPAIGN_NOT_ACTIVE for a campaign that is not active (a draft or a paused one), NO_ACTIVE_EMAIL_ACCOUNTS or NO_ACTIVE_LINKEDIN_ACCOUNTS when no sending account is active for a channel the campaign uses, and CAMPAIGN_LIST_FULL when the campaign holds as many contacts as it can.
Inputs
campaignIdstringrequired- The campaign to add the contacts to; it must be active.
contactIdsarrayrequired- CRM contact ids (24 hex) of the people to add: 1 to 60, each once.
enrichMissingContactDataboolean- When the campaign sends email, first run an email enrichment for each contact without a valid work email, as the app's Add to Campaign does (charged at the enrichment price, only when an email is found). Left false, those contacts are not added and come back as missing_valid_email.default
false
fuseaimcp_add_list_rows_toolAdds existing contacts or companies (by id) to a prospect list.DestructiveAdd List Rows
Adds existing contacts or companies (by id) to a prospect list. Where the list is used, new rows have these effects: an active dynamic campaign that uses it as a source enrolls eligible new contacts and starts messaging them at its next refresh (within 30 minutes), the user's 'Added to list' and 'Company added to list' automations run for each new contact or company, and an active job-change watcher on the list starts tracking the new contacts' LinkedIn profiles with an outside data provider at its next daily sync (billed per check). People and companies watchers that use the list do not pick up added rows until their source lists are next replaced with update_signal_monitor_source_lists. A campaign that uses the list to exclude people leaves the new contacts, or the new companies' contacts, out of its later enrollments; people it already enrolled stay enrolled. Added companies not yet enriched are looked up by domain with an outside company-data provider in the background, and the result is written onto the shared company record every workspace that saved the company sees: it adds the provider's details, fills empty fields, and replaces a name that is still just the domain. Provide the listId plus the ids under `contacts` or `companies` as [{ action: 'add', ids: [...] }]. Takes explicit ids only — it does not import rows straight from a search. Adding contacts costs up to 2 credits each (contacts already saved with an email or phone are free), and the call fails if the balance is too low.
Inputs
listIdstringrequired- No description.
companiesarray- Add operations for companys
contactsarray- Add operations for contacts
fuseaimcp_add_list_to_campaign_toolAdds a further prospect list to an existing campaign (max 10).DestructiveAdd List To Campaign
Adds a further prospect list to an existing campaign (max 10). The campaign must already have at least one list from initialize_campaign_tool — this does not create a campaign without a list. Pass the campaign's id and the REAL list _id (24 hex). Works for system lists like 'All Website Visitors' (use its real _id from list_prospect_lists_tool with filters.includeSystemLists=true, never the slug). On an active campaign the added list is enrolled in the background and the campaign's sequence starts sending to its contacts (on a dynamic campaign, also to people added to the list later); AI writing in those messages is charged in credits as each message is written.
Inputs
campaignIdstringrequired- No description.
listIdstringrequired- No description.
fuseaimcp_approve_or_decline_auto_repliesApprove (queue to send) or decline (stop) pending auto-replies — works for both email and LinkedIn.DestructiveApprove or Decline Auto-Replies
Approve (queue to send) or decline (stop) pending auto-replies — works for both email and LinkedIn. Approving queues each reply to go out to its contact: in a chat with the user, approve only replies whose text they have seen and agreed to send. An approved reply can still be declined until it has gone out. Declining stops the contact's sequence; confirm before declining.
Inputs
isApprovedbooleanrequired- true approves (sends) the replies; false declines and STOPS each contact's sequence.
scheduledEmailIdsarrayrequired- The pending auto-replies' scheduledEmailId values (batch-capable).
fuseaimcp_build_list_from_search_or_listBuilds or appends to a list from a people search (searchFilters, run as a new search that saves up to limit people) or from another list's rows (targetListId).DestructiveBuild List from Search or List
Builds or appends to a list from a people search (searchFilters, run as a new search that saves up to limit people) or from another list's rows (targetListId). Returns the destination list id(s) right away; the rows are added in the background over the next few minutes. Saving from a search costs 1 credit per person saved, and the balance must cover limit credits before it starts; copying another list is free. Saving a person FuseAI already holds, the team's own custom contact for them included, updates that contact's profile fields from the search result. Copied companies that were never looked up before get a free background lookup by domain with an outside company data provider, which fills their empty details and replaces a name that is still just the domain; the tool itself looks up no emails or phone numbers. Each row newly added to the list starts the user's automations that run when a contact or company is added to a list, and people added to a list that an active dynamic campaign uses can be enrolled in it at its next refresh (every 30 minutes) and start receiving its outreach. Provide dataSource and one of existingListId (append to that list) or newListTitle (create a list); given both, it creates the new list and also adds the rows to the existing one. listType is contactList when omitted. It does not take raw contact details: its only data sources are people-search filters or an existing list. useRawFileData has no effect on this tool: its 'skip enrichment' wording applies to imports of raw rows, which this tool does not accept, so whether it is true, false or left out the rows are added the same way and no email or phone lookup runs. Save-by-search imports PEOPLE only; for a COMPANY filter-based save use bulk_save_companies_to_list.
Inputs
dataSourceobjectrequired- Exactly one source: { searchFilters, searchType, limit } for save-by-search, or { targetListId, targetListState, limit, offset } for save-by-list
existingListIdstring- Append to this existing list (provide exactly one of existingListId / newListTitle)
listTypestring- Type of the destination list: contactList or companyListone of
contactListcompanyList newListTitlestring- Create a new list with this title (provide exactly one of existingListId / newListTitle)
useRawFileDataboolean- Skip enrichment and store rows as-is (default false)
fuseaimcp_bulk_disconnect_email_accountsDESTRUCTIVE: Disconnect up to 100 email sending accounts in one call.DestructiveBulk Disconnect Email Accounts
DESTRUCTIVE: Disconnect up to 100 email sending accounts in one call. Each one is removed like disconnect_email_account: an owned account is deleted with its synced emails and conversations in FuseAI, its warmup (removed at the external warmup provider) and its sharing with teammates, and its inbox sync ends; an account shared with the user only loses the user's access. For everyone who loses a mailbox this way (the user, and for an owned account every teammate it was shared with), their campaigns stop sending from it (a running campaign left with no active sending account may stop); while they no longer have it, their campaign emails queued on it, including later steps for contacts already emailed from it, are not sent, and their scheduled one-off emails from it are cancelled when due. Selection: emailAddresses[] OR selectAll:true with filter ({emailAccountType?: "GMAIL"|"OUTLOOK"|"SMTP", onlyActive?}); excludedEmailAddresses[] only combines with selectAll, and a selection matching more than 100 accounts is refused (TOO_MANY_ACCOUNTS_SELECTED), so pass addresses in smaller batches instead. Only call after the user explicitly confirms the accounts.
Inputs
emailAddressesarray- No description.
excludedEmailAddressesarray- No description.
filterobject- No description.
selectAllboolean- No description.one of
true
fuseaimcp_bulk_enrichment_toolEnriches multiple contacts/prospects in bulk.DestructiveBulk Contact Enrichment
Enriches multiple contacts/prospects in bulk. Accepts a listName, a limit, an optional offset and an enrichType (email, phone, all for both, or demographics). Resolves the list by name (fuzzy match), then starts an ASYNC bulk enrichment of `limit` contacts of the list from `offset`, in place. The result says how many contacts the run takes (`totalContacts`, fewer than `limit` at the end of the list), the estimated minutes to completion, and `nextOffset`: pass it as `offset` to enrich the following contacts, since calling again without it takes the same contacts. Works on whole lists, not on a single person. Email, phone and all runs also refresh each searched contact's stored profile from the person's current profile, replacing details such as title, company, location and work history where they differ. On a list that an active dynamic campaign uses, contacts that gain an email or LinkedIn URL from a run can be enrolled and contacted on the campaign's next refresh. Demographics matches each uploaded or manually added contact of the list to a known profile and writes that profile's name, title, company and location over the contact's stored ones; when another contact already holds the person, the uploaded contact is merged into that contact, which takes its place on every list it was on (this can start the user's "Added to list" automations), and the uploaded contact is archived. It finds no emails or phone numbers, and contacts already matched are skipped. Email, phone and all charge only for what they find, per contact: a work email 20 credits when Fuse's own data finds it or 50 when a partner provider does (a personal email 100 when the account puts personal emails first), a phone number 100, or 5 instead when a provider key saved on the account found it; nothing when nothing is found. Demographics charges 1 credit per contact it matches, and nothing when there is no match. The most a run can cost is every contact it takes at the highest of those prices for its enrichType: 50 each for email (100 with personal emails first), 100 for phone, the two added up for all, 1 for demographics. It spends the user's credits: in a conversation, tell the user these prices and that most, and start it only once they have agreed to them (agreeing to another price, such as a search's or an import's, is not agreeing to this one). To run demographics and then email or phone on the same list, start demographics, and start the email or phone run only once the list has finished (its completionStatus is complete again): an email or phone run started before then does not wait for it and reads the contacts before they are matched. A run takes minutes, so an empty result soon after a start means it is still running, never a reason to start it again.
Inputs
enrichTypestringrequired- The type of enrichment to perform. Default to "email": only pass "phone" or "all" when the user explicitly asked for phone numbers. Phone enrichment costs several times more credits per contact. An email, phone or all run starts only when the balance covers every contact taken at the highest price (INSUFFICIENT_CREDITS otherwise), so an unrequested "all" can be refused, or spend the balance on phones nobody asked for. Such a run whose credits are spent elsewhere meanwhile pauses until the user adds credits and resumes. Only pass "demographics" when the user asked for demographics, or to match the list's contacts to known profiles: it finds no emails or phone numbers. It starts when the balance covers one match, and stops early when the balance runs out; running it again later skips the contacts it already matched.one of
emailphonealldemographics listNamestringrequired- The name of the list to enrich.
limitinteger- How many contacts to take, starting at `offset`, in the list's default order (oldest first). Defaults to the number of contacts in the list, capped at 100. Contacts that already hold the requested data, or were searched for it before, are skipped and not charged but still count toward the limit, so calling again with the same offset and limit takes the same contacts again: to reach contacts further down the list, pass an offset (offset 100 with limit 100 takes contacts 101-200).
offsetinteger- How many contacts to skip from the start of the list before taking `limit` of them. Defaults to 0.
fuseaimcp_bulk_save_companies_to_listStart a background job that saves up to `limit` companies matching a filter set, found in an external company database, into one or more lists and the user's saved companies, at 1 credit per company added.DestructiveBulk Save Companies To List
Start a background job that saves up to `limit` companies matching a filter set, found in an external company database, into one or more lists and the user's saved companies, at 1 credit per company added. It answers "Bulk save started" at once, before any company is saved, and the lists show as processing until the job ends. Filter-based (no need to pass company objects) — the preferred way to save company-search results to a list. When FuseAI already holds a company with the same domain, the values the search found replace its stored name, website, industry, size, LinkedIn page, founding year and headquarters location on that shared record, for every user who has saved it. Each company newly added to a list can start the user's active automations that use the "Company added to list" trigger. A company every one of those lists already holds is left out and costs nothing, so `limit` is how many new companies it adds and the same filters sent again add the next ones; a save stops after passing over about 1,000 companies the lists already hold, so a save into a long list can add fewer than `limit`. Send at least one filter (exclusions alone are fine): an empty filter set is refused with 422 SEARCH_FILTERS_INVALID. On the Free plan a save adds at most 100 companies: a larger `limit` is refused with 403 FREE_PLAN_COMPANY_SAVE_LIMIT_EXCEEDED.
Inputs
filtersobjectrequired- No description.
limitintegerrequired- No description.
listIdsarrayrequired- No description.
fuseaimcp_bulk_update_action_feed_statusBulk-apply a status change to the feed items matching a filter, in one call: mark_complete, mark_active (reopens completed items only) or delete (sets their status to deleted; the records are kept).DestructiveBulk Update Smart Actions Status
Bulk-apply a status change to the feed items matching a filter, in one call: mark_complete, mark_active (reopens completed items only) or delete (sets their status to deleted; the records are kept). Each item's linked task is closed, reopened or cancelled to match. Does NOT send messages. jobType must be mark_complete, mark_active or delete; any other value is refused and changes nothing. An empty filter matches EVERYTHING visible to the user's team that is not already deleted; if more than 1000 items match, nothing is changed and BULK_JOB_LIMIT_EXCEEDED is returned. High-impact — confirm before calling unless already approved.
Inputs
jobTypestringrequired- No description.one of
mark_completemark_activedelete filterobject- No description.
fuseaimcp_bulk_update_email_accountsBulk-update profile fields (profileUpdate: {name?, title?, capacity?, signature?} — at least one) across several email accounts the user owns or that teammates shared with them, replacing the current values for everyone who uses those mailboxes.DestructiveBulk Update Email Accounts
Bulk-update profile fields (profileUpdate: {name?, title?, capacity?, signature?} — at least one) across several email accounts the user owns or that teammates shared with them, replacing the current values for everyone who uses those mailboxes. name, title and signature are what recipients see on campaign emails sent from these mailboxes: name is the From name (on emails already queued too; an Outlook mailbox's in-thread follow-ups carry the name set in Microsoft) and the three fill the {Sender …} placeholders. capacity is each mailbox's daily campaign send limit (0 stops its campaign sends). An empty title or signature clears the stored value. Selection: pass emailAddresses[] (specific accounts) OR selectAll:true with filter ({emailAccountType?: "GMAIL"|"OUTLOOK"|"SMTP", onlyActive?}); excludedEmailAddresses[] only combines with selectAll.
Inputs
profileUpdateobjectrequired- No description.
emailAddressesarray- No description.
excludedEmailAddressesarray- No description.
filterobject- No description.
selectAllboolean- No description.
fuseaimcp_bulk_update_linkedin_accountsBulk-update FuseAI sending settings (profileUpdate: {title?, dailyUsageLimits?, weeklyUsageLimits?, inmailMonthly?: {budget}, signature?} — at least one) across several LinkedIn accounts the user owns or that teammates shared with them, replacing the current values for everyone who uses those accounts; an empty title or signature clears it.DestructiveBulk Update LinkedIn Accounts
Bulk-update FuseAI sending settings (profileUpdate: {title?, dailyUsageLimits?, weeklyUsageLimits?, inmailMonthly?: {budget}, signature?} — at least one) across several LinkedIn accounts the user owns or that teammates shared with them, replacing the current values for everyone who uses those accounts; an empty title or signature clears it. Nothing changes on LinkedIn itself, but campaigns sending from these accounts use the new values: title and signature fill {Sender Job Title} and {Sender Signature} in LinkedIn messages written afterwards, and the limits set how many invitations, messages, InMails, profile visits and reactions they send (0 stops that action; inmailMonthly budget null removes the monthly InMail cap). Limits above an account's free or premium maximum are skipped for that account. Selection: pass providerIds[] (specific accounts) OR selectAll:true with filter ({onlyActive?}); excludedProviderIds[] only combines with selectAll.
Inputs
profileUpdateobjectrequired- No description.
excludedProviderIdsarray- No description.
filterobject- No description.
providerIdsarray- No description.
selectAllboolean- No description.
fuseaimcp_cancel_automation_executionCancel one execution of an automation: a run queued, running or waiting (parked on a wait step) ends as cancelled and nothing after its current step runs.DestructiveCancel Automation Execution
Cancel one execution of an automation: a run queued, running or waiting (parked on a wait step) ends as cancelled and nothing after its current step runs. A run that already completed, failed or was cancelled answers EXECUTION_ALREADY_FINISHED.
Inputs
automationIdstringrequired- No description.
executionIdstringrequired- No description.
fuseaimcp_change_campaign_lead_status_toolResume (activate=true) or stop (activate=false) one lead within its campaign, by the lead's targetContactId.DestructiveChange Campaign Lead Status
Resume (activate=true) or stop (activate=false) one lead within its campaign, by the lead's targetContactId. Resuming sets a stopped lead active and immediately schedules its next step, so outreach to that person restarts from the user's connected mailbox or LinkedIn account; the lead's company website may be read first to describe the company, an AI-written or AI-personalized step costs 10 credits, and if credits or a usable sender account are missing the whole campaign is paused instead, with an in-app notice (and a Slack message if Slack is connected). Resuming works whatever stopped the lead, an unsubscribe included, and does not remove them from Unsubscribed Contacts; use resubscribe_contact_tool to reverse an unsubscribe. Stopping the lead is not a pause: it cancels every step still pending for the lead, including scheduled sends, AI replies awaiting approval and call tasks, and resuming later writes those steps afresh rather than restoring them. Use to stop or restart outreach to a single person.
Inputs
activatebooleanrequired- true to activate/resume the lead in its campaign, false to pause/deactivate it.
targetContactIdstringrequired- The lead's targetContactId (a 24-hex MongoDB id).
fuseaimcp_complete_taskMark an open task done on behalf of the user, with an optional note and an optional sales signal (interested or not_interested), which is only recorded and does not stop a campaign.DestructiveComplete Task
Mark an open task done on behalf of the user, with an optional note and an optional sales signal (interested or not_interested), which is only recorded and does not stop a campaign. Completing never sends the task's own message, whether or not completedOutsideFuse is set. On a campaign step task it closes that step and moves the contact on to the next campaign step, which Fuse then writes (AI-written steps cost credits) and, if that step is automatic and the campaign is active, sends it to the contact as an email or LinkedIn action from the campaign's connected mailbox or LinkedIn account. Set completedOutsideFuse true when the user already did that campaign step outside Fuse: it records the step as done outside Fuse and keeps an email or LinkedIn step out of send analytics (other tasks ignore it), but does not stop the next step from being written or sent. Group rows such as a sequence step batch, call batches and AI-reply reviews cannot be completed here.
Inputs
taskIdstringrequired- The task id (24-character Mongo id) from list_tasks.
completedOutsideFuseboolean- true when a send step was already done outside Fuse: closes it without sending anything and keeps it out of analytics.default
false notestring- Optional completion note.
outcomestring- Optional sales signal recorded with the completion.one of
interestednot_interested
fuseaimcp_connect_purchased_mailboxConnect a mailbox the team already purchased through FuseAI as a sending account.DestructiveConnect Purchased Mailbox
Connect a mailbox the team already purchased through FuseAI as a sending account. mailboxId is the RESELLER id the user can copy from the app's Domain Management mailboxes table (NOT a Mongo _id). Credentials are resolved server-side, nothing to supply and nothing is bought. The mailbox is added to the account of whoever on the team bought it, takes a seat in the plan's mailbox limit (refused with EMAIL_ACCOUNT_LIMIT_REACHED when full) and is switched on for campaigns at once (15 emails a day), so running campaigns whose steps do not pin specific mailboxes can start sending from it; its inbox may also start syncing into FuseAI. If the address is already connected, its stored password and connection settings are replaced, and a mailbox the system had switched off (after rejected sends or a provider rate limit) is switched back on, resuming its queued campaign emails.
Inputs
mailboxIdstringrequired- No description.
fuseaimcp_connect_purchased_mailboxesConnect several mailboxes the team already purchased through FuseAI as sending accounts in one call.DestructiveConnect Purchased Mailboxes
Connect several mailboxes the team already purchased through FuseAI as sending accounts in one call. mailboxIds are the RESELLER ids from the app's Domain Management mailboxes table (NOT Mongo _ids), max 25. Credentials are resolved server-side, nothing to supply and nothing is bought. Each mailbox is connected as connect_purchased_mailbox does: added to the account of whoever on the team bought it, taking a seat in the plan's mailbox limit (refused for that mailbox with EMAIL_ACCOUNT_LIMIT_REACHED when full) and switched on for campaigns at once, so running campaigns whose steps do not pin specific mailboxes can start sending from it; its inbox may also start syncing into FuseAI. An address that is already connected has its stored password and connection settings replaced and, if the system had switched it off, is switched back on, resuming its queued campaign emails. Returns a per-mailbox breakdown — a partial success is normal when some mailboxes are still provisioning.
Inputs
mailboxIdsarrayrequired- No description.
fuseaimcp_copy_split_arm_toolDESTRUCTIVE for the target arm: replaces every step of one arm of an A/B split (sequenceStepId is the split's id) with copies of the other arm's steps, nested conditions and their lanes included (voice-note scripts are not copied), so the two arms can then differ in one step.DestructiveCopy Split Arm
DESTRUCTIVE for the target arm: replaces every step of one arm of an A/B split (sequenceStepId is the split's id) with copies of the other arm's steps, nested conditions and their lanes included (voice-note scripts are not copied), so the two arms can then differ in one step. fromBranchKey and toBranchKey are 'a' and 'b'. Only before the campaign starts. Only call after the user has confirmed replacing that arm's steps.
Inputs
campaignIdstringrequired- No description.
fromBranchKeystringrequired- The arm to copy from: 'a' or 'b'.one of
ab sequenceStepIdstringrequired- No description.
toBranchKeystringrequired- The arm whose steps are replaced by the copies: the other arm.one of
ab
fuseaimcp_create_automationCreate an automation as a graph.DestructiveCreate Automation
Create an automation as a graph. An active rule runs on its own, as the user, whenever a trigger fires: each run costs 2 credits plus what its steps use, and its steps may add contacts to campaigns (which then email and message them from the user's connected accounts), post to Slack, enrich or validate contacts and find contacts through outside data providers, scrape web pages, run smart columns, create tasks and add records to lists. Steps and triggers that choose their operation only when the rule runs (mcp_call, composio_call, agent_step and run_code steps, and composio_event triggers) are refused here and nothing is saved; rules that need them are built in the FuseAI app. Body: name, summary (one plain sentence), triggers [{nodeKey, label (2-5 words), description (one short sentence on what starts this rule), attributes}] (min 1, any one fires the rule), steps [{stepId, nodeKey, label (2-5 words), description, dependsOn [stepIds], match ('is'|'is_not', gates only), forEach ('{{steps.<id>.output.<array>}}' to run per item), attributes}] (min 1), status (draft|active|paused; omitted = draft — pass 'draft' for a rule the user has not yet activated, 'active' only on their explicit confirmation), optional expiresAt, sourcePrompt (the user's request) and compilerNotes. Attribute values may be literals or bindings: {{trigger.<field>}} (ONLY fields in the trigger node's outputSchema: contactId or companyId, occurredAt, attributes.<key>, meta.<key>; a name, email or LinkedIn URL is not on the event), {{steps.<stepId>.output.<field>}}, {{item}} / {{item.<field>}} inside a forEach. A binding to a field the trigger event does not carry is rejected at save with the fields it does carry. A step below a forEach step already runs once per item: never repeat that forEach on it (FOR_EACH_REPEATED at save), and a step that must run ONCE after a fan-out (a digest) depends on the array's producer, never on the per-item step (FAN_OUT_DEPENDENT_WITHOUT_ITEM_DATA when it uses no per-item data). A gate's path need not be in outputSchema.required; a missing value reads as false. Contact- and company-scoped condition/action nodes take the record from the trigger event when the trigger is about that entity; in any other rule (schedule, scrape, the other entity) bind their contactId / companyId attribute from an earlier step's output, or the save is rejected. nodeKeys and per-node attributes MUST come from list_automation_nodes; id-valued attributes (campaign, list, ICP, signal, email account) must be resolved with the corresponding list tools first.
Inputs
namestringrequired- No description.
stepsarrayrequired- No description.
triggersarrayrequired- No description.
compilerNotesarray- No description.
expiresAtstring- No description.
sourcePromptstring- No description.
statusstring- No description.one of
activepauseddraft summarystring- No description.
fuseaimcp_create_companies_watcher_monitorCreate a Companies Watcher signal monitor: watch the companies in one or more SOURCE lists for job postings, LinkedIn posts, press mentions, funding milestones, new hires with given titles (delivers the hired PEOPLE into a new contact list), status changes (acquired / IPO / closed / subsidiary / operating), newly adopted technologies, acquisitions they make, or layoffs.DestructiveCreate companies-watcher monitor
Create a Companies Watcher signal monitor: watch the companies in one or more SOURCE lists for job postings, LinkedIn posts, press mentions, funding milestones, new hires with given titles (delivers the hired PEOPLE into a new contact list), status changes (acquired / IPO / closed / subsidiary / operating), newly adopted technologies, acquisitions they make, or layoffs. listId/listIds identify the existing SOURCE list(s) to watch; also needs an eventType and a listName that names BOTH the agent and a NEW dedicated results list it creates — listName must be unique and must NOT reuse a source list's name (name it after the agent the user is creating). A job-posting watcher needs at least one keyword, matched against job titles: use keywords (jobPostingTitle and jobPostingDescription are deprecated aliases of keywords and are also matched against job titles only), optionally narrow by jobPostingLocationTypes, and leave out jobPostingRegion, which is retired: any value is refused. A company is watched by its website, or, for a company the user imported with its LinkedIn page and no website, by that page. Unless initializeOnly is true the watcher starts at once: the companies are registered with Fiber, an external signals provider, and every check (every frequency days, daily by default) costs 50 credits per watched company, plus email or phone enrichment credits for each new hire enriched. With enrichEmail or enrichPhone on (new-hires watchers only), each hire newly added to the results list is looked up through FuseAI's contact enrichment, which checks FuseAI's own data and a waterfall of outside email and phone data providers, and each email or phone number found is saved on the contact. Each signal adds the company, or the hired people, to the results list and to the user's saved companies or contacts: a company the user had removed is saved again, a background Fiber lookup can fill a company's empty fields on its shared record and replace a name that is still just the domain, and a saved contact matched to a hire has its profile fields replaced with the delivered profile. Each signal also fires the Company signal match trigger (Signal match for each hire), so the user's automations on that trigger run, which may add people to campaigns or act in connected apps. It pauses once maxContacts results have been added and stops after one month unless expirationDate is set (null for no end date).
Inputs
eventTypestringrequired- No description.one of
company-watch-linkedin-job-postingscompany-watch-linkedin-postscompany-watch-press-mentionscompany-watch-funding-milestonescompany-watch-new-hirescompany-watch-status-changescompany-watch-new-technologiescompany-watch-acquisitionscompany-watch-layoffs listNamestringrequired- Name for BOTH the agent and the NEW dedicated results list this watcher creates. Must be unique — never reuse a source list's name; name it after the agent the user is creating.
maxCreditSpendintegerrequired- Required. The most credits one check may cost (1000 to 200000). Every check costs 50 credits per watched contact or company, so the watcher watches at most maxCreditSpend / 50 of them (rounded down) from its source lists; the rest are not watched or billed.
enrichEmailboolean- company-watch-new-hires only (ignored for any other eventType): whether to enrich the hires' emails, which campaigns need.
enrichPhoneboolean- company-watch-new-hires only (ignored for any other eventType): whether to enrich the hires' phone numbers.
expirationDatestring- No description.
frequencyinteger- No description.default
1 initializeOnlyboolean- Omit or false to create and START the monitor immediately (sets up the live subscription and spends credits). Set true to save it as a DRAFT for the user to review and activate later.default
false jobPostingDescriptionarray- Keywords matched against JOB TITLES upstream. At least one is required for job-posting watchers.
jobPostingLocationTypesarray- Work arrangement filter — any of: remote, on_site, hybrid. Omit to match any arrangement.
jobPostingRegionstring- No description.
jobPostingTitlearray- Keywords matched against JOB TITLES upstream. At least one is required for job-posting watchers.
keywordsarray- Keywords matched against JOB TITLES upstream. At least one is required for job-posting watchers.
listIdstring- The existing SOURCE list(s) to watch, by id — the people/companies whose changes are tracked.
listIdsarray- The existing SOURCE list(s) to watch, by id — the people/companies whose changes are tracked.
maxContactsinteger- Max Results: the watcher pauses once this many results have been added to its results list. It does not limit how many contacts or companies are watched or what a check costs; maxCreditSpend does.
signalFiltersobject- Filters on the watched signals. REQUIRED for company-watch-new-hires, which needs titleKeywords; optional for the other eventTypes. Each key names the eventType that takes it and is refused with any other.
fuseaimcp_create_company_department_headcount_monitorCreate a Company Department Headcount signal monitor: track companies whose named department is within a headcount range.DestructiveCreate company-department-headcount monitor
Create a Company Department Headcount signal monitor: track companies whose named department is within a headcount range. Unless initializeOnly is true it starts at once: CrustData, an external company data provider, checks daily, and each matching company it delivers is added to a new company list named listName (the name must not already exist) and to the user's saved companies, at 100 credits per company newly added to the list. A company the user had removed is saved again, and a background lookup with Fiber, an external company data provider, then fills empty fields on those shared company records and replaces a name that is still just the domain. Each match can start the user's active automations that use the Company signal match trigger. It pauses after maxContacts companies (default 200) and stops after one month unless expirationDate is set (null for no end date). notificationCount, if set, caps how many companies one daily run delivers; despite its name, it sends no alerts to the user or anyone else.
Inputs
departmentHeadcountobjectrequired- No description.
industriesarrayrequired- No description.
listNamestringrequired- No description.
regionsarrayrequired- No description.
annualRevenueobject- No description.
companyHeadcountarray- No description.
expirationDatestring- No description.
initializeOnlyboolean- Omit or false to create and START the monitor immediately (sets up the live subscription and spends credits). Set true to save it as a DRAFT for the user to review and activate later.default
false maxContactsinteger- No description.
maxCreditSpendinteger- Maximum credits this agent may spend in a single run or refresh (1000 to 200000). CrustData monitors run once a day, so for them it is a daily limit; job-change watchers watch only as many contacts as one check within it pays for; web monitors stop once their total delivered results reach it. Omit for no limit.
notificationCountinteger- No description.
fuseaimcp_create_company_employee_job_location_in_two_countries_monitorCreate an Employee-Job-Location-in-Two-Countries signal monitor: track companies whose employees work across two countries, scoped to one industry/region/headcount.DestructiveCreate employee-job-location monitor
Create an Employee-Job-Location-in-Two-Countries signal monitor: track companies whose employees work across two countries, scoped to one industry/region/headcount. No input chooses the two countries: CrustData reports a company once its employees work in two countries, and regions filters on the company's headquarters location. Unless initializeOnly is true it starts at once: CrustData, an external company data provider, checks daily, and each matching company it delivers is added to a new company list named listName (the name must not already exist) and to the user's saved companies, at 100 credits per company newly added to the list. A company the user had removed is saved again, and a background lookup with Fiber, an external company data provider, then fills empty fields on those shared company records and replaces a name that is still just the domain. Each match can start the user's active automations that use the Company signal match trigger. It pauses after maxContacts companies (default 200) and stops after one month unless expirationDate is set (null for no end date). notificationCount, if set, caps how many companies one daily run delivers; despite its name, it sends no alerts to the user or anyone else. companyHeadcountGrowth is accepted but not used as a filter.
Inputs
companyHeadcountarrayrequired- Company-size bucket — use one of exactly: 1-10, 11-50, 51-200, 201-500, 501-1,000, 1,001-5,000, 5,001-10,000, 10,001+.
industriesarrayrequired- No description.
listNamestringrequired- No description.
regionsarrayrequired- No description.
annualRevenueobject- No description.
companyHeadcountGrowthobject- No description.
expirationDatestring- No description.
initializeOnlyboolean- Omit or false to create and START the monitor immediately (sets up the live subscription and spends credits). Set true to save it as a DRAFT for the user to review and activate later.default
false maxContactsinteger- No description.
maxCreditSpendinteger- Maximum credits this agent may spend in a single run or refresh (1000 to 200000). CrustData monitors run once a day, so for them it is a daily limit; job-change watchers watch only as many contacts as one check within it pays for; web monitors stop once their total delivered results reach it. Omit for no limit.
notificationCountinteger- No description.
fuseaimcp_create_company_from_domainSave a company for this user by website domain.DestructiveCreate Company From Domain
Save a company for this user by website domain. Pass `website` (a domain or URL); if no company record exists for the domain it is created, named by its domain, then the company is added to the user's saved companies (a company the user had removed is saved again). After the call returns, a background job sends each saved company that lacks basic firmographic data, and was not looked up before, to an external company data provider, then fills the company's empty fields and its placeholder name; this lookup charges no credits, and the returned record does not include that data yet. Optional `listNames` adds the company to those company lists, creating a list when its name does not exist yet (these additions do not start the user's "Company added to list" automations); optional `source` labels how it was acquired. Optional `watcherType` is written to the company record that every FuseAI account with this company shares, replacing any stored value; it only labels the record and starts no signal monitoring. Optional `websites` saves several companies the same way, without adding them to lists or returning them. Returns { data, outcome } where data is the company record (data._id is the company id) and outcome.isNewCompany is true when the company record was created by this call and false when it already existed; outcome.lists (when listNames given) is an array of { listName, listId, isNewList, isNewCompanyAddition }. Use it whenever the company must show up in the user's CRM.
Inputs
listNamesarray- No description.
sourcestring- No description.one of
prospect_searchcompany_searchwebsite_visitorssalesgptagentscsv_uploadmanualautomationsderived_from_contactsbusiness_apihubspotsalesforcezohoattiopipedrivegohighlevelnotion watcherTypearray- No description.
websitestring- No description.
websitesarray- No description.
fuseaimcp_create_company_headcount_growth_monitorCreate a Company Headcount Growth signal monitor: track companies growing headcount within a % range, scoped to one industry/region/headcount bucket.DestructiveCreate company-headcount-growth monitor
Create a Company Headcount Growth signal monitor: track companies growing headcount within a % range, scoped to one industry/region/headcount bucket. Unless initializeOnly is true, it starts at once: it creates a new company list named listName and a daily watch with CrustData, an external company data provider, that runs until expirationDate (one month out by default; null for no end). Each company found is added to the user's saved companies (a company the user had removed is saved again) and to the list, at 100 credits per company added. A background lookup with Fiber, an external company data provider, then fills empty fields on those shared company records and replaces a name that is still just the domain. Each match can start the user's active automations that use the Company signal match trigger.
Inputs
companyHeadcountarrayrequired- Company-size bucket — use one of exactly: 1-10, 11-50, 51-200, 201-500, 501-1,000, 1,001-5,000, 5,001-10,000, 10,001+.
companyHeadcountGrowthobjectrequired- No description.
industriesarrayrequired- No description.
listNamestringrequired- No description.
regionsarrayrequired- No description.
annualRevenueobject- No description.
companyCountriesarray- No description.
expirationDatestring- No description.
initializeOnlyboolean- Omit or false to create and START the monitor immediately (sets up the live subscription and spends credits). Set true to save it as a DRAFT for the user to review and activate later.default
false maxContactsinteger- No description.
maxCreditSpendinteger- Maximum credits this agent may spend in a single run or refresh (1000 to 200000). CrustData monitors run once a day, so for them it is a daily limit; job-change watchers watch only as many contacts as one check within it pays for; web monitors stop once their total delivered results reach it. Omit for no limit.
notificationCountinteger- No description.
fuseaimcp_create_company_headcount_growth_over_baseline_monitorCreate a Headcount-Growth-Over-Baseline signal monitor: track companies that grow past a baseline headcount by a given percentage (companyHeadcountGrowthFromBaseline) within a timeframe; regions takes one value formatted '<state>, <country>' or '<city>, <state>, <country>', and other formats are refused.DestructiveCreate headcount-growth-over-baseline monitor
Create a Headcount-Growth-Over-Baseline signal monitor: track companies that grow past a baseline headcount by a given percentage (companyHeadcountGrowthFromBaseline) within a timeframe; regions takes one value formatted '<state>, <country>' or '<city>, <state>, <country>', and other formats are refused. Unless initializeOnly is true it starts at once: CrustData, an external company data provider, checks daily, and each matching company it delivers is added to a new company list named listName (the name must not already exist) and to the user's saved companies, at 100 credits per company newly added to the list. A company the user had removed is saved again, and a background lookup with Fiber, an external company data provider, then fills empty fields on those shared company records and replaces a name that is still just the domain. Each match can start the user's active automations that use the Company signal match trigger. It pauses after maxContacts companies (default 200) and stops after one month unless expirationDate is set (null for no end date). notificationCount, if set, caps how many companies one daily run delivers; despite its name, it sends no alerts to the user or anyone else.
Inputs
baselineHeadcountarrayrequired- No description.
companyHeadcountGrowthFromBaselinearrayrequired- No description.
industriesarrayrequired- No description.
listNamestringrequired- No description.
regionsarrayrequired- No description.
timeframearrayrequired- Timeframe — use one of exactly: YoY, MoM, QoQ, 6M, 2Y.
annualRevenueobject- No description.
expirationDatestring- No description.
initializeOnlyboolean- Omit or false to create and START the monitor immediately (sets up the live subscription and spends credits). Set true to save it as a DRAFT for the user to review and activate later.default
false maxContactsinteger- No description.
maxCreditSpendinteger- Maximum credits this agent may spend in a single run or refresh (1000 to 200000). CrustData monitors run once a day, so for them it is a daily limit; job-change watchers watch only as many contacts as one check within it pays for; web monitors stop once their total delivered results reach it. Omit for no limit.
notificationCountinteger- No description.
fuseaimcp_create_contact_noteAdd a plain-text note to a CRM contact's activity timeline in Fuse on behalf of the user, e.g.DestructiveCreate Contact Note
Add a plain-text note to a CRM contact's activity timeline in Fuse on behalf of the user, e.g. a summary of an offline conversation. The note is not pushed to a connected CRM such as HubSpot or Salesforce, and the call itself sends nothing, calls no outside service and starts no automation or campaign. The whole team sees it, and it counts toward the contact's lead score at Fuse's next daily rescore, which can move the contact into the Hot Leads list; that move does not start the user's "Added to list" automations, but an active dynamic campaign that uses Hot Leads as a source list can then enroll the contact and start messaging them by email or LinkedIn. If the team's CRM push conditions use lead score, the score change can also get the contact created in the team's connected CRM at the next scheduled push. When Fuse Brain's 'Call transcripts & notes' source is on, Fuse Brain reads it and counts the note's author as a billed source in its monthly credit charge. Returns the created note.
Inputs
bodystringrequired- Plain-text note body, up to 10000 characters.
contactIdstringrequired- CRM contact id the note is about (the id list and contact tools return).
fuseaimcp_create_first_person_hired_in_department_monitorCreate a First-Person-Hired-in-Department signal monitor: detect the first hire into a given department at companies matching the filters.DestructiveCreate first-person-hired-in-department monitor
Create a First-Person-Hired-in-Department signal monitor: detect the first hire into a given department at companies matching the filters. Unless initializeOnly is true, it starts at once: it creates a new company list named listName and a daily watch with CrustData, an external company data provider, that runs until expirationDate (one month out by default; null for no end). Each company found is added to the user's saved companies (a company the user had removed is saved again) and to the list, at 100 credits per company added. A background lookup with Fiber, an external company data provider, then fills empty fields on those shared company records and replaces a name that is still just the domain. Each match can start the user's active automations that use the Company signal match trigger.
Inputs
companyDepartmentarrayrequired- Department — use one of exactly: Accounting, Administrative, Arts and Design, Business Development, Community and Social Services, Consulting, Education, Engineering, Entrepreneurship, Finance, Healthcare Services, Human Resources, Information Technology, Legal, Marketing, Media and Communication, Military and Protective Services, Operations, Product Management, Program and Project Management, Purchasing, Quality Assurance, Real Estate, Research, Sales, Customer Success and Support.
companyHeadcountarrayrequired- Company-size bucket — use one of exactly: 1-10, 11-50, 51-200, 201-500, 501-1,000, 1,001-5,000, 5,001-10,000, 10,001+.
industriesarrayrequired- No description.
listNamestringrequired- No description.
regionsarrayrequired- No description.
annualRevenueobject- No description.
expirationDatestring- No description.
initializeOnlyboolean- Omit or false to create and START the monitor immediately (sets up the live subscription and spends credits). Set true to save it as a DRAFT for the user to review and activate later.default
false maxContactsinteger- No description.
maxCreditSpendinteger- Maximum credits this agent may spend in a single run or refresh (1000 to 200000). CrustData monitors run once a day, so for them it is a daily limit; job-change watchers watch only as many contacts as one check within it pays for; web monitors stop once their total delivered results reach it. Omit for no limit.
notificationCountinteger- No description.
numOfFollowersarray- No description.
fuseaimcp_create_first_person_hired_internationally_monitorCreate a First-Person-Hired-Internationally signal monitor: detect a company's first international hire, scoped to one industry/region/headcount.DestructiveCreate first-person-hired-internationally monitor
Create a First-Person-Hired-Internationally signal monitor: detect a company's first international hire, scoped to one industry/region/headcount. Unless initializeOnly is true, it starts at once: it creates a new company list named listName and a daily watch with CrustData, an external company data provider, that runs until expirationDate (one month out by default; null for no end). Each company found is added to the user's saved companies (a company the user had removed is saved again) and to the list, at 100 credits per company added. A background lookup with Fiber, an external company data provider, then fills empty fields on those shared company records and replaces a name that is still just the domain. Each match can start the user's active automations that use the Company signal match trigger.
Inputs
companyHeadcountarrayrequired- Company-size bucket — use one of exactly: 1-10, 11-50, 51-200, 201-500, 501-1,000, 1,001-5,000, 5,001-10,000, 10,001+.
industriesarrayrequired- No description.
listNamestringrequired- No description.
regionsarrayrequired- No description.
annualRevenueobject- No description.
companyHeadcountGrowthobject- No description.
expirationDatestring- No description.
initializeOnlyboolean- Omit or false to create and START the monitor immediately (sets up the live subscription and spends credits). Set true to save it as a DRAFT for the user to review and activate later.default
false maxContactsinteger- No description.
maxCreditSpendinteger- Maximum credits this agent may spend in a single run or refresh (1000 to 200000). CrustData monitors run once a day, so for them it is a daily limit; job-change watchers watch only as many contacts as one check within it pays for; web monitors stop once their total delivered results reach it. Omit for no limit.
notificationCountinteger- No description.
fuseaimcp_create_funding_announcement_monitorCreate a Funding Announcement signal monitor: track companies that announce funding rounds.DestructiveCreate funding-announcement monitor
Create a Funding Announcement signal monitor: track companies that announce funding rounds. Pass fundingRoundTypes ([] for all rounds) and at least one account filter (companyHeadcount, industries or companyCountries); a call with none is refused. companyHeadcount accepts every value in its enum, not only the eight bands its field description lists: also the same bands written without commas (501-1000, 1001-5000, 5001-10000, 10001+) and the two parts of 1-10 (myself only, 2-10). Unless initializeOnly is true, it starts at once: it creates a new company list named listName (the name must not already exist) and a daily watch with CrustData, an external company data provider, that runs until expirationDate (one month out by default; null for no end). Each company found is added to the user's saved companies (a company the user had removed is saved again) and to the list, at 100 credits per company newly added to the list, or 300 per company with realtime (a live search for same-day results, off by default; the 300x in the realtime field is the data provider's cost, not what the user is charged). A background lookup with Fiber, an external company data provider, then fills empty fields on those shared company records and replaces a name that is still just the domain. Each match can start the user's active automations that use the Company signal match trigger. maxCreditSpend caps the credits one run spends; notificationCount caps the companies one run delivers and, despite its name, sends no alerts (a realtime agent delivers 5 a run when it is omitted); maxContacts pauses the monitor after that many companies in total (no limit when omitted). With none of these three set, a run that is not realtime is limited only by the provider's per-run maximum. The monitor also pauses when the balance cannot cover a delivery.
Inputs
fundingRoundTypesarrayrequired- Funding round types — use values from: pre_seed, seed, angel, series_a, series_b, series_c, series_d, series_e, series_f, series_g, series_h, series_i, series_j, series_unknown, private_equity, debt_financing, convertible_note, grant, corporate_round, equity_crowdfunding, product_crowdfunding, secondary_market, post_ipo_equity, post_ipo_debt, post_ipo_secondary, non_equity_assistance, initial_coin_offering, undisclosed. Pass [] to match all rounds.
listNamestringrequired- No description.
companyCountriesarray- No description.
companyHeadcountarray- Company-size bucket — use one of exactly: 1-10, 11-50, 51-200, 201-500, 501-1,000, 1,001-5,000, 5,001-10,000, 10,001+.
expirationDatestring- No description.
industriesarray- No description.
initializeOnlyboolean- Omit or false to create and START the monitor immediately (sets up the live subscription and spends credits). Set true to save it as a DRAFT for the user to review and activate later.default
false maxContactsinteger- No description.
maxCreditSpendinteger- Maximum credits this agent may spend in a single run or refresh (1000 to 200000). CrustData monitors run once a day, so for them it is a daily limit; job-change watchers watch only as many contacts as one check within it pays for; web monitors stop once their total delivered results reach it. Omit for no limit.
notificationCountinteger- No description.
realtimeboolean- Run a realtime agent: a live search for same-day results at about 300x the provider cost of the default. Off by default; only set it when the user asks for realtime. A realtime agent delivers 5 results a run unless notificationCount is set.
fuseaimcp_create_g2_review_scrapeFind the people who reviewed a software product on G2 and save them to a new contact list named listName.DestructiveCreate G2 review scrape
Find the people who reviewed a software product on G2 and save them to a new contact list named listName. Unless initializeOnly is true, it starts a one-time background run at once: an outside scraping service reads the product's public G2 reviews. Requires productUrl (a https://www.g2.com/products/<product> URL) and a listName that no existing list or scrape uses. Reads the newest maxReviews reviews (optionally only those rated at or below maxRating stars, or published in the last lookbackDays days). G2 shows only a reviewer's first name, last initial, job title and company size, so each named reviewer is searched on Fiber, an external people-data provider, and only those matched to exactly one LinkedIn profile are added; anonymous reviewers, reviewers with no job title and ambiguous matches are skipped. Each contact added costs 50 credits, and 2 more when the user has no email or phone saved for them, plus enrichment credits when enrichEmail or enrichPhone is on; the credit balance and maxCreditSpend can lower how many are added. A reviewer who is already a contact has their saved profile fields (such as name, title and company) overwritten with the matched LinkedIn profile, a reviewer the user had deleted from their contacts is saved again, and each added contact can start the user's automations, which can add them to campaigns or act in connected apps.
Inputs
listNamestringrequired- No description.
productUrlstringrequired- The G2 product whose reviewers to find, e.g. https://www.g2.com/products/outreach/reviews
enrichEmailboolean- No description.default
false enrichPhoneboolean- No description.default
false initializeOnlyboolean- Omit or false to create and START the scrape immediately (spends credits). Set true to save it as a DRAFT for the user to review and activate later.default
false lookbackDaysinteger- Read only reviews published in the last this many days
maxCreditSpendinteger- Maximum credits this agent may spend in a single run or refresh (1000 to 200000). CrustData monitors run once a day, so for them it is a daily limit; job-change watchers watch only as many contacts as one check within it pays for; web monitors stop once their total delivered results reach it. Omit for no limit.
maxRatinginteger- Read only reviews rated at or below this many stars, e.g. 3 for unhappy users
maxReviewsinteger- How many of the newest reviews to readdefault
100
fuseaimcp_create_google_maps_scrapeFind local businesses of one type on Google Maps and save them to a NEW company list named listName (shown as a 'Google Maps Search' agent).DestructiveCreate Google Maps search
Find local businesses of one type on Google Maps and save them to a NEW company list named listName (shown as a 'Google Maps Search' agent). Give businessType (e.g. 'dentist', never a location) plus locations: fully qualified city names as city, state or region, country (e.g. 'Austin, Texas, United States'); or give mapsLink, a Google Maps search, place, map-view or maps.app.goo.gl link, which supplies the area (and, for a search link, the term). maxResults caps how many businesses are collected. Unless mode is 'draft', the search starts at once and fills the list in the background, at 50 credits per business saved. A business whose website matches a company already in FuseAI is linked to that company, and every saved business is added to the user's saved companies (one the user had removed is saved again). A background lookup with Fiber, an external company data provider, then fills empty fields on the saved companies' shared records and replaces a name that is still just the domain. Each business added can start the user's active automations that use the Company signal match or Company added to list trigger.
Inputs
listNamestringrequired- No description.
businessTypestring- What kind of business to find, e.g. 'dentist' or 'HVAC contractor'. Never a location: cities go in `locations`. Optional with a search link, which carries its own term.
locationsarray- Up to 25 fully qualified city names as city, state or region, country (e.g. Austin, Texas, United States). Each city's full footprint is searched; a name that does not resolve to a city is skipped and reported, and the request is refused only when none resolve.
mapsLinkstring- A Google Maps link to search from instead of, or as well as, `locations`: a search link (its term and map area are used), a place link (its city footprint), a map-view link (a circle around its centre), or a maps.app.goo.gl share link. Directions, Street View and place-id links are refused.
maxCreditSpendinteger- Maximum credits this agent may spend in a single run or refresh (1000 to 200000). CrustData monitors run once a day, so for them it is a daily limit; job-change watchers watch only as many contacts as one check within it pays for; web monitors stop once their total delivered results reach it. Omit for no limit.
maxResultsinteger- How many businesses to collect across all locations (1 to 250).default
100 modestring- "draft" saves the search as a DRAFT for the user to review and activate later. Omit it, or send "start", to create and START the search immediately (spends credits).one of
draftstart radiusMilesnumber- Radius in miles around a link's map position (1 to 250); defaults to one derived from the link's zoom. Only with mapsLink.
fuseaimcp_create_icpCreate an Ideal Customer Profile (ICP) in the user's FuseAI account.DestructiveCreate ICP
Create an Ideal Customer Profile (ICP) in the user's FuseAI account. It never changes the user's other ICPs, and a name one of them already has (ignoring case) is refused. The new ICP counts at once wherever any of the user's ICPs count: automation ICP-match conditions not tied to one ICP, CRM push conditions set to any of their ICPs, and the ICP match on lists; lead scores, which count only the team owner's ICPs, count it when contacts are next rescored. Through those conditions it can change which contacts the user's active automations act on (their steps can add contacts to campaigns that email and message them, post to Slack or act in the user's connected apps) and which new contacts FuseAI creates in the team's connected CRM, such as HubSpot or Salesforce. Criteria come in three sections. People: the person (name, location, job title keywords and titles, seniority, the person's own industry, languages, skills, experience, education, LinkedIn URL, open to work, unemployed). Company: the person's current company (name, keywords its name or description must or must not carry, department, industry, headquarters location, size, website, revenue, funding, employee count). Past job: companies, titles, seniority and departments the person had before. Include/exclude criteria take { status: 'include' | 'exclude', value, label } entries. The person's location and the company's headquarters location are separate criteria, each at three levels: for the person, country, locationRegion (state) and personalLocation (city); for the company, hqCountry, jobCompanyLocationRegion (state) and location (city). A country is a lowercase full name such as "germany". A city is written "city, state, country", and a bare place name in a city criterion is refused. The industry criterion is the person's own industry; for the industries of the companies the ICP targets, use jobCompanyIndustry.
Inputs
namestringrequired- Name for the ICP.
actualEmployeeRangeobject- Company: number of employees at the person's current company, as { min, max }, both required.
certificationstring- People: a certification the person holds, in lowercase, e.g. "pmp".
companyKeywordsarray- Company: words the company's name or description must carry, any of them, for the kind of company when no industry value names it, e.g. "fintech" or "agency". Each is a lowercase word or short phrase, matched as a whole word with its plural; a term with * or ? is refused. At most 25.
countryarray- People location, country level: countries the person lives in. Each value is a lowercase full country name, e.g. "germany" or "united states".
currentCompanyNamearray- Company: the person's current company, by name, e.g. "stripe".
currentCompanyWebsitearray- Company: the person's current company, by website domain, e.g. "stripe.com".
currentJobTitlearray- People: exact current job titles in lowercase, e.g. "vp of sales". Use keywords to match part of a title.
currentRevenuearray- Company: annual revenue band of the person's current company. Values: $0-$1M, $1M-$10M, $10M-$25M, $25M-$50M, $50M-$100M, $100M-$250M, $250M-$500M, $500M-$1B, $1B-$10B, $10B+.
degreearray- People: degrees the person holds, in lowercase, e.g. "bachelors", "masters" or "master of business administration". Use include entries only.
departmentarray- Company: the person's department at their current company. Values: advisory, analyst, creative, education, engineering, finance, fulfillment, health, hospitality, human_resources, legal, manufacturing, marketing, operations, partnerships, product, professional_service, public_service, research, sales, sales_engineering, support, trade, unemployed.
descriptionstring- A sentence or two on who the ICP targets.
excludedCompanyKeywordsarray- Company: words that remove a company whose name or description carries any of them, e.g. "agency". Same matching as companyKeywords. At most 25.
excludedKeywordsarray- People: words that remove a person whose current job title matches any of them, e.g. "intern". Same matching as keywords. At most 25 terms, and a term needs text other than the wildcards * and ?.
fullNamestring- People: the person's full name in lowercase, e.g. "jane doe".
fundingRaisedobject- Company: total funding the person's current company has raised, in US dollars, as { min, max }, both required.
hqCountryarray- Company location, country level: countries where the person's current company has its headquarters. Each value is a lowercase full country name, e.g. "germany" or "united states".
industryarray- People: the person's own industry (Person Industry in the app), in lowercase, e.g. "computer software". It is not the industry of their company: for the industries the ICP targets, use jobCompanyIndustry.
isOpenToWorkboolean- People: true keeps only people whose LinkedIn profile shows Open to Work. Leave it out otherwise.one of
true isUnemployedboolean- People: true keeps only people who hold no current job. Leave it out otherwise.one of
true jobCompanyIndustryarray- Company: the industry of the person's current company (Company Industry in the app). Use it for the industries the ICP targets. Same lowercase values as industry, e.g. "computer software".
jobCompanyLocationRegionarray- Company location, state level: states or regions where the person's current company has its headquarters, in lowercase, e.g. "texas".
keywordsarray- People: words to look for in the person's current job title, e.g. "growth". A bare word matches anywhere in the title; a term with * or ? must match the whole title, e.g. "manager*" for titles starting with manager. Matching ignores case.
languagesSpokenarray- People: languages the person speaks, as lowercase names, e.g. "german".
levelarray- People: seniority of the person's current job. Values: cxo, vp, director, manager, senior, entry, owner, partner, training, unpaid.
linkedinUrlstring- People: the person's LinkedIn profile URL, e.g. "linkedin.com/in/jane-doe".
locationarray- Company location, city level: cities where the person's current company has its headquarters. Write each as "city, state, country" in lowercase, e.g. "austin, texas, united states", or as "city, country" for a place with no state. A bare place name is refused: use hqCountry for a whole country and jobCompanyLocationRegion for a state.
locationRegionarray- People location, state level: states or regions the person lives in, in lowercase, e.g. "bavaria" or "california".
numbersOfMonthInCurrentRolenumber- People: number of months the person has been in their current role.
pastCompanyNamearray- Past job: companies the person worked at before, by name, e.g. "oracle".
pastCompanyWebsitearray- Past job: companies the person worked at before, by website domain, e.g. "oracle.com".
pastJobDepartmentarray- Past job: department of a job the person held before. Same values as department.
pastJobLevelsarray- Past job: seniority of a job the person held before. Same values as level.
pastJobTitlearray- Past job: exact job titles the person held before, in lowercase, e.g. "account executive".
personalLocationarray- People location, city level: cities the person lives in. Write each as "city, state, country" in lowercase, e.g. "munich, bavaria, germany", or as "city, country" for a place with no state. A bare place name is refused: use country for a whole country and locationRegion for a state.
schoolNamearray- People: schools the person attended, in lowercase, e.g. "stanford university".
sizearray- Company: headcount band of the person's current company. Values: 1-10, 11-50, 51-200, 201-500, 501-1000, 1001-5000, 5001-10000, 10001+.
skillsarray- People: skills on the person's profile, in lowercase, e.g. "python".
totalYearsOfExperienceobject- People: the person's total years of work experience, as { min, max }, both required.
fuseaimcp_create_job_changes_watcher_monitorCreate a Job Changes watcher: track the shared contacts with a LinkedIn URL on one or more SOURCE prospect lists for a new employer, a promotion or a new title (custom contacts are not watched).DestructiveCreate job-changes watcher monitor
Create a Job Changes watcher: track the shared contacts with a LinkedIn URL on one or more SOURCE prospect lists for a new employer, a promotion or a new title (custom contacts are not watched). Each contact's public LinkedIn profile is re-checked through Fiber, an external data provider that receives the contacts' LinkedIn URLs, monthly by default (frequency 30; 7 for weekly, 1 for daily) until the watcher is paused or deleted, and every check costs 50 credits per watched contact; each change is added to a new dedicated results list once a check finds it, and the person's new title and employer are written onto their CRM contact records, replacing the stored ones. Each change can start the user's active automations that use the Signal match trigger, and enrichEmail/enrichPhone also look up the email/phone of each changed contact, at additional credits. listId or listIds (give exactly one) identifies the existing SOURCE list(s) to watch; listName is the name of BOTH the agent and the NEW results list it creates — it must be unique and must NOT reuse a source list's name (name it after the agent the user is creating, e.g. watching 'fuseai-team' → an agent/list called 'fuseai-team Job Changes').
Inputs
listNamestringrequired- Name for BOTH the agent and the NEW dedicated results list this watcher creates. Must be unique — never reuse a source list's name; name it after the agent the user is creating.
maxCreditSpendintegerrequired- Required. The most credits one check may cost (1000 to 200000). Every check costs 50 credits per watched contact or company, so the watcher watches at most maxCreditSpend / 50 of them (rounded down) from its source lists; the rest are not watched or billed.
enrichEmailboolean- No description.default
false enrichPhoneboolean- No description.default
false frequencynumber- How often each watched contact is re-checked, in days: 30 (monthly, the default), 7 (weekly) or 1 (daily). Every check costs 50 credits per watched contact.one of
1730default30 initializeOnlyboolean- Omit or false to create and START the monitor immediately (sets up the live subscription and spends credits). Set true to save it as a DRAFT for the user to review and activate later.default
false listIdstring- The existing SOURCE list(s) to watch, by id — the people/companies whose changes are tracked.
listIdsarray- The existing SOURCE list(s) to watch, by id — the people/companies whose changes are tracked.
maxContactsinteger- Max Results: the watcher pauses once this many results have been added to its results list. It does not limit how many contacts or companies are watched or what a check costs; maxCreditSpend does.
fuseaimcp_create_job_posting_in_location_monitorCreate a Job-Posting-in-Location signal monitor: find companies posting jobs in one region, optionally narrowed by industry, headcount, HQ country or last funding round.DestructiveCreate job-posting-in-location monitor
Create a Job-Posting-in-Location signal monitor: find companies posting jobs in one region, optionally narrowed by industry, headcount, HQ country or last funding round. Unless initializeOnly is true, it starts at once: it creates a new company list named listName and a daily watch with CrustData, an external company data provider, that runs until expirationDate (one month out by default; null for no end). Each hiring company found is added to the user's saved companies (a company the user had removed is saved again) and to the list with a link to the posting, at 100 credits per company added (300 with realtime). A background lookup with Fiber, an external company data provider, then fills empty fields on those shared company records and replaces a name that is still just the domain. Each match can start the user's active automations that use the Company signal match trigger.
Inputs
listNamestringrequired- No description.
regionsarrayrequired- No description.
companyCountriesarray- No description.
companyHeadcountarray- Company-size bucket — use one of exactly: 1-10, 11-50, 51-200, 201-500, 501-1,000, 1,001-5,000, 5,001-10,000, 10,001+.
expirationDatestring- No description.
industriesarray- No description.
initializeOnlyboolean- Omit or false to create and START the monitor immediately (sets up the live subscription and spends credits). Set true to save it as a DRAFT for the user to review and activate later.default
false lastFundingRoundTypesarray- No description.
maxContactsinteger- No description.
maxCreditSpendinteger- Maximum credits this agent may spend in a single run or refresh (1000 to 200000). CrustData monitors run once a day, so for them it is a daily limit; job-change watchers watch only as many contacts as one check within it pays for; web monitors stop once their total delivered results reach it. Omit for no limit.
notificationCountinteger- No description.
realtimeboolean- Run a realtime agent: a live search for same-day results at about 300x the provider cost of the default. Off by default; only set it when the user asks for realtime. A realtime agent delivers 5 results a run unless notificationCount is set.
fuseaimcp_create_job_posting_monitorCreate a Job Postings signal monitor: find companies hiring for specific titles/keywords in a region.DestructiveCreate job-posting monitor
Create a Job Postings signal monitor: find companies hiring for specific titles/keywords in a region. Requires a region and at least one of titles or descriptions. Unless initializeOnly is true, it starts at once: it creates a new company list named listName and a daily watch with CrustData, an external company data provider, that runs until expirationDate (one month out by default; null for no end). Each hiring company found is added to the user's saved companies (a company the user had removed is saved again) and to the list with a link to the posting, at 100 credits per company added (300 with realtime). A background lookup with Fiber, an external company data provider, then fills empty fields on those shared company records and replaces a name that is still just the domain. Each match can start the user's active automations that use the Company signal match trigger.
Inputs
listNamestringrequired- No description.
regionsarrayrequired- No description.
companyCountriesarray- No description.
companyHeadcountarray- Company-size bucket — use one of exactly: 1-10, 11-50, 51-200, 201-500, 501-1,000, 1,001-5,000, 5,001-10,000, 10,001+.
descriptionsarray- No description.
exactIndustryMatchboolean- No description.default
false exactKeywordMatchboolean- No description.default
true expirationDatestring- No description.
industriesarray- No description.
initializeOnlyboolean- Omit or false to create and START the monitor immediately (sets up the live subscription and spends credits). Set true to save it as a DRAFT for the user to review and activate later.default
false lastFundingRoundTypesarray- No description.
maxContactsinteger- No description.
maxCreditSpendinteger- Maximum credits this agent may spend in a single run or refresh (1000 to 200000). CrustData monitors run once a day, so for them it is a daily limit; job-change watchers watch only as many contacts as one check within it pays for; web monitors stop once their total delivered results reach it. Omit for no limit.
notificationCountinteger- No description.
realtimeboolean- Run a realtime agent: a live search for same-day results at about 300x the provider cost of the default. Off by default; only set it when the user asks for realtime. A realtime agent delivers 5 results a run unless notificationCount is set.
titlesarray- No description.
fuseaimcp_create_knowledge_hub_profileCreate the Knowledge Hub profile of the user's team from one or more website URLs (async — it returns 'processing' and the profile finishes building in the background): it scrapes those websites from the public web, with their subpages when includeSubPages is set, and writes the profile's pages with AI.DestructiveCreate Knowledge Hub profile
Create the Knowledge Hub profile of the user's team from one or more website URLs (async — it returns 'processing' and the profile finishes building in the background): it scrapes those websites from the public web, with their subpages when includeSubPages is set, and writes the profile's pages with AI. At modelEffort 'high' it costs credits: a one-time charge per website URL when the build succeeds, plus a monthly charge for each of refreshEnabled and learning while it is on. At 'fast' (the default) the build, refreshEnabled and learning are all free. refreshEnabled re-scrapes the sources weekly and learning reads the team's emails, LinkedIn messages and call transcripts stored in Fuse every day; both regenerate the profile's pages, replacing manual edits to them. A team owns one profile, shared by everyone on it: this fails with PROFILE_ALREADY_EXISTS once the team has a built or building one, so update that one instead; a draft (saved during onboarding or left by a failed build) does not block it and is replaced by the new profile.
Inputs
rootUrlsarrayrequired- The website(s) to build the profile from.
additionalInfoarray- Extra freeform context to include.
learningobject- Opt-in learning from team communications (default off).
modelEffortstring- Synthesis/learning quality tier; drives model choice and credit rates (default fast).one of
fasthigh profileNamestring- Name for the new profile.
refreshEnabledboolean- Weekly source refresh opt-in (default off; bills monthly while enabled).default
false
fuseaimcp_create_linkedin_post_monitorCreate a LinkedIn Posts signal monitor: find people posting about specific topics/keywords on LinkedIn, optionally narrowed by author company/title/location and industry.DestructiveCreate LinkedIn-post monitor
Create a LinkedIn Posts signal monitor: find people posting about specific topics/keywords on LinkedIn, optionally narrowed by author company/title/location and industry. Requires 1-5 keywords and at least one of companyHeadcount or companyCountries; when the user named neither, pass all eight companyHeadcount buckets ['1-10', '11-50', '51-200', '201-500', '501-1,000', '1,001-5,000', '5,001-10,000', '10,001+'] to cover any company size. Unless initializeOnly is true, it starts at once and searches public LinkedIn posts through an outside data provider once a day until expirationDate (one month by default; null means no end date) or until paused, adding the people it finds to a new list named listName. Each person added costs 100 credits (300 with realtime), and 2 more when the user has no email or phone saved for them, plus enrichment credits when enrichEmail or enrichPhone is on. A person who is already a contact has their saved profile fields (such as name, title and company) overwritten with the found profile, and each person added can start the user's automations, which can add them to campaigns or act in connected apps.
Inputs
keywordsarrayrequired- No description.
listNamestringrequired- No description.
actorTypearray- No description.
authorCompanyarray- No description.
authorLocationarray- No description.
authorTitlearray- No description.
companyCountriesarray- No description.
companyHeadcountarray- Company-size bucket — use one of exactly: 1-10, 11-50, 51-200, 201-500, 501-1,000, 1,001-5,000, 5,001-10,000, 10,001+.
currentCompanyarray- No description.
currentTitlearray- No description.
detailedReactorDataboolean- No description.
enrichEmailboolean- No description.default
false enrichPhoneboolean- No description.default
false excludeAuthorCompaniesarray- No description.
excludeAuthorTitlesarray- No description.
excludePostCategoriesarray- No description.
excludePostIntentarray- No description.
expirationDatestring- No description.
industriesarray- No description.
initializeOnlyboolean- Omit or false to create and START the monitor immediately (sets up the live subscription and spends credits). Set true to save it as a DRAFT for the user to review and activate later.default
false leadCompanyCountriesarray- No description.
leadCompanyHeadcountarray- Company-size bucket — use one of exactly: 1-10, 11-50, 51-200, 201-500, 501-1,000, 1,001-5,000, 5,001-10,000, 10,001+.
maxContactsinteger- No description.
maxCreditSpendinteger- Maximum credits this agent may spend in a single run or refresh (1000 to 200000). CrustData monitors run once a day, so for them it is a daily limit; job-change watchers watch only as many contacts as one check within it pays for; web monitors stop once their total delivered results reach it. Omit for no limit.
notificationCountinteger- No description.
pastCompanyarray- No description.
pastTitlearray- No description.
postCategoryarray- No description.
postedWithinstring- How far back each daily run looks for posts: now-24h, now-7d (default), now-1M, now-3M or now-1y.one of
now-24hnow-7dnow-1Mnow-3Mnow-1y postIntentarray- No description.
reactorsboolean- No description.
realtimeboolean- Run a realtime agent: a live search for same-day results at about 300x the provider cost of the default. Off by default; only set it when the user asks for realtime. A realtime agent delivers 5 results a run unless notificationCount is set.
fuseaimcp_create_linkedin_post_scrapeScrape the people who liked or commented on specific public LinkedIn posts into a new list named listName — 'LinkedIn Post Engagement'.DestructiveCreate LinkedIn post scrape
Scrape the people who liked or commented on specific public LinkedIn posts into a new list named listName — 'LinkedIn Post Engagement'. Unless initializeOnly is true, it starts a background scrape through an outside scraping service at once; enableRecurring scrapes the same posts once more 7 days later and adds new people to that list. Requires postUrls (or a single postUrl), fieldsToScrape and a listName that no existing list or LinkedIn scrape uses. Each person saved costs 50 credits, and 2 more when the user has no email or phone saved for them, plus enrichment credits when enrichEmail or enrichPhone is on. A person who is already a contact has their saved profile fields (such as name, title and company) overwritten with the scraped profile, and each saved person can start the user's automations, which can add them to campaigns or act in connected apps.
Inputs
fieldsToScrapearrayrequired- No description.
listNamestringrequired- No description.
enableRecurringboolean- No description.default
false enrichEmailboolean- No description.default
false enrichPhoneboolean- No description.default
false initializeOnlyboolean- Omit or false to create and START the scrape immediately (spends credits). Set true to save it as a DRAFT for the user to review and activate later.default
false maxCreditSpendinteger- Maximum credits this agent may spend in a single run or refresh (1000 to 200000). CrustData monitors run once a day, so for them it is a daily limit; job-change watchers watch only as many contacts as one check within it pays for; web monitors stop once their total delivered results reach it. Omit for no limit.
postUrlstring- No description.
postUrlsarray- No description.
fuseaimcp_create_linkedin_profile_engagement_scrapeScrape the people who liked or commented on the recent public posts of specific LinkedIn profiles (not the profiles themselves) into a new list named listName.DestructiveCreate LinkedIn Profile Engagement Scrape
Scrape the people who liked or commented on the recent public posts of specific LinkedIn profiles (not the profiles themselves) into a new list named listName. Unless initializeOnly is true, it starts a background scrape through an outside scraping service at once; enableRecurring repeats it every day until recurringExpiresAt, or until paused when that is not set. Requires profileUrls (or a single profileUrl), fieldsToScrape and a listName that no existing list or LinkedIn scrape uses. Each person saved costs 50 credits, and 2 more when the user has no email or phone saved for them, plus enrichment credits when enrichEmail or enrichPhone is on. A person who is already a contact has their saved profile fields (such as name, title and company) overwritten with the scraped profile, and each saved person can start the user's automations, which can add them to campaigns or act in connected apps.
Inputs
fieldsToScrapearrayrequired- No description.
listNamestringrequired- No description.
enableRecurringboolean- No description.default
false enrichEmailboolean- No description.default
false enrichPhoneboolean- No description.default
false includeRepostsboolean- No description.default
false initializeOnlyboolean- Omit or false to create and START the scrape immediately (spends credits). Set true to save it as a DRAFT for the user to review and activate later.default
false maxCreditSpendinteger- Maximum credits this agent may spend in a single run or refresh (1000 to 200000). CrustData monitors run once a day, so for them it is a daily limit; job-change watchers watch only as many contacts as one check within it pays for; web monitors stop once their total delivered results reach it. Omit for no limit.
maxPostsinteger- No description.default
5 postedLimitstring- No description.one of
24hweekmonth3monthsyeardefaultyear profileUrlstring- No description.
profileUrlsarray- No description.
recurringExpiresAtstring- No description.
fuseaimcp_create_people_watcher_monitorCreate a People Watcher signal monitor: watch the people in one or more SOURCE prospect lists for profile/job changes, new LinkedIn posts, comments and reactions on LinkedIn posts, turning on Open to Work, or turning on Hiring.DestructiveCreate people-watcher monitor
Create a People Watcher signal monitor: watch the people in one or more SOURCE prospect lists for profile/job changes, new LinkedIn posts, comments and reactions on LinkedIn posts, turning on Open to Work, or turning on Hiring. listId/listIds identify the existing SOURCE list(s) to watch; also needs an eventType and a listName that names BOTH the agent and a NEW dedicated results list it creates — listName must be unique and must NOT reuse a source list's name (name it after the agent the user is creating). Unless initializeOnly is true, it starts at once: the watched people's LinkedIn profile URLs are sent to an outside tracking provider, which checks them every `frequency` days until expirationDate (one month by default; null means no end date) or until paused. Each check costs 50 credits per watched contact (maxCreditSpend caps how many are watched), each delivered person costs 2 credits when the user has no email or phone saved for them, and enrichment credits apply when enrichEmail or enrichPhone is on. Each delivered person is saved again from the profile found for them, overwriting their saved contact's profile fields (such as title and company after a job change), and each delivery can start the user's automations, which can add people to campaigns or act in connected apps.
Inputs
eventTypestringrequired- No description.one of
linkedin-person-profile-updateslinkedin-person-post-updateslinkedin-person-post-engagementlinkedin-person-open-to-worklinkedin-person-hiring listNamestringrequired- Name for BOTH the agent and the NEW dedicated results list this watcher creates. Must be unique — never reuse a source list's name; name it after the agent the user is creating.
maxCreditSpendintegerrequired- Required. The most credits one check may cost (1000 to 200000). Every check costs 50 credits per watched contact or company, so the watcher watches at most maxCreditSpend / 50 of them (rounded down) from its source lists; the rest are not watched or billed.
enrichEmailboolean- No description.default
false enrichPhoneboolean- No description.default
false expirationDatestring- No description.
frequencyinteger- No description.default
1 initializeOnlyboolean- Omit or false to create and START the monitor immediately (sets up the live subscription and spends credits). Set true to save it as a DRAFT for the user to review and activate later.default
false listIdstring- The existing SOURCE list(s) to watch, by id — the people/companies whose changes are tracked.
listIdsarray- The existing SOURCE list(s) to watch, by id — the people/companies whose changes are tracked.
maxContactsinteger- Max Results: the watcher pauses once this many results have been added to its results list. It does not limit how many contacts or companies are watched or what a check costs; maxCreditSpend does.
signalFiltersobject- Optional filters on the watched signals. Each key names the eventType that takes it and is refused with any other; omit for every signal of the type.
fuseaimcp_create_person_discovery_via_filters_monitorCreate a Person-Discovery-via-Filters signal monitor: once a day, find people in an outside people-data provider who newly match a filter set and add them to a new list named listName.DestructiveCreate person-discovery monitor
Create a Person-Discovery-via-Filters signal monitor: once a day, find people in an outside people-data provider who newly match a filter set and add them to a new list named listName. At least one filter must be provided. Unless initializeOnly is true, it starts at once and runs until expirationDate (one month by default; null means no end date) or until paused. Each person added costs 100 credits (300 with realtime), and 2 more when the user has no email or phone saved for them, plus enrichment credits when enrichEmail or enrichPhone is on. A person who is already a contact has their saved profile fields (such as name, title and company) overwritten with the found profile, and each person added can start the user's automations, which can add them to campaigns or act in connected apps.
Inputs
listNamestringrequired- No description.
companyCountriesarray- No description.
companyHeadcountarray- Company-size bucket — use one of exactly: 1-10, 11-50, 51-200, 201-500, 501-1,000, 1,001-5,000, 5,001-10,000, 10,001+.
currentCompanyarray- No description.
currentTitlearray- No description.
enrichEmailboolean- No description.default
false enrichPhoneboolean- No description.default
false excludeCompanyCountriesarray- No description.
excludeCountriesarray- No description.
excludeCurrentCompaniesarray- No description.
excludeCurrentTitlesarray- No description.
excludeIndustriesarray- No description.
excludePastCompaniesarray- No description.
excludePastTitlesarray- No description.
excludeSchoolsarray- No description.
excludeSeniorityLevelsarray- No description.
expirationDatestring- No description.
firstNamestring- No description.
headlineKeywordstring- No description.
industriesarray- No description.
initializeOnlyboolean- Omit or false to create and START the monitor immediately (sets up the live subscription and spends credits). Set true to save it as a DRAFT for the user to review and activate later.default
false lastCompanyarray- No description.
lastNamestring- No description.
maxContactsinteger- No description.
maxCreditSpendinteger- Maximum credits this agent may spend in a single run or refresh (1000 to 200000). CrustData monitors run once a day, so for them it is a daily limit; job-change watchers watch only as many contacts as one check within it pays for; web monitors stop once their total delivered results reach it. Omit for no limit.
mentionedInNewsboolean- No description.
notificationCountinteger- No description.
pastCompanyarray- No description.
pastTitlearray- No description.
realtimeboolean- Run a realtime agent: a live search for same-day results at about 300x the provider cost of the default. Off by default; only set it when the user asks for realtime. A realtime agent delivers 5 results a run unless notificationCount is set.
recentlyChangedJobsboolean- No description.
recentlyPostedOnLinkedInboolean- No description.
regionsarray- No description.
schoolsarray- No description.
seniorityLevelsarray- Seniority of the current role — use values from: Owner / Partner, CXO, Vice President, Director, Experienced Manager, Entry Level Manager, Strategic, Senior, Entry Level, In Training.
summaryKeywordstring- No description.
yearsAtCurrentCompanystring- Years at the current employer — one of: Less than 1 year, 1 to 2 years, 3 to 5 years, 6 to 10 years, More than 10 years.one of
Less than 1 year1 to 2 years3 to 5 years6 to 10 yearsMore than 10 years yearsInCurrentPositionstring- Years in the current role — one of: Less than 1 year, 1 to 2 years, 3 to 5 years, 6 to 10 years, More than 10 years.one of
Less than 1 year1 to 2 years3 to 5 years6 to 10 yearsMore than 10 years yearsOfExperiencestring- Total years of experience — one of: Less than 1 year, 1 to 2 years, 3 to 5 years, 6 to 10 years, More than 10 years.one of
Less than 1 year1 to 2 years3 to 5 years6 to 10 yearsMore than 10 years
fuseaimcp_create_person_starting_new_job_monitorCreate a Person-Starting-New-Job signal monitor: once a day, find people whose public LinkedIn posts announce a new job, through an outside data provider, scoped to an industry plus optional company/title filters, and add them to a new list named listName.DestructiveCreate person-starting-new-job monitor
Create a Person-Starting-New-Job signal monitor: once a day, find people whose public LinkedIn posts announce a new job, through an outside data provider, scoped to an industry plus optional company/title filters, and add them to a new list named listName. Unless initializeOnly is true, it starts at once and runs until expirationDate (one month by default; null means no end date) or until paused. Each person added costs 100 credits (300 with realtime), and 2 more when the user has no email or phone saved for them, plus enrichment credits when enrichEmail or enrichPhone is on. A person who is already a contact has their saved profile fields (such as name, title and company) overwritten with the found profile, and each person added can start the user's automations, which can add them to campaigns or act in connected apps.
Inputs
industriesarrayrequired- No description.
listNamestringrequired- No description.
authorCompanyarray- No description.
authorTitlearray- No description.
currentCompanyarray- No description.
currentTitlearray- No description.
detailedReactorDataboolean- No description.
enrichEmailboolean- No description.default
false enrichPhoneboolean- No description.default
false excludeAuthorCompaniesarray- No description.
excludeAuthorTitlesarray- No description.
expirationDatestring- No description.
initializeOnlyboolean- Omit or false to create and START the monitor immediately (sets up the live subscription and spends credits). Set true to save it as a DRAFT for the user to review and activate later.default
false leadCompanyCountriesarray- No description.
leadCompanyHeadcountarray- Company-size bucket — use one of exactly: 1-10, 11-50, 51-200, 201-500, 501-1,000, 1,001-5,000, 5,001-10,000, 10,001+.
maxContactsinteger- No description.
maxCreditSpendinteger- Maximum credits this agent may spend in a single run or refresh (1000 to 200000). CrustData monitors run once a day, so for them it is a daily limit; job-change watchers watch only as many contacts as one check within it pays for; web monitors stop once their total delivered results reach it. Omit for no limit.
notificationCountinteger- No description.
pastCompanyarray- No description.
pastTitlearray- No description.
postedWithinstring- How far back each daily run looks for posts: now-24h, now-7d (default), now-1M, now-3M or now-1y.one of
now-24hnow-7dnow-1Mnow-3Mnow-1y reactorsboolean- No description.
realtimeboolean- Run a realtime agent: a live search for same-day results at about 300x the provider cost of the default. Off by default; only set it when the user asks for realtime. A realtime agent delivers 5 results a run unless notificationCount is set.
fuseaimcp_create_prospect_listCreate a new list.DestructiveCreate Prospect List
Create a new list. Provide name (required; it must not match a system list name or one of the user's existing lists), entityType (contactList, the default, or companyList), and optionally seed members via contacts or companies (arrays of ids). status public (the default) shows the list to the user's teammates too; private keeps it out of their list views. isDynamic (default false) sets the dynamic label FuseAI gives lists that its monitors keep updating, but does not make this list fill or refresh itself. completionStatus processing (the default is complete) makes the list report that it is still being filled, and this call never changes it to complete. Seeding contacts can cost up to 2 credits per contact, and the call fails with INSUFFICIENT_BALANCE when the balance is too low. Seeded members do not start the user's 'Added to list' or 'Company added to list' automations. When a company list is seeded with companies, a background job looks up each not-yet-enriched company by its domain with an outside company-data provider and writes the result onto that shared company record, which every workspace that saved the company sees: it adds the provider's details, fills empty fields, and replaces a name that is still just the domain. Returns the created list with its id.
Inputs
namestringrequired- No description.
companiesarray- No description.
completionStatusstring- No description.one of
processingcomplete contactsarray- No description.
entityTypestring- No description.one of
contactListcompanyList isDynamicboolean- No description.
statusstring- No description.one of
publicprivate
fuseaimcp_create_sequence_step_toolAdds a new sequence step to a campaign at an explicit position, specifying channel, wait days, type, and content.DestructiveCreate Sequence Step
Adds a new sequence step to a campaign at an explicit position, specifying channel, wait days, type, and content. A step placed first in the root list sets the previous first step's wait to 3 days if it was 0. Requires campaignId, sequenceNumber, sequenceStepType ('manual' uses the content given; 'ai_generated' has AI write each contact's message) and daysBetween (whole days to wait after the previous step); channel defaults to 'email'. For sequenceStepType 'ai_generated' on an email, LinkedIn message or InMail step, AI writes the step's topic and title from the campaign's focus and company details, and any topic passed is ignored. sequenceNumber is the 0-based position within the target list: the root list, or the lane named by parentStepId (a condition step's id) and branchKey ('yes' or 'no'; 'a' or 'b' under an A/B split). Nothing can be placed after a list's last step when that step is a condition. To branch, add a step with channel 'condition', conditionType 'invite_accepted' (needs a LinkedIn connection step earlier on its path) or 'clicked_link' (needs an email step earlier on its path), sequenceStepType 'manual' and daysBetween as the window (1-30 days); it is appended to its list, and its lanes are then filled by adding steps with its id as parentStepId. To A/B test two paths, add a step with channel 'condition', conditionType 'ab_split', sequenceStepType 'manual', daysBetween 0 and optionally splitArms [{ key: 'a', label?, weight }, { key: 'b', label?, weight }] (whole percents adding up to 100; omitted means 50/50); it may be the first step, and each contact reaching it goes to one arm in the weights' proportions, by a fixed hash of the contact and the split, and stays there. Fill each arm with steps whose parentStepId is the split's id and branchKey 'a' or 'b'; an arm may stay empty to hold contacts out. Splits and conditions can only be added before the campaign starts, and no step can be added once a campaign with a split or condition has started. On a started campaign without them, the new step joins the live sequence: contacts still in it are sent the new step as well, from the user's connected mailbox or LinkedIn account. To choose which of the user's mailboxes send an email step, pass allowedInboxes built from list_email_accounts, as [{ inboxId: the mailbox's email address, emailAccountType: its account value }]; omitted, any of the user's usable mailboxes may send it. Other fields: template { subject, body, format 'editor' or 'html' } is a 'manual' step's message, and topic the brief an AI-written step is written from; invitationText (linkedin_connection, up to 200 characters) is the note sent with the invite; voiceNoteScript (linkedin_message, up to 100 words) adds a voice note in the sender's cloned voice, and each one sent costs credits; a linkedin_like step reacts with reactionType (like, celebrate, support, love, insightful, funny, or auto for AI's pick; default like) to the contact's newest post from the last postRecencyDays days (1-365, default 30), and is skipped if there is none; inmailExhaustionBehavior (inmail) is 'skip' (default) or 'defer' (once, to the next monthly refill) for when the sending LinkedIn account has no InMail credits left; cc (email) lists extra email addresses copied on every email the step sends; replyInSameThread (email, inmail) false starts a new thread instead of replying in the contact's earlier one; executionMode 'manual' (always for dialer and todo, never for linkedin_visit or linkedin_like) raises a task for a person and holds the contact until it is done or skipped, while 'automatic' (the default) has the campaign perform the step; autoSkipAfterDays (1-365) skips such a task left unstarted that many days after it is due, and taskInstructions and taskLinkUrl (an http(s) link) are shown on it; channelSequenceNumber has no effect, as the server renumbers every step. In email/message content, only these personalization tokens are substituted (single braces, exact names): {Contact Full Name}, {Contact First Name}, {Contact Last Name}, {Contact Company}, {Contact Job Title}, {Contact Industry}, {Sender Full Name}, {Sender First Name}, {Sender Last Name}, {Sender Job Title}, {Sender Signature}. The {Sender …} tokens take the name, title and signature of the mailbox or LinkedIn account that sends the step; the campaign's sender name and job title fill in only when that account has none. Two more token kinds are resolved per contact: {Column: <list column name>} inserts the contact's value from that column of the campaign's lists, and {{<instruction>}} has AI write that part for each contact from their data ({{<instruction> :: <original text>}} rewrites existing words, keeping them as the fallback). Anything else is sent to the prospect literally. When the campaign writes each message, an 'ai_generated' step or a {{...}} part costs credits.
Inputs
campaignIdstringrequired- No description.
sequenceNumberintegerrequired- 0-based position within the target list (the root list, or the lane named by parentStepId + branchKey). Later steps of that list shift down; a condition step is always appended. The server then renumbers the whole campaign in sequence-tree order.
allowedInboxesstring- No description.
autoSkipAfterDaysstring- No description.
branchKeystring- Which lane of parentStepId. Under an invite_accepted or clicked_link condition: 'yes' (the signal arrived in time) or 'no'. Under an ab_split: the arm, 'a' or 'b'. Sent together with parentStepId.one of
yesnoab ccstring- No description.
channelstring- No description.one of
emaillinkedin_connectionlinkedin_messagelinkedin_visitlinkedin_likeinmaildialertodocondition channelSequenceNumberinteger- No description.
conditionTypestring- No description.
daysBetweenstring- No description.
executionModestring- No description.
inmailExhaustionBehaviorstring- No description.
invitationTextstring- No description.
parentStepIdstring- The condition step whose lane this list is. Omit (or null) for the root list. Sent together with branchKey.
postRecencyDaysstring- No description.
reactionTypestring- No description.
replyInSameThreadstring- No description.
sequenceStepTypestring- No description.
splitArmsstring- No description.
taskInstructionsstring- No description.
taskLinkUrlstring- No description.
templatestring- No description.
topicstring- No description.
voiceNoteScriptstring- No description.
fuseaimcp_create_smart_column_toolAdds an AI 'smart column' that researches/scores each row of a list from a prompt (standard and deep_research search the web), or (engine writer) writes a message for each row.DestructiveCreate Smart Column
Adds an AI 'smart column' that researches/scores each row of a list from a prompt (standard and deep_research search the web), or (engine writer) writes a message for each row. Campaigns that use the list read these values when they prepare each contact's next message: a step with a {Column: <columnName>} token holds each contact until their row is filled and then inserts the value as written, with no review step, and AI-written steps use the values as context. Starts a background job and returns a jobId; poll get_smart_column_job_status_tool. Costs credits for each row computed: 20 (standard), 100 (deep_research) or 10 (writer), and is refused when the balance cannot cover every target row. Each non-empty value also triggers the user's 'Smart column updated' automations. Provide listId, prompt, columnName and runOn (first_1|first_10|first_100|first_1000|all). columnName must be new: a name already on the list or a reserved column name is refused, so it never changes an existing column. Optional filters (the same row filters as get_prospect_list_rows_tool) and sort ({ sortBy, sortOrder }) pick and order the target rows, runOn takes the first 1, 10, 100, 1,000 or all of them, and only those rows are computed and charged; if none match, the call is refused unless allowEmpty is true, which keeps the empty column and charges nothing. providerIds (deep_research only; ids from get_provider_key_catalog) narrows which outside data providers the research may call, at no extra credits. When deep_research calls a provider the user saved their own API key for, it uses that key, so that provider bills the user's own account.
Inputs
columnNamestringrequired- No description.
listIdstringrequired- No description.
promptstringrequired- No description.
runOnstringrequired- No description.one of
first_1first_10first_100first_1000all allowEmptyboolean- No description.default
false enginestring- How each row is filled. standard (the default): a quick web lookup finds the row's value. deep_research: in-depth research across many sources, slower. writer: writes a message for each row, such as an email opener, a LinkedIn note or a subject line, from the row's other columns, the team's Knowledge Hub and, when the team's People Brain is on, what it knows about the contact and their company, in the user's writing style and with no web research; the prompt says what to write. A row the writer cannot fill is left failed, never given a placeholder.one of
standarddeep_researchwriterdefaultstandard filtersobject- No description.
providerIdsarray- No description.
sortobject- No description.
fuseaimcp_create_taskCreate a task in the Tasks hub: a to-do, call, email or LinkedIn step for the caller or a teammate (assignedToUserId from list_team_members; defaults to the caller and must be an active team member).DestructiveCreate Task
Create a task in the Tasks hub: a to-do, call, email or LinkedIn step for the caller or a teammate (assignedToUserId from list_team_members; defaults to the caller and must be an active team member). Optional contactId links it to a CRM contact; dueAt defaults to end of today in the given timezone. An email or LinkedIn task is only a reminder for the assignee: creating it sends nothing to the contact and adds nothing to a campaign. If the assignee linked their phone to Dash, FuseAI's assistant, on iMessage and has urgent-task alerts on, a high or urgent task created between 8am and 8pm their local time may also make Dash text them about it; a sent text cannot be unsent. Tasks the caller assigns to themselves are not texted, except calls linked to a contact. Returns the created task.
Inputs
titlestringrequired- Short title shown in the Tasks hub.
assignedToUserIdstring- Teammate user id (from list_team_members) to assign the task to; defaults to the caller. Must be the team owner or an active member of the caller's team, else TASK_ASSIGNEE_NOT_ALLOWED.
contactIdstring- CRM contact the task is about, if any.
dueAtstring- ISO 8601 due date-time; defaults to end of today in the caller's timezone.
instructionsstring- Free-text instructions or context shown on the task.
linkUrlstring- Optional http(s) link attached to the task.
prioritystring- Task priority.one of
lowmediumhighurgent taskTypestring- Kind of work; defaults to todo.one of
todocallemaillinkedin_connectionlinkedin_messagelinkedin_interactdefaulttodo timezonestring- IANA time zone used to default dueAt to end of day, e.g. America/New_York.
fuseaimcp_create_web_monitorCreate a Web Monitor: a recurring plain-language web search that surfaces PEOPLE or COMPANIES matching the described signal into a new dedicated results list.DestructiveCreate web monitor
Create a Web Monitor: a recurring plain-language web search that surfaces PEOPLE or COMPANIES matching the described signal into a new dedicated results list. `entityType` picks which ('person' or 'company') and cannot be changed later. `prompt` describes the signal in natural language and should mention the year for recency; `searchPeriod` sets how often it re-runs; `numResults` caps web pages read per run (not matches). For people, only those whose LinkedIn profile URL is found can be added. Unless initializeOnly is true, it sets up the search at an outside web-search provider and runs it at once, then re-runs it every searchPeriod until expirationDate, or until paused when none is set (there is no default end date); it pauses once it has added maxContacts results or as many as maxCreditSpend pays for in total. Each result newly added costs 50 credits, and a person 2 more when the user has no email or phone saved for them. A person who is already a contact has their saved profile fields (such as name, title and company) overwritten with the found profile, and each result can start the user's automations, which can add people to campaigns or act in connected apps.
Inputs
entityTypestringrequired- What the monitor should find: 'person' for people, 'company' for companies.one of
personcompany listNamestringrequired- No description.
promptstringrequired- Plain-language description of the signal to watch for on the open web (e.g. 'companies fined by a regulator over late supplier payments'). Including the year — 'in 2026' — materially improves recency.
expirationDatestring- No description.
initializeOnlyboolean- Omit or false to create and START the monitor immediately (sets up the live subscription and spends credits). Set true to save it as a DRAFT for the user to review and activate later.default
false maxContactsinteger- No description.
maxCreditSpendinteger- Maximum credits this agent may spend in a single run or refresh (1000 to 200000). CrustData monitors run once a day, so for them it is a daily limit; job-change watchers watch only as many contacts as one check within it pays for; web monitors stop once their total delivered results reach it. Omit for no limit.
numResultsinteger- Web pages to read per run (1-100, defaults to 10). This caps PAGES, not results — one page can yield several matches.
searchPeriodstring- How often the search re-runs — use one of exactly: 1h, 6h, 1d, 7d. Defaults to 1d.one of
1h6h1d7d
fuseaimcp_deactivate_campaign_lead_toolStop one lead's sequence inside a campaign, cancelling every step still pending for them.DestructiveStop Campaign Lead
Stop one lead's sequence inside a campaign, cancelling every step still pending for them. Takes a targetContactId from the campaign's lead list.
Inputs
targetContactIdstringrequired- No description.
fuseaimcp_deactivate_mailboxDESTRUCTIVE: Deactivate a purchased mailbox: its monthly credit charge stops and it is permanently deleted at the mailbox provider, at once if its paid month has already ended, otherwise automatically one day before its next monthly renewal.DestructiveDeactivate Mailbox
DESTRUCTIVE: Deactivate a purchased mailbox: its monthly credit charge stops and it is permanently deleted at the mailbox provider, at once if its paid month has already ended, otherwise automatically one day before its next monthly renewal. It is not disconnected from FuseAI, so campaigns can keep sending from it until it is deleted, and until then reactivate_mailbox can restore it. A deleted mailbox cannot be recovered. Identify it by domainName + email (both), or by the reseller mailboxId if the user supplies it from the app's Domain Management table. Only call after the user explicitly confirms.
Inputs
domainNamestring- No description.
emailstring- No description.
mailboxIdstring- No description.
fuseaimcp_deactivate_sending_domainDESTRUCTIVE: Deactivate a purchased sending domain by domainName together with all its mailboxes: their credit charges stop and they are permanently deleted at the provider.DestructiveDeactivate Sending Domain
DESTRUCTIVE: Deactivate a purchased sending domain by domainName together with all its mailboxes: their credit charges stop and they are permanently deleted at the provider. If the domain's paid term has already ended, its provider subscription is cancelled and the domain and every mailbox are deleted at once; otherwise each mailbox is deleted automatically one day before its next monthly renewal and the domain's whole workspace one day before the domain's renewal date. The mailboxes are not disconnected from FuseAI, so campaigns can keep sending from them until they are deleted; until then reactivate_sending_domain plus reactivate_mailbox for each mailbox can restore them. Deleted domains and mailboxes cannot be recovered. Only call after the user explicitly confirms.
Inputs
domainNamestringrequired- No description.
fuseaimcp_deactivate_twilio_phone_numberDESTRUCTIVE: Release/deactivate a power-dialer number — rented or verified caller-ID — (phoneNumberId = the number's _id from the rented-numbers list).DestructiveDeactivate Twilio Phone Number
DESTRUCTIVE: Release/deactivate a power-dialer number — rented or verified caller-ID — (phoneNumberId = the number's _id from the rented-numbers list). A rented number is first unlinked from the team's Twilio Trust Hub registrations, then released at Twilio and given up permanently. A verified caller-ID is deleted from Twilio and must be verified again before it can be used. Either way the number is archived in FuseAI and stops being used for calls, and removing a verified caller-ID also turns off inbound call forwarding to it. Only call after the user explicitly confirms.
Inputs
phoneNumberIdstringrequired- No description.
fuseaimcp_deep_research_startStarts an asynchronous Fuse Deep Research run: over several minutes it researches the request across the web in the background and produces a markdown report, plus — when the request is to find people or companies — a Fuse prospect list: a new one, or the earlier run's list it finds more for.DestructiveStart Deep Research
Starts an asynchronous Fuse Deep Research run: over several minutes it researches the request across the web in the background and produces a markdown report, plus — when the request is to find people or companies — a Fuse prospect list: a new one, or the earlier run's list it finds more for. Runs bill credits. ONLY call this when the user EXPLICITLY asks for "deep research", or confirms they want Deep Research after you offered it. For a general research/find-X request that does not say deep, do NOT call this — use lead_researcher_tool to build a list of people or companies from the web, or ask the user whether they want quick research (immediate answer) or Deep Research (background, 5-20 minutes, report + prospect list) and wait for their choice. Returns immediately with a runId; it can NEVER wait for or preview results in the same call. After starting, call deep_research_status with the runId to check progress (suggest the user checks back in a few minutes). Pass the user's full request as `query`, an optional short `listName` when a list is expected, and `effort` if specified. Without `targetCount` a run finds a first batch, not every match. To add more to the list an earlier run made, pass that run's runId as `continueRunId` and no listName. A run stops at a credit limit: 1000, 2000 or 5000 credits for low, medium or high effort, plus 40 per row of targetCount; each provider step it runs costs about 20 credits. Steps on the user's own key cost 5 each and sit outside that limit. Saving the run's list costs credits too: 1 per row matched to a known person or company. Saving also overwrites the stored profile details of contacts and companies already in FuseAI that its rows match, and each new list member can start the user's 'Added to list' automations. With continueRunId the rows go into the earlier run's list, so an active campaign with live list refresh that uses that list can enroll and message them.
Inputs
querystringrequired- The full deep-research request in the user's own words — the topic to research across the web.
continueRunIdstring- No description.
effortstring- No description.
idempotencyKeystring- Optional retry-safety key: calls with the same key return the SAME run instead of starting (and billing) a second one. Omit to always start a fresh run.
listNamestring- No description.
targetCountstring- No description.
fuseaimcp_delete_action_feed_itemDelete a Smart Actions feed item by id: sets its status to deleted, which removes it from the feed and cancels its linked task.DestructiveDelete Smart Actions Feed Item
Delete a Smart Actions feed item by id: sets its status to deleted, which removes it from the feed and cancels its linked task. The record is kept and can still be listed with status deleted. Destructive — confirm with the user before calling unless they already approved.
Inputs
actionFeedIdstringrequired- The Smart Actions feed card id (24-character Mongo id) — NOT the actionId.
fuseaimcp_delete_all_sequence_steps_toolDESTRUCTIVE: deletes ALL sequence steps from a campaign (clears its sequence) and permanently deletes their attachment files, which a duplicated campaign may still use.DestructiveDelete All Sequence Steps
DESTRUCTIVE: deletes ALL sequence steps from a campaign (clears its sequence) and permanently deletes their attachment files, which a duplicated campaign may still use. Refused once a campaign with a condition or A/B split has started. On a started campaign without them, contacts get no further steps, but messages already queued still go out, without those steps' attachments and, for emails, as replies in the contact's existing thread. Only call after the user has explicitly confirmed.
Inputs
campaignIdstringrequired- No description.
fuseaimcp_delete_automationDESTRUCTIVE: Delete an automation by automationId; it cannot be restored.DestructiveDelete Automation
DESTRUCTIVE: Delete an automation by automationId; it cannot be restored. No new runs start, its queued runs and runs waiting on a wait step are cancelled when they come due, any triggers it registered in the user's connected apps are turned off at Composio, and it disappears from the list. Only call after the user explicitly confirms the deletion. A teammate's rule may be deleted only by the team's owner or an admin (403 AUTOMATION_ACCESS_DENIED otherwise).
Inputs
automationIdstringrequired- No description.
fuseaimcp_delete_campaign_templates_toolDESTRUCTIVE: permanently delete saved campaign templates by id, including teammates' templates the user can see under the team's campaign visibility settings.DestructiveDelete Campaign Templates
DESTRUCTIVE: permanently delete saved campaign templates by id, including teammates' templates the user can see under the team's campaign visibility settings. Ids the user cannot see are skipped, so deletedCount can be lower than requestedCount. Only call after the user has explicitly confirmed which templates to delete.
Inputs
campaignTemplateIdsarrayrequired- No description.
fuseaimcp_delete_contact_noteDelete a contact note by noteId, removing it from the contact's timeline.DestructiveDelete Contact Note
Delete a contact note by noteId, removing it from the contact's timeline. Only the note's author may delete it.
Inputs
noteIdstringrequired- The note id from list_contact_notes or the contact timeline.
fuseaimcp_delete_email_tagDelete one of the user's custom FuseAI inbox tags by name and remove it from every email and conversation that carries it (re-creating the tag does not restore those assignments).DestructiveDelete Email Tag
Delete one of the user's custom FuseAI inbox tags by name and remove it from every email and conversation that carries it (re-creating the tag does not restore those assignments). Built-in tags cannot be deleted, and nothing changes in the connected Gmail, Outlook or SMTP mailbox. Removing it from campaign replies FuseAI has not processed yet can change how those replies are classified, and so whether the campaign keeps emailing the prospect or sends them an AI-written reply. Destructive — confirm with the user before using.
Inputs
tagNamestringrequired- No description.
fuseaimcp_delete_icpPermanently delete the user's Ideal Customer Profile (ICP) with this exact, case-sensitive name; it cannot be restored.DestructiveDelete ICP
Permanently delete the user's Ideal Customer Profile (ICP) with this exact, case-sensitive name; it cannot be restored. If more than one of the user's ICPs has exactly this name, all of them are deleted; normally only one does, because create_icp and update_icp refuse a name another of the user's ICPs already has (ignoring case). Automation ICP-match conditions that name it then treat every contact as not matching it, so a 'does not match ICP' condition on it passes every contact; Find contacts at company steps that use it fail; and CRM push conditions that name it never match again. Automation and CRM push conditions set to any of the user's ICPs, and the ICP match on lists, stop counting it at once; lead scores, which count only the team owner's ICPs, stop counting it when contacts are next rescored. Through those it can change which contacts the user's active automations act on (their steps can add contacts to campaigns that email and message them, post to Slack or act in the user's connected apps) and which new contacts FuseAI creates in the team's connected CRM, such as HubSpot or Salesforce. DESTRUCTIVE — only after the user explicitly confirms.
Inputs
namestringrequired- The exact name of the ICP to delete.
fuseaimcp_delete_list_toolDeletes an entire prospect list; only the list's creator can delete it, and the contacts and companies on it stay saved.DestructiveDelete List
Deletes an entire prospect list; only the list's creator can delete it, and the contacts and companies on it stay saved. Campaigns on the user's team that use the list as a source stop messaging its contacts, unless another active list in the same campaign also holds them. If the list excludes contacts from a campaign, they stop being excluded, and an active dynamic campaign can enroll and message them at its next refresh (within 30 minutes). Destructive and irreversible — only call after the user has explicitly confirmed.
Inputs
listIdstringrequired- No description.
fuseaimcp_delete_provider_keyDESTRUCTIVE: Permanently delete the user's saved API key for a Deep Research provider (by providerId).DestructiveDelete Provider Key
DESTRUCTIVE: Permanently delete the user's saved API key for a Deep Research provider (by providerId). From the next provider call, Deep Research runs (including ones in progress) and Fuse chat's data-provider calls switch to Fuse's platform key, billed at Fuse's per-call credit rate instead of the flat own-key rate, or stop working for byok-required providers, and a run or job already using the key may fail. This plugin cannot save the key again; the user has to re-enter it in Fuse. Only call after the user explicitly confirms.
Inputs
providerIdstringrequired- No description.
fuseaimcp_delete_saved_searchDelete a saved search.DestructiveDelete Saved Search
Delete a saved search. Destructive — only call after the user has explicitly confirmed.
Inputs
savedSearchIdstringrequired- The id of the saved search to delete (a 24-hex MongoDB id).
fuseaimcp_delete_sequence_step_toolDESTRUCTIVE: deletes a single sequence step by its id.DestructiveDelete Sequence Step
DESTRUCTIVE: deletes a single sequence step by its id. Deleting a condition step (an A/B split included) also deletes both of its lanes (a split's two arms) and every step in them. The deleted steps' attachment files are permanently deleted too, also for any other campaign whose steps use them: a copy made with duplicate_campaign_tool shares its original's files, so they are deleted for both the copy and the original. That other campaign's steps still list the deleted files, and its messages for those steps, queued or later, then fail to send. It also works on a running campaign whose sequence has no condition step (once a campaign has started, a sequence with a condition cannot lose any step): messages of this step that are already queued still go out, without its attachments and, for emails, as replies in the contact's existing thread; contacts waiting on this step's manual task (calls excepted) skip it, and their next step is written and queued. Writing that next step costs 10 credits when it is AI-written or AI-personalized, and if credits or a usable sender account are missing the campaign is paused instead, with an in-app notice (and a Slack message if Slack is connected). Only call after the user has explicitly confirmed removing that step.
Inputs
sequenceStepIdstringrequired- No description.
fuseaimcp_delete_signal_monitorPermanently delete a signal monitor and stop it at its data provider: its watches are paused or deleted there, and a running Google Maps search or recurring LinkedIn scrape is stopped.DestructiveDelete signal monitor
Permanently delete a signal monitor and stop it at its data provider: its watches are paused or deleted there, and a running Google Maps search or recurring LinkedIn scrape is stopped. Its results list and the records already delivered to it are kept. A monitor still being set up cannot be deleted, except a Google Maps search, whose run this stops. DESTRUCTIVE — only call after the user has explicitly confirmed the deletion.
Inputs
documentIdstringrequired- The signal monitor's documentId.
fuseaimcp_disconnect_all_linkedin_accountsDESTRUCTIVE: Disconnect every LinkedIn account the user owns at once (accounts teammates shared with the user are not touched).DestructiveDisconnect All LinkedIn Accounts
DESTRUCTIVE: Disconnect every LinkedIn account the user owns at once (accounts teammates shared with the user are not touched). Each is removed at the LinkedIn connection provider, its synced chats, connections and invitations are permanently deleted from FuseAI, and teammates it was shared with lose access; campaigns and inbox sync stop for every account, queued campaign LinkedIn steps from them fail, and a campaign left without a LinkedIn sender can be stopped. An account the provider does not confirm removing stays connected and is not counted in deletedAccountsCount. Reconnecting requires the interactive login in the FuseAI app. Only call after the user explicitly confirms they mean all accounts.
Inputs
This tool takes no inputs.
fuseaimcp_disconnect_email_accountDESTRUCTIVE: Disconnect an email sending account (by emailAddress).DestructiveDisconnect Email Account
DESTRUCTIVE: Disconnect an email sending account (by emailAddress). An account the user owns is deleted with its synced emails and conversations, its warmup (removed at the warmup provider) and its sharing with teammates; its inbox sync ends and campaigns stop sending from it (a running campaign left with no active sending account may stop). For an account shared with the user, only the user's access is removed. Only call after the user explicitly confirms. Reconnecting Gmail/Outlook later requires the OAuth flow in the FuseAI app.
Inputs
emailAddressstringrequired- No description.
fuseaimcp_disconnect_integrationDESTRUCTIVE: disconnect the calling user from an integration and remove their stored connection in FuseAI.DestructiveDisconnect Integration
DESTRUCTIVE: disconnect the calling user from an integration and remove their stored connection in FuseAI. HubSpot and Notion tokens are also revoked at the provider, FuseAI's webhook is deleted in Attio, and the FuseAI app is uninstalled from the GoHighLevel sub-account when no other connection uses it; Salesforce, Zoho and Pipedrive grants are not revoked at the provider. When no teammate stays connected to that CRM, the team's CRM sync settings are archived and its scheduled syncs stop. `service` ∈ hubspot|salesforce|zoho|pipedrive|attio|gohighlevel|notion|slack. Confirm with the user before calling. Slack: archives only the user's own connection, and nothing is revoked or uninstalled at Slack; the team owner's, which a teammate without one posts with, is the owner's to disconnect. The owner's disconnect also turns off FuseAI's Slack agent in every Slack workspace of the team, and teammates without their own connection can no longer post; anyone else's turns off the agent in workspaces they installed it in or are connected to, once no other teammate is connected there. Without a connection of their own, a member gets 403 INSUFFICIENT_PERMISSION, and an admin gets 400 MISSING_SLACK_CREDENTIALS unless an agent they installed is turned off.
Inputs
servicestringrequired- The integration to disconnectone of
hubspotsalesforceattiozohopipedrivegohighlevelnotionslack
fuseaimcp_disconnect_linkedin_accountDESTRUCTIVE: Disconnect one LinkedIn account (by accountId).DestructiveDisconnect LinkedIn Account
DESTRUCTIVE: Disconnect one LinkedIn account (by accountId). For an account the user owns: it is removed at the LinkedIn connection provider, its synced chats, connections and invitations are permanently deleted from FuseAI, teammates it was shared with lose access, campaigns stop using it and its inbox sync ends, its queued campaign LinkedIn steps fail, and a campaign left without a LinkedIn sender can be stopped; reconnecting requires the interactive login in the FuseAI app. For an account a teammate shared with the user, it only removes the user's own access (their queued campaign steps from it fail); the account stays connected for its owner. Only call after the user explicitly confirms.
Inputs
accountIdstringrequired- No description.
fuseaimcp_duplicate_campaign_toolDuplicates an existing campaign (creates a copy with its sequence).DestructiveDuplicate Campaign
Duplicates an existing campaign (creates a copy with its sequence). Pass the campaignId to clone. The copy is a new draft campaign named 'Copy of <original name>' (with a number added if the user already has a campaign by that name) and owned by the user, with fresh copies of all the original's sequence steps, the same lists and exclusions, and its AI prompt and company context, schedule and sending settings (its sender mailboxes, LinkedIn accounts and delegates only when the user owns the original), but none of its leads or sending history. The copy enrolls no one and sends nothing until it is started (update_campaign_status_tool). The call itself changes nothing in the original, but the copy's steps reuse the original's attachment files instead of getting copies of them, tying the two campaigns' attachments together: later deleting a step that holds those files, in either campaign, with delete_sequence_step_tool, delete_all_sequence_steps_tool or update_sequence_tool's delete_step action permanently deletes them for both, and the other campaign's messages for steps that list them then fail to send.
Inputs
campaignIdstringrequired- No description.
fuseaimcp_enrich_actionFind an email address for the contact behind an email send_message action that still needs one, by actionId — useful before send_action_email when the recipient email is missing.DestructiveEnrich Smart Action Contact
Find an email address for the contact behind an email send_message action that still needs one, by actionId — useful before send_action_email when the recipient email is missing. Uses an email already saved on the contact if there is one; otherwise looks the person up through external enrichment providers and saves what they find to the contact, which can overwrite its stored profile fields (such as name, job title, company, location and LinkedIn URL). Then writes the email into the action's recipient, replacing any recipient email already there. Works only on email send_message actions flagged as needing enrichment; returns ENRICHMENT_NO_EMAIL_FOUND when no email is found. Charges no credits, even when a provider finds the email. Sends nothing.
Inputs
actionIdstringrequired- The action id (24-character Mongo id) — NOT the actionFeedId. Feed items surface it as `actionId`.
fuseaimcp_enrich_contactLook up a person's professional email address, phone number, or both by querying a waterfall of data providers (Fuse, Apollo, Hunter, ContactOut, Datagma, and others).DestructiveEnrich Contact
Look up a person's professional email address, phone number, or both by querying a waterfall of data providers (Fuse, Apollo, Hunter, ContactOut, Datagma, and others). Needs a LinkedIn URL (input.url, highest overall match rate), or the person's name (input.firstName + input.lastName, or input.name) plus their company domain (input.domain, best for email matching). A company name alone (input.company) is used too: the company's domain is looked up from it for the email finders. Returns a JSON object with primaryEmail, primaryPhone, and arrays of all found emails and phones. It also saves the result to the user's contacts, which their team shares: to enrich a saved contact, pass its id as input.contactId; without it the person is saved as a contact even when no email or phone is found (a saved contact with the same LinkedIn profile is updated instead). The emails and phones found are merged into the contact, which can change a re-found address's status and the primary email and drop stored work emails at a past employer's domain, and when a provider returns the person's profile, the contact's name, LinkedIn URL, job title, company and location are overwritten with it. When the enriched contact is on a list that an active dynamic campaign uses, an email or LinkedIn URL it gains can get the person enrolled in that campaign at its next refresh (every 30 minutes) and then sent its outreach. Fields are null or empty arrays when no data is found. It costs credits when it finds a new deliverable email or valid phone number (values the given contact already holds are not charged) and nothing when it finds none. This is a slow operation (up to about 30 seconds) because it queries the providers one after another. Pass the person identifiers inside `input` (input.firstName, input.lastName, input.domain for the company domain, input.url for the LinkedIn URL, input.company for the company name) plus `enrichType` (email | phone | all). Optional inputs: input.customContactId is an older name for input.contactId (send one, not both); input.location is the person's location, used to help match them by name; extendedEnrich true keeps asking providers after the first one that finds a value and skips results cached from earlier lookups; resumeAfterApiUsed (enrichType email or phone, without extendedEnrich) takes the apiUsed of a returned email or phone and continues with the providers after it until one finds a new value; both are charged by the credit rule above. listId names the list the lookup was started from and only updates that list's last-activity time, without adding the person to the list; source, used only without input.contactId, records how the saved person was acquired as one of FuseAI's source labels (such as agents) and never replaces a label already set.
Inputs
enrichTypestringrequired- The type of enrichment to perform. Default to "email" — only pass "phone" or "all" when the user explicitly asked for phone numbers (phone enrichment costs several times more credits per contact).one of
emailphoneall inputobjectrequired- No description.
extendedEnrichboolean- No description.
listIdstring- No description.
resumeAfterApiUsedstring- No description.
sourcestring- No description.
fuseaimcp_exclude_contact_from_campaigns_toolStop a CRM contact's sequence inside one or more campaigns (contactId + campaignIds), cancelling every step still pending for them.DestructiveExclude Contact From Campaigns
Stop a CRM contact's sequence inside one or more campaigns (contactId + campaignIds), cancelling every step still pending for them. Use when you know the contact (e.g. from the CRM) but not their targetContactId in each campaign's lead list.
Inputs
campaignIdsarrayrequired- The campaigns to stop the contact's sequence in.
contactIdstringrequired- The CRM contact id to exclude (24-character Mongo id) — not a campaign targetContactId.
fuseaimcp_generate_ai_sequence_toolAI-generate a campaign's outreach sequence, for when the user wants the sequence written for them.DestructiveGenerate AI Sequence
AI-generate a campaign's outreach sequence, for when the user wants the sequence written for them. It deletes all of the campaign's current steps (hand-written content and attachments included) and replaces them with one new step per channel in the sequence list, with AI-written step topics and automatic wait days. The sequence list takes at most 10 channels, each one of email, linkedin_connection, linkedin_message, linkedin_visit, linkedin_like, dialer, inmail or todo; a longer list or any other channel name is refused and nothing changes, and an omitted or empty list gives three email steps. Each lead's message is then AI-written from its step's topic when that step is scheduled, and each AI-written message costs credits. It also runs on a campaign that is already sending: its leads continue on the new steps, and because every step is new, leads part-way through may be sent the new sequence from its first step, while messages already queued still go out as written. With a linked knowledge hub it may also replace the campaign's stored company description. Refused for a sequence with conditions or A/B splits once the campaign has started. It is the same operation as generate_campaign_sequence_tool, so use one or the other: each call replaces the steps again.
Inputs
campaignIdstringrequired- The campaign to generate an AI sequence for (its 24-hex MongoDB id).
sequencearray- Optional ordered list of step channels, e.g. ["email","linkedin_message","email"]. Omit to let the system choose.
fuseaimcp_generate_campaign_sequence_toolWrite a fresh AI-generated sequence for a campaign, deleting all of its current steps (hand-written content and attachments included) and replacing them with new steps that hold AI-written topics and automatic wait days.DestructiveGenerate Campaign Sequence
Write a fresh AI-generated sequence for a campaign, deleting all of its current steps (hand-written content and attachments included) and replacing them with new steps that hold AI-written topics and automatic wait days. Optionally takes the ordered channel list to generate for; without one it writes three email steps. Each contact's message is then AI-written from its step's topic when that step is scheduled, and each AI-written message costs credits. It also runs on a campaign that is already sending: its enrolled contacts continue on the new steps, and because every step is new, contacts part-way through may be sent the new sequence from its first step, while messages already queued still go out as written. When the campaign has a linked knowledge hub, it may also replace the campaign's stored company description with the hub's context. Refused with SEQUENCE_STRUCTURE_LOCKED for a sequence with conditions or A/B splits once the campaign has started. It is the same operation as generate_ai_sequence_tool, so use one or the other: each call replaces the steps again.
Inputs
campaignIdstringrequired- No description.
sequencearray- No description.
fuseaimcp_get_prospect_contact_activity_toolOne contact's outreach-activity timeline (emails, calls, LinkedIn touches, the team's notes, CRM activity, upcoming campaign steps) by CRM contactId — the id a contact event (campaign_reply, add_to_list, ...) carries.DestructiveProspect Contact Activity
One contact's outreach-activity timeline (emails, calls, LinkedIn touches, the team's notes, CRM activity, upcoming campaign steps) by CRM contactId — the id a contact event (campaign_reply, add_to_list, ...) carries. If your team has HubSpot or GoHighLevel connected, it may read the contact's activity directly from that CRM; for GoHighLevel, the first page also saves the contact's notes and tasks into FuseAI's activity history, replacing the copies earlier calls saved, and saved notes count toward the contact's lead score. Use to see everything that has happened with a single person, e.g. to recap a conversation before drafting a follow-up.
Inputs
contactIdstringrequired- The CRM contact id whose outreach activity to fetch.
cursorstring- Opaque `${occurredAtISO}_${activityId}` cursor from a previous page's nextCursor.
limitinteger- Page size (hub-served timeline only). Capped server-side.
fuseaimcp_get_user_infoRead the user + company record: company name/website, enrichment preferences, and subscription.DestructiveGet User Info
Read the user + company record: company name/website, enrichment preferences, and subscription. Normally it changes nothing, but if the account has no current plan it first starts the free plan: it saves a new subscription and, for a solo user or team owner, resets the credit balance, which forfeits any remaining expiring credits.
Inputs
This tool takes no inputs.
fuseaimcp_get_website_creditsThe team's website-intent (visitor de-anonymization) credits used this billing period and monthly credit cap (the team owner's, for every teammate); a null cap means no custom cap is set, so the plan's default cap applies once the team has a website-tracking setup.DestructiveGet Website Credits
The team's website-intent (visitor de-anonymization) credits used this billing period and monthly credit cap (the team owner's, for every teammate); a null cap means no custom cap is set, so the plan's default cap applies once the team has a website-tracking setup. For a team owner, or a user on no team, who has no website-tracking setup yet, the call first creates an empty one, which registers the account for FuseAI's background visitor identification and website-intent credit charges: visitors to the team's website are then looked up hourly with Fiber, an outside people-data provider, and the people identified are added to the CRM's website-visitor lists, where each new person can trigger a Slack visitor alert and website-visitor automations. Confirm with the user before calling it when the team may have no website-tracking setup yet (get_website_allowed_domains returns null in that case and changes nothing).
Inputs
This tool takes no inputs.
fuseaimcp_import_companies_to_listSave companies you already have (read off a web page you scraped, from text the user pasted, or from your search results) into a company list: a new one (newListTitle) or an existing company list (existingListId).DestructiveImport Companies To List
Save companies you already have (read off a web page you scraped, from text the user pasted, or from your search results) into a company list: a new one (newListTitle) or an existing company list (existingListId). mode DEMOGRAPHICS enriches them in the background, first reading the website of a company that came without one off its page on the source (Profile URL); it fills in company details, never emails or phone numbers, and costs 1 credit per company that matches and nothing for the rest, so call it only once the user has agreed to that cost. mode RAW saves them without enrichment, free; a domain that is not the company's own website is left empty in either mode. Give each company's LinkedIn page (Company LinkedIn URL) whenever the source shows it: it is kept on the company (one matched to a known company keeps that company's page), and a company watcher uses that page when the company has no website. Answers at once with the list (listId, listName) and a rough estimatedMinutes: the list fills in the background and the user gets a notification when it is complete. Each company newly added to the list starts the user's automations that run when a company is added to a list. More companies than one call takes go in further calls, with existingListId set to the listId the first call answered. A value with no field of its own (a deal stage, a rival, a score) goes in the row's customColumns as a name/value pair: each name becomes a column on the list, or fills the list's own column of that name when every value fits it (a number, a date, a yes/no, or one of its options) and it is neither a smart nor a link column; an existing list takes columns only when the user created it. Never put such a value in another field. A company already on the list is saved again. When the row matches a known company by its website, the company keeps its row and its other columns, and the columns the call fills take the row's values (an empty value clears one); otherwise the row replaces it, and its other columns, smart ones included, are lost. Mode DEMOGRAPHICS matches and charges it again.
Inputs
modestringrequired- DEMOGRAPHICS matches each company to a known company and fills in its profile, charging 1 credit for every company it matches. RAW saves the companies without enrichment, free. In either mode a domain that is not the company's own website is left empty.one of
RAWDEMOGRAPHICS rowsarrayrequired- One object per company, at most 1000 per call, and at most 20 different column names across them.
existingListIdstring- Add to this existing list, by its 24-hex id. It must be a list of the same type. Give exactly one of newListTitle and existingListId.
listTypestring- Always 'companyList'.one of
companyListdefaultcompanyList newListTitlestring- Create a new list with this name. Give exactly one of newListTitle and existingListId.
sourceUrlstring- The page the rows were scraped from, when they were. It keeps that site's own links out of the websites the lookup finds.
fuseaimcp_import_people_to_listImport rows of people from a file or CSV the user gave you, or from a web page you scraped (a team, speaker or directory page), into a people list: a new one (newListTitle) or an existing people list (existingListId).DestructiveImport People To List
Import rows of people from a file or CSV the user gave you, or from a web page you scraped (a team, speaker or directory page), into a people list: a new one (newListTitle) or an existing people list (existingListId). People named in the chat or found in a search are saved with save_people_to_list, not added here as new rows: RAW would keep them as custom contacts with no database profile. Once they are on the list, this tool adds column values to them only with mode RAW and onDuplicate merge; with mode RAW the default, replace, takes them off the list. mode DEMOGRAPHICS matches each row to a known profile and fills it in, without finding emails or phone numbers; it costs 1 credit per person who matches and nothing for the rest, and that price never covers an email or a phone. EMAIL, PHONE and EMAIL_AND_PHONE match each row the same way for free, then find those contact details, charged per detail found: a work email 20 credits when Fuse's own data finds it or 50 when a partner provider does (a personal email 100 when the account puts personal emails first), a phone number 100, or 5 instead when a provider key saved on the account found it; nothing when nothing is found. Call a paid mode only once the user has agreed to that cost. mode RAW keeps each row exactly as given, free, as a custom contact with no database profile behind it. Give each person's LinkedIn URL whenever you know it: matching tries it first. Answers at once with the list (listId, listName) and a rough estimatedMinutes: the list fills in the background and the user gets a notification when it is complete. Each person newly added to the list starts the user's automations that run when someone is added to a list, and people added to a list that an active dynamic campaign uses can be enrolled in it at its next refresh (every 30 minutes) and start receiving its outreach. More rows than one call takes go in further calls, with existingListId set to the listId the first call answered. A value with no field of its own (a rival, a deal stage, a score) goes in the row's customColumns as a name/value pair: each name becomes a column on the list, or fills the list's own column of that name when every value fits it (a number, a date, a yes/no, or one of its options) and it is neither a smart nor a link column; an existing list takes columns only when the user created it. Never put such a value in another field. In every mode, a row matching a person already on the list replaces them by default: the row is saved again (with mode RAW as a new contact; a paid mode matches it again, and DEMOGRAPHICS charges for it again) and the person is taken off the list unless the row saves onto that same contact; a contact mode RAW takes off can also be archived when it is on no other list. onDuplicate skip leaves them as they are and drops the row. With mode RAW only, onDuplicate merge keeps them instead: the row fills their empty fields and adds its emails and phones, and its column values replace theirs, free.
Inputs
modestringrequired- DEMOGRAPHICS matches each person to a known profile and fills it in, charging 1 credit for every person it matches. EMAIL, PHONE or EMAIL_AND_PHONE match each person for free, then find those contact details. RAW keeps each row exactly as given, free, as a custom contact with no database profile.one of
RAWDEMOGRAPHICSEMAILPHONEEMAIL_AND_PHONE rowsarrayrequired- One object per person, at most 1000 per call, and at most 20 different column names across them.
existingListIdstring- Add to this existing list, by its 24-hex id. It must be a list of the same type. Give exactly one of newListTitle and existingListId.
listTypestring- Always 'contactList'.one of
contactListdefaultcontactList newListTitlestring- Create a new list with this name. Give exactly one of newListTitle and existingListId.
onDuplicatestring- What a row that matches someone already on the list does (matched by LinkedIn URL, then email, then name plus company domain). replace (the default): the row is saved as a new contact and the person on the list is taken off it. merge, with mode RAW only: the person stays on the list, the row fills their empty fields and adds its emails and phones, and its column values replace theirs. skip: the person is left as they are and the row is dropped.one of
mergeskipreplace
fuseaimcp_initialize_campaign_toolCreates a DRAFT outreach campaign.DestructiveInitialize Campaign
Creates a DRAFT outreach campaign. For ai-generated campaigns it starts a background job in which an AI model analyzes the optional userPrompt (it may search the public web) and generates topics and sequence steps; a step order named in the prompt replaces stepChannels, and when templateId is given the template's steps are used instead. Manual campaigns get empty starter steps (or the template's) and no background job. The new campaign is initialized, NOT active — it sends nothing, and creating it charges no credits. Each call also saves this campaign's timezone, sending days and hours (and ctaLink, if given) as the user's remembered campaign defaults, replacing the previous ones. Activating it is a separate, explicit act this tool never performs: only activate when the user explicitly asks, and confirm first. Returns 202 with the new campaignId and initializationStatus ('pending' for ai-generated campaigns, 'completed' for manual ones). Required: campaignName, campaignType (ai-generated | manual), stepChannels, campaignStartDate (ISO), timezone (IANA, e.g. America/New_York), emailDays (Monday...Sunday), excludeEngagedContacts (true skips contacts who have already replied to any of the user's campaigns; out-of-office auto-replies do not count), isDynamic (true: while the campaign is active, contacts later added to its source lists are enrolled automatically and start the sequence; false: later additions are not enrolled), listIds (at least one CRM prospect list id — a campaign cannot be created without a source list; add_list_to_campaign_tool only adds further lists later). Optional calendar link: ctaLink (http(s) booking URL), only when the user gave one; a campaign gets no link unless one is passed and does not need one. Optional schedule: emailStartHour / emailEndHour (24-hour H:MM; if omitted, the user's remembered hours apply, else 09:00–17:00). The campaign also inherits the user's remembered auto-pilot setting: when it is on, follow-ups to contacts who replied are sent without waiting for approval once the campaign is active. Optional AI context: writingStyleId (ai-generated campaigns only; one of the user's styles from list_writing_styles, else the user's default style applies), vectorStoreId (a Knowledge Hub profile's vectorStoreId from list_knowledge_hub_profiles, whose company knowledge the AI draws on when generating the campaign's topics; omitted, the team's Knowledge Hub is used) and useLinkedinPosts (default false; when true, each contact's public LinkedIn posts are fetched once the campaign is active and AI-written messages draw on them). Optional audience filters: excludeLists (CRM prospect list ids whose contacts are not enrolled), excludeCompanyLists (up to 20 company or people list ids; contacts at those companies, or at the companies of those people, are not enrolled), excludeContacted (default none, which skips nobody; ever, last_1_month, last_3_months, last_6_months or last_1_year skips contacts already contacted ever or within that window; true means ever) and sendOnlyToValidatedEmailsEnabledAt (an ISO 8601 timestamp turns it on, null leaves it off: when the sequence has an email step, only contacts whose email is already verified valid are enrolled, so catch-all and unverified addresses are skipped). Optional launch-time email finding: findMissingEmailsEnabledAt (an ISO 8601 timestamp turns it on, null leaves it off): when the campaign is launched and its sequence has an email step, work emails are first looked up for contacts on its lists never searched for one; each email found is saved to its contact and charged in credits to the campaign owner. Optional send-time check: validateEmailsBeforeSendEnabledAt (an ISO 8601 timestamp turns it on, null leaves it off): once the campaign is active, each address is verified just before its email is sent and each address verified is charged in credits; an email whose address fails is skipped, that contact is stopped and an invalid address is marked invalid, and the campaign is deactivated when credits run out. Optional tracking: addUnsubscribeLink (default true), openRateTrackingEnabledAt / clickTrackingEnabledAt (default off; pass true to enable), trackingDomainId (from list_tracking_domains_tool; omit/null uses the team's default ACTIVE domain at send time). Click tracking must be on for a custom tracking domain to rewrite email links. Optional daily caps: dailyEmailLimit / dailyLinkedinConnectionLimit / dailyLinkedinMessageLimit (sends per calendar day in the campaign's timezone across all its accounts; omit/null = no campaign-level cap, account limits always apply). Optional co-owners: delegates (user ids of teammates from list_team_members who co-own the campaign and can edit it; the first one owns its dialer (call) tasks).
Inputs
campaignNamestringrequired- No description.
campaignStartDatestringrequired- No description.
campaignTypestringrequired- No description.one of
ai-generatedmanual emailDaysarrayrequired- Weekdays on which the campaign may send. English names with a capital first letter: Monday ... Sunday. At least one day required.
excludeEngagedContactsbooleanrequired- No description.
isDynamicbooleanrequired- No description.
stepChannelsarrayrequired- No description.
timezonestringrequired- IANA time zone in which sending days and the daily window are evaluated, e.g. America/New_York. Invalid zones are rejected (the scheduler would otherwise silently fall back to UTC).
addUnsubscribeLinkboolean- When true, every email step appends a one-click unsubscribe footer (email channel only). Default true.default
true clickTrackingEnabledAtstring- Rewrite links in outgoing emails to tracked redirects. Must be on for a custom tracking domain to appear in the email; otherwise links stay as the original URL. Pass true to turn on, false or null to leave off (default off). An ISO 8601 timestamp is also accepted.
ctaLinkstring- The calendar/booking link (absolute http(s) URL) recipients follow to book a meeting. Optional: omitted or null, the campaign has no link.
dailyEmailLimitinteger- Maximum emails this campaign sends per calendar day in its timezone, across all its mailboxes and excluding replies. Mailbox limits and sharing with other campaigns still apply. Null (default) means no campaign-level cap.
dailyLinkedinConnectionLimitinteger- Maximum LinkedIn connection requests this campaign sends per calendar day in its timezone, across all its LinkedIn accounts. Account limits still apply. Null (default) means no campaign-level cap.
dailyLinkedinMessageLimitinteger- Maximum LinkedIn messages (InMails included) this campaign sends per calendar day in its timezone, across all its LinkedIn accounts. Account limits still apply. Null (default) means no campaign-level cap.
delegatesarray- User ids of teammates who co-own the campaign, from list_team_members. The FIRST id owns the campaign's dialer (call) tasks: the campaign appears on that person's dialer page and its call batch is created for them instead of the creator. Every id must be the team owner or an active member of the user's team.
emailEndHourstring- Local end of the daily sending window as 24-hour H:MM or HH:MM (e.g. 17:00). Default 17:00 if omitted.
emailStartHourstring- Local start of the daily sending window as 24-hour H:MM or HH:MM (e.g. 09:00). Default 09:00 if omitted.
excludeCompanyListsarray- No description.
excludeContactedstring- No description.default
none excludeListsarray- No description.
findMissingEmailsEnabledAtstring- Find missing work emails at launch: an ISO 8601 timestamp turns it on, null (default) leaves it off. When on and the sequence has an email step, contacts in the campaign's lists never searched for an email are enriched before enrollment, charged to the campaign owner (only emails found are charged).
listIdstring- Deprecated single-list form. Prefer listIds. If sent alone it is promoted to listIds[0]. One of listId or listIds is required.
listIdsarray- CRM prospect list ids to source contacts from (1-10). Required at create (or send legacy listId). To add another list after create, use add_list_to_campaign_tool.
openRateTrackingEnabledAtstring- Enable the open-tracking pixel on outgoing emails. Pass true to turn on, false or null to leave off (default off). An ISO 8601 timestamp is also accepted.
sendOnlyToValidatedEmailsEnabledAtstring- No description.
templateIdstring- No description.
trackingDomainIdstring- Id of an ACTIVE team tracking domain (from list_tracking_domains_tool). Null (default) does not mean the Fuse origin: click-tracked links use the team's default ACTIVE domain automatically if one exists. Click tracking must also be enabled or links are not rewritten.
useLinkedinPostsboolean- No description.default
false userPromptstring- No description.
validateEmailsBeforeSendEnabledAtstring- No description.
vectorStoreIdstring- No description.
writingStyleIdstring- No description.
fuseaimcp_invite_team_memberInvite one or more people to the team, each with a role (ADMIN or MEMBER, default MEMBER).DestructiveInvite Team Member
Invite one or more people to the team, each with a role (ADMIN or MEMBER, default MEMBER). Owner or admin only. Creates a pending invitation for each person and immediately emails each one an invitation from FuseAI, which cannot be recalled. On plans with a seat limit each pending invitation takes a seat, and the whole batch is refused if there are not enough free seats; it is also refused if anyone in it is already on the team or already invited. If an email fails to send, the call returns an error but the invitations stay pending.
Inputs
membersarrayrequired- The people to invite
fuseaimcp_launch_campaign_toolLaunch a campaign that has not been launched yet (or whose launch failed): the campaign is set active, the contacts on its lists are enrolled, and its sequence starts sending real emails and LinkedIn actions to them from the user's connected accounts on the campaign's schedule.DestructiveLaunch Campaign
Launch a campaign that has not been launched yet (or whose launch failed): the campaign is set active, the contacts on its lists are enrolled, and its sequence starts sending real emails and LinkedIn actions to them from the user's connected accounts on the campaign's schedule. Only call it when the user explicitly asks to launch, after confirming with them. Enrolment runs in the background, so this returns as soon as the work is queued. AI writing in the campaign's messages is charged in credits as each message is written; when the campaign's find-missing-emails setting is on, work emails are first looked up for list contacts never searched for one, also charged in credits.
Inputs
campaignIdstringrequired- No description.
fuseaimcp_launch_draft_campaign_toolLaunches the draft campaign given by campaignId (one not launched yet, or whose launch failed): validates readiness, enrolls the attached lists' contacts as its targets, and flips the campaign to active.DestructiveLaunch Draft Campaign
Launches the draft campaign given by campaignId (one not launched yet, or whose launch failed): validates readiness, enrolls the attached lists' contacts as its targets, and flips the campaign to active. Outreach then begins on its schedule: the sequence's emails and LinkedIn actions go out to the enrolled contacts from the user's connected mailboxes and LinkedIn accounts. Enrolment runs in the background, so this returns as soon as the work is queued. A campaign that fails the readiness checks, or is already active, paused or stopped, is refused and left unchanged. It is the same launch as launch_campaign_tool, so use one or the other. This is a launch action, not a data-prep step: never call it while creating or editing a campaign; only on an explicit user request to launch, and confirm with the user first. Launching spends credits: each AI-written message and AI personalization is charged as it is written, and when the campaign's find-missing-emails setting is on, work emails are first looked up from external data providers for list contacts never searched for one and saved to the lists, charged to the owner's credits.
Inputs
campaignIdstringrequired- The campaign to (re)generate target contacts for (its 24-hex MongoDB id).
fuseaimcp_lead_researcher_toolStarts a background web search for people or companies matching the query and saves what it finds, over several minutes, to a FuseAI list: a new list named listName (with a number added when the user already has a list of that name), or the existing list existingListId, which shows as processing until the search finishes.DestructiveLead Research
Starts a background web search for people or companies matching the query and saves what it finds, over several minutes, to a FuseAI list: a new list named listName (with a number added when the user already has a list of that name), or the existing list existingListId, which shows as processing until the search finishes. With existingListId, that list keeps its name; listName is used only if the search returns a different kind of result than that list holds, and those results then go to a new list of that name. Saving each person costs 2 credits unless the user already has an email or phone saved for them, and also overwrites the stored profile fields of a matching contact already in FuseAI. Each result gets a match/miss value per search criterion in the list's columns; on an existing list, columns with the same name are reused and their values replaced, including for people or companies already on it. New list members can start the user's 'Added to list' and 'Company added to list' automations, whose steps can send messages, add people to campaigns or act in connected apps. Any active campaign with live list refresh that uses the list can also enroll the people added when it refreshes (every 30 minutes) and send them its emails, LinkedIn messages and connection requests, which cannot be undone. Each call starts a new search, so repeating a call (for example after a timeout) repeats these effects: it creates another list when existingListId is omitted, can charge credits again and can start automations for the people it adds. Before repeating a call that may already have started, check the user's lists with list_prospect_lists_tool.
Inputs
listNamestringrequired- The name of the list to save the results to.
querystringrequired- The query to search the web for.
resultTypestringrequired- What the user asked to find: people ('person') or companies ('company'). The search looks for exactly this.one of
personcompany countnumber- The number of results to find. Defaults to 10. Max 500.default
10 existingListIdstring- Id of an EXISTING list to add the results to, when the user asked to add to a list they named (find its id by name first). Omit to build a new list named listName. The list must hold what resultType asks for: people go into contact lists, companies into company lists.
fuseaimcp_list_writing_stylesList the user's writing styles (id, name, rules, isDefault, extractionStatus, extractionError, learning settings, modelEffort).DestructiveList Writing Styles
List the user's writing styles (id, name, rules, isDefault, extractionStatus, extractionError, learning settings, modelEffort). The default style is applied when a campaign has none selected: those campaigns write their AI messages and AI replies to contacts with its rules. If the user has no default style yet, this call first creates it with generic rules and style learning switched on at the 'fast' effort (modelEffort null), which charges no credits. A daily background job then starts adding rules learned from the user's sent emails and LinkedIn messages stored in FuseAI (newer learned rules can replace earlier ones), and smart columns with engine writer start applying its rules to the messages they write. The call itself never changes or removes a style that already exists.
Inputs
This tool takes no inputs.
fuseaimcp_merge_duplicate_contact_groupsMerge the duplicate groups listed in selectedGroupIds (their groupId values) from a completed duplicate scan or review (scanJobId).DestructiveMerge Duplicate Contact Groups
Merge the duplicate groups listed in selectedGroupIds (their groupId values) from a completed duplicate scan or review (scanJobId). Each group keeps one contact (a shared profile over a team-only record, then the most recently created): its empty fields are filled from the others, all emails and phones are combined onto it, the others' list memberships, campaign enrollments, CRM links, owner assignments, tasks and activity move to it (where it already has the same one, the other's copy is deleted; an enrollment in a campaign it is already in is instead stopped along with its pending sends), and the other contacts are archived. Moved enrollments continue their sequences using the kept contact's details, and a campaign that enrolls new members of its lists may start contacting it. Runs in the background and answers { data: { jobId } }; that job records how many groups were merged and contacts archived. Contacts are shared, so a merge changes them for everyone who has them, and it cannot be undone: only merge groups the user has seen and approved.
Inputs
scanJobIdstringrequired- The jobId of a completed duplicate scan or duplicate review whose groups to merge
selectedGroupIdsarrayrequired- The groups to merge, by their groupId
fuseaimcp_modify_scheduled_replyEdit a pending auto-reply's body/subject and set when it goes out: scheduledDateUTC is required and replaces the reply's send time.DestructiveModify Pending Auto-Reply
Edit a pending auto-reply's body/subject and set when it goes out: scheduledDateUTC is required and replaces the reply's send time. The reply goes to its contact from the campaign's connected mailbox or LinkedIn account, and keeps its approval state: one awaiting approval still needs approving, while one already approved goes out with the new text at the new time without being approved again. Returns a NEW scheduledEmailId that replaces the old one — the old id becomes invalid.
Inputs
bodystringrequired- The new message body.
scheduledDateUTCstringrequired- When to (re)schedule the reply, as an ISO UTC datetime.
scheduledEmailIdstringrequired- The pending reply's scheduledEmailId.
subjectstring- New subject line (email channel only).
fuseaimcp_pause_email_warmupPause managed deliverability / email warmup for a sending account.DestructivePause Email Warmup
Pause managed deliverability / email warmup for a sending account. While warmup runs, the external email warmup service Mailivery sends warmup emails and replies from the mailbox to build its sending reputation. This call pauses the mailbox's warmup at Mailivery, which stops those emails until warmup is started again with start_email_warmup, and updates FuseAI's stored copy of the warmup. `id` is the account's managedDeliverabilityCampaignId from the email-accounts list, passed AS A STRING — the field is numeric, so quote it (e.g. "18234"; an unquoted number is rejected).
Inputs
idstringrequired- No description.
fuseaimcp_preview_campaign_sequence_toolPreview how a campaign's sequence step renders for sample prospects on its lists (variables filled in).DestructivePreview Campaign Sequence
Preview how a campaign's sequence step renders for sample prospects on its lists (variables filled in). The previews are only returned: it sends nothing, stores no messages, changes no campaign, contact or list, and charges no credits, with one exception: when it previews an AI-written step and the campaign's owner has no writing style yet, it first creates that owner's default writing style with style learning switched on, and a daily job then keeps adding and replacing that style's rules, which every later AI-written campaign message and reply follows. To write the preview it may look up the sample prospects' company websites and public LinkedIn posts and search the web. sequenceNumber is the step's sequenceNumber from get_campaign_sequence_steps_tool; samples is how many prospects to preview (1-10, default 3). Optional sequenceStepType renders the step as ai_generated (an AI-written message), ai_reply (an AI-written reply) or manual (the step's saved template with variables filled in) instead of its saved type; the saved step is not changed. Optional fromEmail, for email steps: the address of one of the user's connected mailboxes whose saved sender name, title and signature fill the preview's sender variables; nothing is sent from it. Useful to show the user the copy before activating; AI-written steps are written afresh for each contact when their message is queued, so the sent wording will differ.
Inputs
campaignIdstringrequired- No description.
sequenceNumberintegerrequired- No description.
fromEmailstring- No description.
samplesinteger- No description.default
3 sequenceStepTypestring- No description.one of
ai_generatedai_replymanual
fuseaimcp_pull_crm_activitiesStart a background pull of up to 6 months of activities (calls, emails, meetings, notes, tasks) from the connected CRM into the FuseAI activities hub (async; nothing is written to the CRM).DestructivePull CRM Activities
Start a background pull of up to 6 months of activities (calls, emails, meetings, notes, tasks) from the connected CRM into the FuseAI activities hub (async; nothing is written to the CRM). It adds new activities, overwrites ones stored earlier with the CRM's values, and deletes stored copies of FuseAI's own pushed activities and, for GoHighLevel, activities the CRM withdrew. Campaigns that skip already-contacted contacts count pulled calls, emails and meetings, so a pull can change who they enroll. Not available for Notion: crmIdentifier "notion" returns the error ACTIVITY_PULL_NOT_SUPPORTED and starts no pull.
Inputs
crmIdentifierstringrequired- The connected CRMone of
hubspotsalesforceattiozohopipedrivegohighlevelnotion
fuseaimcp_pull_crm_companiesStart a background pull of companies from the connected CRM into FuseAI (async; nothing is written to the CRM).DestructivePull CRM Companies
Start a background pull of companies from the connected CRM into FuseAI (async; nothing is written to the CRM). It creates FuseAI companies for new CRM records and overwrites the team's own company records that are linked to a CRM company or share its domain with the CRM's name, domain, description, industry, LinkedIn URL, revenue, headcount and address; FuseAI's shared company catalog is never changed.
Inputs
crmIdentifierstringrequired- The connected CRMone of
hubspotsalesforceattiozohopipedrivegohighlevelnotion
fuseaimcp_pull_crm_contactsStart a background pull of contacts from the connected CRM into FuseAI (async; nothing is written to the CRM).DestructivePull CRM Contacts
Start a background pull of contacts from the connected CRM into FuseAI (async; nothing is written to the CRM). It creates FuseAI contacts and companies for new CRM records, brings back contacts a team member deleted in FuseAI while the CRM still holds them, and overwrites stored data with the CRM's values: the CRM owner and CRM property values, each pulled contact's primary email and phone as stored for the team member whose CRM connection the pull uses (stored work emails at the contact's past employers may be dropped), the name, job title, company and address of an existing contact matched by LinkedIn URL, and the name, revenue and industry of the team's own matching companies.
Inputs
crmIdentifierstringrequired- The connected CRMone of
hubspotsalesforceattiozohopipedrivegohighlevelnotion
fuseaimcp_push_crm_companiesHIGH-IMPACT: push companies FROM FuseAI INTO the customer's CRM (async).DestructivePush CRM Companies
HIGH-IMPACT: push companies FROM FuseAI INTO the customer's CRM (async). Confirm with the user first. Empty selection = push all enriched companies (fills empty CRM fields); or scope by companyIds / customCompanyIds (overwrites existing CRM values). overrideExisting: true makes any push overwrite existing CRM values; false makes it fill empty fields instead. listIds is not supported yet, so leave it out: sent without ids it is refused (COMPANY_LIST_PUSH_UNSUPPORTED), and sent with ids it is ignored.
Inputs
crmIdentifierstringrequired- The connected CRMone of
hubspotsalesforceattiozohopipedrivegohighlevelnotion companyIdsarray- At most 5000 ids across companyIds and customCompanyIds
customCompanyIdsarray- At most 5000 ids across companyIds and customCompanyIds
listIdsarray- No description.
overrideExistingboolean- No description.
fuseaimcp_push_crm_contactsHIGH-IMPACT: push contacts FROM FuseAI INTO the customer's CRM (async).DestructivePush CRM Contacts
HIGH-IMPACT: push contacts FROM FuseAI INTO the customer's CRM (async). Confirm with the user first. Empty selection = push all enriched contacts (fills empty CRM fields only); or scope by contactIds / customContactIds (overwrites existing CRM values, including the CRM owner where one is mapped) / listIds (a list id or "all_enriched_contacts"; fills empty CRM fields only). Send ids and listIds in separate calls: sent together, the whole push overwrites existing CRM values (every enriched contact's when "all_enriched_contacts" is included), and next to real lists alone the ids are not pushed at all. Push all and "all_enriched_contacts" create only contacts that match the team's contactPushConditions (see get_crm_settings); contactIds and the members of real lists are pushed regardless, also when sent alongside "all_enriched_contacts". Pushed contacts stay linked to their CRM records, so later scheduled syncs keep writing to those records even when they match no contactPushConditions: contact auto push (every 6 hours) and activity sync (hourly, only while auto push is on) when the team has turned them on, and the hourly Fuse property sync when the team has Fuse property mappings.
Inputs
crmIdentifierstringrequired- The connected CRMone of
hubspotsalesforceattiozohopipedrivegohighlevelnotion contactIdsarray- No description.
customContactIdsarray- No description.
listIdsarray- No description.
fuseaimcp_push_fuse_propertiesStart a sync of the team's mapped Fuse properties (e.g.DestructivePush Fuse Properties
Start a sync of the team's mapped Fuse properties (e.g. LinkedIn connection status) to the connected CRM now instead of at the next hourly run (async). Writes only to contact records that already exist in the CRM, overwriting the mapped properties there and clearing a value FuseAI wrote earlier when the status moves to one the team left unmapped. HubSpot only for now: any crmIdentifier other than "hubspot" is refused (422 FUSE_PROPERTY_SYNC_NOT_SUPPORTED). Owner or admin only.
Inputs
crmIdentifierstringrequired- The connected CRMone of
hubspotsalesforceattiozohopipedrivegohighlevelnotion
fuseaimcp_reactivate_mailboxReactivate a purchased mailbox that was deactivated but not yet deleted at the provider: this cancels its scheduled deletion at the mailbox provider and resumes its monthly renewal charge in credits.DestructiveReactivate Mailbox
Reactivate a purchased mailbox that was deactivated but not yet deleted at the provider: this cancels its scheduled deletion at the mailbox provider and resumes its monthly renewal charge in credits. The mailbox is kept there instead of being deleted by FuseAI's daily job one day before its next monthly renewal, so campaigns sending from it can keep sending to prospects after that date. Nothing is sent to the mailbox provider when this tool runs. A deleted mailbox cannot be reactivated, and if its domain was deactivated too, also reactivate the domain with reactivate_sending_domain or the mailbox is still deleted with it. Identify it by domainName + email (both), or by the reseller mailboxId if the user supplies it from the app's Domain Management table.
Inputs
domainNamestring- No description.
emailstring- No description.
mailboxIdstring- No description.
fuseaimcp_reactivate_sending_domainReactivate a purchased sending domain by domainName that was deactivated but not yet deleted at the provider: this cancels its scheduled deletion at the domain provider and resumes its renewal charge in credits.DestructiveReactivate Sending Domain
Reactivate a purchased sending domain by domainName that was deactivated but not yet deleted at the provider: this cancels its scheduled deletion at the domain provider and resumes its renewal charge in credits. FuseAI's daily job then no longer cancels the domain's provider subscription one day before its renewal date, which would delete the domain and its whole workspace there. Nothing is sent to the domain provider when this tool runs. Mailboxes deactivated with it stay deactivated and are still deleted unless reactivate_mailbox is called for each.
Inputs
domainNamestringrequired- No description.
fuseaimcp_regenerate_queued_campaign_messages_toolRewrite, in a background job, every message a campaign has queued but not yet sent (emails, LinkedIn messages, InMails and connection-request notes) from its current sequence steps, replacing their wording, including edits the user made to them.DestructiveRegenerate Queued Campaign Messages
Rewrite, in a background job, every message a campaign has queued but not yet sent (emails, LinkedIn messages, InMails and connection-request notes) from its current sequence steps, replacing their wording, including edits the user made to them. It does not change the sequence steps or messages already sent, and does nothing for a campaign with no queued messages (such as a draft). AI writing a message has not yet paid for is charged in credits; if the balance cannot cover it, that message is stopped and the campaign is paused, and the user is notified in the app and on Slack if connected.
Inputs
campaignIdstringrequired- No description.
fuseaimcp_regenerate_step_topics_toolRe-generates the per-step topics of an AI campaign's EXISTING sequence steps from the campaign's current prompt/focus.DestructiveRegenerate Step Topics
Re-generates the per-step topics of an AI campaign's EXISTING sequence steps from the campaign's current prompt/focus. Keeps every step's channel, order, wait days, templates, and attachments — the titles/topics of its AI-generated steps are overwritten, and when the campaign has a focus and a linked knowledge hub, its stored company description may also be replaced with the context the topics were written from. A lighter alternative to deleting and rebuilding the sequence. It also works on a running campaign: messages not yet written for a contact are written from the new topics, while messages already queued keep their text.
Inputs
campaignIdstringrequired- No description.
fuseaimcp_remove_domain_forwardingDESTRUCTIVE: Remove a purchased sending domain's web forwarding (by domainName) — visitors to the domain stop being redirected.DestructiveRemove Domain Forwarding
DESTRUCTIVE: Remove a purchased sending domain's web forwarding (by domainName) — visitors to the domain stop being redirected. Only call after the user explicitly confirms.
Inputs
domainNamestringrequired- No description.
fuseaimcp_remove_list_rows_toolRemoves contacts or companies from a prospect list and deletes their values in the list's custom and smart columns (adding them back does not restore those values); the contacts and companies themselves stay saved.DestructiveRemove List Rows
Removes contacts or companies from a prospect list and deletes their values in the list's custom and smart columns (adding them back does not restore those values); the contacts and companies themselves stay saved. Campaign sequences already running for them are not stopped, but later messages that use this list's columns are written without those values. If the list is one of an active dynamic campaign's exclusion lists (for example 'Unsubscribed Contacts'), the people it no longer excludes can be enrolled and messaged at the campaign's next refresh (within 30 minutes). Destructive — only call after the user has confirmed. Provide the listId plus the ids under `contacts` or `companies` as [{ action: 'remove', ids: [...] }].
Inputs
listIdstringrequired- No description.
companiesarray- Remove operations for companys
contactsarray- Remove operations for contacts
fuseaimcp_rename_list_toolRenames a prospect list.DestructiveRename List
Renames a prospect list. Provide the listId and the new name. Renaming a list FuseAI fills automatically, such as 'Unsubscribed Contacts', 'Not Interested' or 'All Website Visitors', detaches it: new entries then go to a newly created list with the original name. Campaigns that exclude the renamed list stop excluding people who unsubscribe or decline afterwards, so an active dynamic campaign can enroll and message them, and campaigns or automations that use the renamed list stop receiving new entries.
Inputs
listIdstringrequired- No description.
namestringrequired- The new list name
fuseaimcp_reorder_list_columns_toolReorders the columns of a prospect list.DestructiveReorder List Columns
Reorders the columns of a prospect list. Provide listId and the complete ordered array of column ids.
Inputs
columnOrderarrayrequired- No description.
listIdstringrequired- No description.
fuseaimcp_reorder_sequence_steps_toolReorders the steps of one list of a campaign's sequence.DestructiveReorder Sequence Steps
Reorders the steps of one list of a campaign's sequence. Pass the complete list of that list's sequenceStepIds in the new order: the root list by default, or a lane when parentStepId (the condition step's id) and branchKey ('yes' or 'no'; the arm 'a' or 'b' of an A/B split) are given. A condition step must stay the last step of its list. In the root list the new first step's wait is set to 0 days and a step moved out of first place gets a 3-day wait; other steps keep their waits. Once a campaign with a condition or A/B split has started, its steps cannot be reordered; on a started campaign without them, contacts still in the sequence get their remaining steps in the new order.
Inputs
campaignIdstringrequired- No description.
sequenceStepIdsarrayrequired- Every step of one list (the root list, or the lane named by parentStepId + branchKey), in the new order, each once.
branchKeystring- Which lane of parentStepId. Under an invite_accepted or clicked_link condition: 'yes' (the signal arrived in time) or 'no'. Under an ab_split: the arm, 'a' or 'b'. Sent together with parentStepId.one of
yesnoab parentStepIdstring- The condition step whose lane this list is. Omit (or null) for the root list. Sent together with branchKey.
fuseaimcp_rerun_smart_column_toolRe-runs an existing smart column and cancels any run of it still in progress.DestructiveRerun Smart Column
Re-runs an existing smart column and cancels any run of it still in progress. mode all (the default) recomputes the rows selected by its most recent run's settings (row limit, sort, filters) and replaces their current values, manual edits included; new_only fills only rows without a completed value, with no row limit. Starts a background job and returns a jobId; poll get_smart_column_job_status_tool. Costs the column's per-row credit rate (20 standard, 100 deep_research, 10 writer) for each row computed, and is refused when the balance cannot cover every target row (a run in progress is cancelled even then). Campaigns that use the list then use the new values in their messages (a step with the column's {Column: <name>} token holds a contact while their row is recomputed), and each non-empty value triggers the user's 'Smart column updated' automations. Provide listId, the existing smart column's columnId, and mode (all|new_only).
Inputs
columnIdstringrequired- No description.
listIdstringrequired- No description.
modestring- No description.one of
allnew_onlydefaultall
fuseaimcp_resolve_scheduled_email_approvals_toolApprove or decline drafts awaiting review, in bulk.DestructiveResolve Scheduled Email Approvals
Approve or decline drafts awaiting review, in bulk. isApproved defaults to true: a call that omits it approves every listed draft that is awaiting review; pass isApproved false to decline. Approving queues them to send to their contacts from the campaign's connected mailbox or LinkedIn account: in a chat with the user, approve only drafts whose text they have seen and agreed to send. Declining stops any listed send that has not gone out yet (a draft awaiting review or any queued send, approved or not), and that lead's sequence does not move on to a next step. Reports how many rows the decision actually applied to.
Inputs
scheduledEmailIdsarrayrequired- No description.
isApprovedboolean- No description.default
true
fuseaimcp_respond_to_linkedin_invitationAccept or decline a received LinkedIn connection invitation on the LinkedIn account that received it, which can be one a teammate shared with the user.DestructiveRespond to LinkedIn Invitation
Accept or decline a received LinkedIn connection invitation on the LinkedIn account that received it, which can be one a teammate shared with the user. It takes effect on LinkedIn right away: accepting connects that account with the inviter, and a decline cannot be undone. After an accept, a background sync adds the new connection to FuseAI's All LinkedIn Connections list, creating a contact or updating a matching saved contact's name and LinkedIn URL, where the user's 'Added to list' automations or an active campaign that targets that list can pick them up. Requires the accountId, id (pass it as invitationId), sharedSecret and provider of a received invitation from list_linkedin_invitations.
Inputs
accountIdstringrequired- No description.
actionstringrequired- No description.one of
acceptdecline invitationIdstringrequired- No description.
providerstringrequired- No description.one of
LINKEDINLINKEDIN_RECRUITERLINKEDIN_SALES_NAVIGATOR sharedSecretstringrequired- No description.
fuseaimcp_respond_to_team_invitationAccept or reject a pending team invitation the user has received (invitationId is the `_id` from list_team_invitations).DestructiveRespond To Team Invitation
Accept or reject a pending team invitation the user has received (invitationId is the `_id` from list_team_invitations). Rejecting declines it for good; the team would have to invite the user again. Accepting cannot be fully undone: the user joins the inviting team with the invited role; their own team is set aside until they leave and the invitations it had sent stop working; their credit balance is reset, with unused credits from a paid plan or bought credit packs moving to the inviting team's owner and free or trial credits lost; any paid FuseAI subscription they hold is cancelled with FuseAI's billing provider, Stripe, and will not renew; they are signed out of their FuseAI sessions; and a background job copies the team owner's contact enrichments onto their contacts. Accepting is refused if the user is already a member of another team, still has members on their own team, holds a lifetime deal, or the inviting team has no free seat.
Inputs
actionstringrequired- No description.one of
acceptreject invitationIdstringrequired- The invitation's _id (a 24-hex id)
fuseaimcp_resubscribe_contact_toolReverses an accidental campaign unsubscribe for a CRM contact.DestructiveResubscribe Contact
Reverses an accidental campaign unsubscribe for a CRM contact. Removes them from the Unsubscribed Contacts system list (so new campaigns and dynamic enrollment stop skipping them) and resumes every campaign lead of yours that was stopped with outcome Unsubscribe, so those campaigns send them their remaining steps again. Each resumed lead's next step is written right away: an AI-written or AI-personalized step costs 10 credits, and if credits or a usable sender account are missing that campaign is paused instead, with an in-app notice (and a Slack message if Slack is connected). Does not undo a manual stop or bounce. Idempotent if they are already resubscribed. Pass the CRM contactId (the same id list tools and get_prospect_contact_activity_tool use) — not a campaign targetContactId.
Inputs
contactIdstringrequired- The CRM contact id to resubscribe (24-character Mongo id). Same id list tools and get_prospect_contact_activity_tool use — not a campaign targetContactId.
fuseaimcp_save_companies_to_listSave specific company objects to one or more lists (each company wrapped as { fiberDetails: company }).DestructiveSave Companies To List
Save specific company objects to one or more lists (each company wrapped as { fiberDetails: company }). Takes explicit company objects only — it does not save by filter. The companies are also added to the user's saved companies. When FuseAI already holds a company with the same domain, the values sent replace its stored name, website, industry, size, LinkedIn page, founding year and headquarters location on that shared record, for every user who has saved it. Each company newly added to a list can start the user's active automations that use the "Company added to list" trigger, whose steps can post to the user's connected Slack, act in other apps the user connected, or find contacts at the company and add them to a campaign that messages them. The save itself charges no credits.
Inputs
companiesarrayrequired- No description.
listIdsarrayrequired- No description.
fuseaimcp_save_people_to_listSave or add people (contacts, prospects or leads) to a list: people search_people found, people an earlier answer listed, or people the user names.DestructiveSave People To List
Save or add people (contacts, prospects or leads) to a list: people search_people found, people an earlier answer listed, or people the user names. Pass up to 25 people with everything you know: each one's linkedinUrl whenever you have it, else their first and last name with their company (or its domain), and their email when you know it. It matches each person to their database profile the way list imports do: the profile must carry both names given and a current or past job at the company, so a first name alone ("Steve the CRO at Datadog") is not enough: find the person with search_people first, or ask the user for their full name. It charges 1 credit per person it saves to the list (the lookups are free, and so are people already in the list and custom contacts); creditsCharged says what it cost. It answers once the people are in the list (a campaign can use them right away): saved, and notFound (for reason no_match, say so in one sentence and offer to save them as custom contacts, then call again with saveUnmatchedAsCustom true only once the user agrees; for needs_more_detail, ask for their full name and company or their LinkedIn URL; it never saves a temporarily_unavailable person as custom, so for that reason offer to try again later instead). needsChoice is always empty. Give exactly one list: newListName (created, or the user's contact list of exactly that name), listId, or existingListName (the user's contact list of that name or, failing an exact match, the closest name containing every word of it). Each person newly added to the list starts the user's automations that run when someone is added to a list, and people added to a list that an active dynamic campaign uses can be enrolled in it at its next refresh (every 30 minutes) and start receiving its outreach. Saving a person matched to a contact FuseAI already holds can update that contact's profile fields (such as title, company and location) and can turn the team's custom contact for them into a shared database contact. Files and CSVs of rows go through import_people_to_list instead.
Inputs
peoplearrayrequired- The people to save, at most 25, each with everything you know about them. Each answer refers to a person by their index in this array.
existingListNamestring- Save into the user's existing contact list with this name.
listIdstring- Save into this existing contact list, by its 24-hex id (the list.id an earlier call answered).
newListNamestring- Save into a new contact list with this name; when the user already has a contact list of exactly this name, the people are added to it. Give exactly one of newListName, listId and existingListName.
saveUnmatchedAsCustomboolean- Save the people no profile matched (notFound reason no_match or needs_more_detail) as custom contacts (their details as given, no database profile). Only after the user agreed to that for those people. A person whose lookup failed (temporarily_unavailable) stays in notFound whatever this says.default
false
fuseaimcp_save_searchCreate a saved search (a named filter set the user can reuse), or, with savedSearchId, overwrite one: the title, filters and type sent replace the stored ones, and the previous values are not kept.DestructiveSave Search
Create a saved search (a named filter set the user can reuse), or, with savedSearchId, overwrite one: the title, filters and type sent replace the stored ones, and the previous values are not kept.
Inputs
filtersstringrequired- Filter object to persist. People filters use FLAT dotted keys with {status, value} entries for include/exclude fields — the same shape the search tools take. They are stored the way the app shows a saved search, so a few people filters hold less here: full_name, work_email, mobile_phone, personal_emails, certifications.name and linkedin_url hold ONE include value; location_region and job_company_location_region hold include values only; job_start_date holds one start month (YYYY-MM) in the past, kept as months in role. first_name and last_name cannot be saved. Anything else is refused with 422 SAVED_SEARCH_FILTERS_UNSUPPORTED naming the filter.
searchTypestringrequired- Whether these are people ('contacts') or company filters.one of
contactscompanies titlestringrequired- Name for the saved search.
savedSearchIdstring- Provide to UPDATE an existing saved search; omit to create a new one.
fuseaimcp_search_and_save_people_to_listSearches an external data provider's database of professional profiles by job title, seniority, company, industry, location, skills, and other attributes, then saves the matching contacts to a Prospect List in a background job that takes a few minutes.DestructiveSearch and Save People to List
Searches an external data provider's database of professional profiles by job title, seniority, company, industry, location, skills, and other attributes, then saves the matching contacts to a Prospect List in a background job that takes a few minutes. Returns the created or updated list result only. It costs 1 credit per person saved, plus 1 credit for the lookup that finds the first match, and starts only when the balance covers every person asked for. When FuseAI already holds a matched person as a contact, that contact's stored name, title, company, location and work history are replaced with the profile data found. Each person newly added to a list can start the user's active "Added to list" automations, and an active dynamic campaign that uses the list enrolls them and starts contacting them on its next refresh. ADDING TO AN EXISTING LIST: the people already on it are left out of the search, so `count` is how many people new to the list it adds, and the same filters sent again add the next ones; `skippedListMemberCount` says how many it left out (up to 1,000, so on a bigger list some can be found again). SEARCH WITHIN A LIST: with `withinListId` (the list's id) or `withinListName`, the search runs over the members of that list only (their LinkedIn connections list, a saved list, any contact list they or a teammate shares), looked up by their LinkedIn profiles, and saves the members who match. Use it for which of my connections / which people in list X work at a kind of company or hold a kind of role: it answers from the database's data about them, not from the list's own fields, and never needs the list's rows read.
Inputs
filtersobjectrequired- People-search filters as FLAT dotted keys (e.g. job_title_levels, location_region, education.school.name — never a nested education/experience object). Include/exclude fields take arrays of {status: include|exclude, value} objects; a bare string or an array of strings are includes, and an exclude is always an entry. location_name/skills/education.school.name take string arrays; keywords is a string array; numeric ranges go under rangeInputs. Unknown keys are rejected, inside company_filters too. These filters are no longer served and are ignored if sent, so they never narrow the search: sex, github_url, github_username, facebook_url, job_title_sub_role, experience.title.sub_role, experience.company.industry, experience.company.location.country, experience.company.location.name, experience.company.location.continent, experience.company.type, location_continent, job_company_location_continent.
countinteger- The number of prospects to find. Defaults to 10. Max 500. For an existing list, only people not on it yet count.
existingListNamestring- Existing Prospect List name when the user asks to add to a list.
newListNamestring- Name for a new Prospect List. Infer this when the user does not provide one.
withinListIdstring- The id of the contact list to search WITHIN: only its members are searched, by their LinkedIn profiles, so the matches are the members who fit the filters ("which of my LinkedIn connections work at YC companies" -> the id of their LinkedIn connections list). The user's own list or a teammate's shared one. Prefer it to withinListName. A list of more than 10000 members can't be searched this way, and a `linkedin_url` filter cannot be sent with it. The matches are saved to the destination list (newListName or existingListName) like any search.
withinListNamestring- The name of one of the user's OWN contact lists to search within, when its id is not at hand (withinListId is preferred: the name a list is shown under can differ from its own). Same rules as withinListId.
fuseaimcp_send_action_emailExecute (send) a single suggested action from the Smart Actions feed, by its id.DestructiveSend Smart Action Email
Execute (send) a single suggested action from the Smart Actions feed, by its id. Sends the action's drafted email to its recipient from the action owner's connected Gmail, Outlook or SMTP mailbox (a teammate's, if the action is theirs), which cannot be undone, then marks the feed item and its linked task completed: in a chat with the user, execute only an action whose message they have seen and agreed to send. Only email send_message actions can be sent (LinkedIn message and connection actions return INVALID_CHANNEL_FOR_SEND), and an action whose recipient email has not been found yet must go through enrich_action first. A sent email can start the user's automations that trigger on emails sent to a contact. If the provider rate-limits the mailbox or rejects its credentials, FuseAI switches the mailbox off or disconnects it and notifies its owner; campaign steps that send from it are paused until it is turned back on or reconnected.
Inputs
actionIdstringrequired- The action id (24-character Mongo id) — NOT the actionFeedId. Feed items surface it as `actionId`.
fuseaimcp_send_emailSend an email from the user's connected Gmail, Outlook or SMTP mailbox.DestructiveSend Email
Send an email from the user's connected Gmail, Outlook or SMTP mailbox. If the provider rate-limits the mailbox or rejects its credentials, FuseAI switches the mailbox off or disconnects it and notifies its owner; campaign steps that send from it are paused until it is turned back on or reconnected. 'from' must be one of the user's connected sending addresses; to/cc/bcc are arrays of email addresses; content is the message body. Pass scheduledDateTime (ISO) to schedule the send instead of sending immediately; a scheduled send goes out as a separate email to each 'to' address, each copied to every cc and bcc address, and cannot be cancelled once scheduled. To reply within an existing conversation, pass inReplyToEmailId (the emailId of the message being replied to) or conversationKey. The email reaches its recipients as sent and cannot be recalled: in a chat with the user, send only an email whose recipients and text they have seen and agreed to send. Sending or scheduling also stops the pending campaign sequence steps and discards the pending AI reply drafts of any 'to' recipient who is a contact in a campaign of this mailbox's owner or of a teammate it is shared with, and each sent email can start the user's automations that trigger on emails sent to a contact. Each call sends or schedules a new email, so calling again, including a retry after an error or timeout, can send the same email twice. Before retrying an immediate send, check list_emails with labelType 'sent'; if the email is not listed there, ask the user before calling again, because it may still have gone out: when the user hides non-campaign emails, a sent email to someone outside their campaigns is not listed, and after a timeout the sent copy can appear only later. A scheduled send is not listed in list_emails, not even under labelType 'scheduled', until it goes out, so after an error or timeout on a scheduled send, ask the user before calling again.
Inputs
contentstringrequired- No description.
fromstringrequired- No description.
subjectstringrequired- No description.
toarrayrequired- No description.
bccarray- No description.
ccarray- No description.
conversationKeystring- No description.
inReplyToEmailIdstring- No description.
scheduledDateTimestring- No description.
fuseaimcp_send_linkedin_invitationSend a LinkedIn connection invitation to a profile URL from a connected LinkedIn account: the user's own, or one a teammate shared with them, in which case it goes out under the teammate's name.DestructiveSend LinkedIn Invitation
Send a LinkedIn connection invitation to a profile URL from a connected LinkedIn account: the user's own, or one a teammate shared with them, in which case it goes out under the teammate's name. Each send uses one of that account's daily and weekly invitations, the same allowance its campaigns use, and fails once they are used up. The invitation, and its note when there is one, reaches the person right away and cannot be taken back: in a chat with the user, send only an invitation they have seen and agreed to send.
Inputs
accountIdstringrequired- No description.
linkedinUrlstringrequired- No description.
invitationTextstring- No description.
fuseaimcp_send_linkedin_messageSend a message in an existing LinkedIn conversation.DestructiveSend LinkedIn Message
Send a message in an existing LinkedIn conversation. chatId is the LinkedIn conversation's id; accountId is the id of the connected LinkedIn account that sends the message. The message reaches the contact right away and cannot be unsent: in a chat with the user, send only text they have seen and agreed to send. Sending also stops the pending campaign sequence steps and discards the pending AI reply drafts (with their approval cards and follow-up tasks) of the contact if they are in a campaign of this account's owner or of a teammate the owner shares LinkedIn accounts with for campaigns. Each message counts against the account's daily and weekly LinkedIn message limits.
Inputs
accountIdstringrequired- No description.
chatIdstringrequired- No description.
messageTextstringrequired- No description.
fuseaimcp_send_slack_messageSend a Slack message to the user's connected workspace: a channel (channelId or #name), a DM (slackUserId = Slack member id), or the integration's configured channel when neither is given.DestructiveSend Slack Message
Send a Slack message to the user's connected workspace: a channel (channelId or #name), a DM (slackUserId = Slack member id), or the integration's configured channel when neither is given. A teammate without their own Slack posts with the team owner's connection. Optional items [{ title, url?, note? }] (max 25) render as a bulleted list under the text. Returns messageTs and channel. Each call posts a new message, which this tool cannot unsend or edit.
Inputs
textstringrequired- The message text (Slack mrkdwn)
channelIdstring- Channel id or #name; defaults to the integration's channel
itemsarray- Rendered as a bulleted list under the text
slackUserIdstring- Slack member id to DM instead of a channel
fuseaimcp_send_test_sequence_email_toolSend a real test email of one of a campaign's email steps from one of the user's connected mailboxes (fromEmail if given) to toEmail, so the user can see how it looks before activating.DestructiveSend Test Sequence Email
Send a real test email of one of a campaign's email steps from one of the user's connected mailboxes (fromEmail if given) to toEmail, so the user can see how it looks before activating. Merge fields are filled from a sample contact on the campaign's lists, and the step's CC addresses, if any, also receive it. Email steps only; nothing is recorded on the campaign.
Inputs
campaignIdstringrequired- No description.
sequenceNumberintegerrequired- No description.
toEmailstringrequired- No description.
fromEmailstring- No description.
sequenceStepTypestring- No description.one of
ai_generatedai_replymanual
fuseaimcp_set_crm_owner_mappingMap CRM owners to FuseAI users, replacing each listed owner's current mapping.DestructiveSet CRM Owner Mapping
Map CRM owners to FuseAI users, replacing each listed owner's current mapping. The mapping is saved in FuseAI and decides whose CRM records each user sees in FuseAI (for example "My HubSpot Contacts") and which CRM owner FuseAI sets on contacts it later pushes to the connected CRM, including the scheduled auto-push when it is on. Pass `mappings` as an array of { crmOwnerId, mappedUserId }: `crmOwnerId` is the `_id` of an owner row from list_crm_owners (not the CRM's own user id); `mappedUserId` is the FuseAI user id (null to unmap). Needs crmIdentifier + crmAccountId.
Inputs
crmAccountIdstringrequired- No description.
crmIdentifierstringrequired- No description.one of
hubspotsalesforcezohoattiopipedrivegohighlevelnotion mappingsarrayrequired- No description.
fuseaimcp_set_custom_propertiesSet which custom contact properties of the connected CRM FuseAI pulls onto the team's contacts (max 100).DestructiveSet Custom Properties
Set which custom contact properties of the connected CRM FuseAI pulls onto the team's contacts (max 100). Send the full list: it replaces the current one, and an enabled property left out is removed. Each crmPropertyName must be a custom property that exists on the connected CRM. Removing a property deletes its pulled values from all of the team's contacts in FuseAI (the CRM itself is not changed); confirm with the user before removing any.
Inputs
crmIdentifierstringrequired- The connected CRMone of
hubspotsalesforceattiozohopipedrivegohighlevelnotion propertiesarrayrequired- No description.
fuseaimcp_set_domain_forwardingSet or update a purchased sending domain's web forwarding, replacing any current target: domainName + forwardTo (http/https URL).DestructiveSet Domain Forwarding
Set or update a purchased sending domain's web forwarding, replacing any current target: domainName + forwardTo (http/https URL). The change is made live at the domain provider, so every visitor to the domain is redirected to forwardTo. Active domains only.
Inputs
domainNamestringrequired- No description.
forwardTostringrequired- No description.
fuseaimcp_set_email_account_active_statusToggle whether an email account is used in campaigns ("active"): emailAddress + isActive.DestructiveSet Email Account Active Status
Toggle whether an email account is used in campaigns ("active"): emailAddress + isActive. A mailbox a teammate shared with the user (with campaign permission) can be switched too, and the change applies to everyone who uses it. Activating resumes the campaign emails queued for it, lets running campaigns pick it as a sender again, and lifts a switch-off the system applied after rejected sends or a provider rate limit. Deactivating holds its queued campaign emails and stops new campaign sends from it; a running campaign left with no active sending account may stop.
Inputs
emailAddressstringrequired- No description.
isActivebooleanrequired- No description.
fuseaimcp_set_fuse_property_mappingsHIGH-IMPACT: set which Fuse properties FuseAI writes onto the team's CRM contact records, and into which CRM property each goes.DestructiveSet Fuse Property Mappings
HIGH-IMPACT: set which Fuse properties FuseAI writes onto the team's CRM contact records, and into which CRM property each goes. Send the full list; a Fuse property left out stops being written. Each crmPropertyName must be an existing, writable contact property of the connected CRM. Saving a change starts a background sync right away (it also runs hourly) that overwrites that property on every existing CRM contact record FuseAI has a value for, and clears a value FuseAI wrote there when the contact's status moves to one left unmapped. Confirm with the user first. HubSpot only for now: any crmIdentifier other than "hubspot" is refused (422 FUSE_PROPERTY_SYNC_NOT_SUPPORTED). Owner or admin only.
Inputs
crmIdentifierstringrequired- The connected CRMone of
hubspotsalesforceattiozohopipedrivegohighlevelnotion mappingsarrayrequired- No description.
fuseaimcp_single_enrichment_toolEnriches a single contact by LinkedIn URL or by name plus company, querying a waterfall of data providers (Fuse, Apollo, Hunter, ContactOut, Datagma, and others).DestructiveSingle Contact Enrichment
Enriches a single contact by LinkedIn URL or by name plus company, querying a waterfall of data providers (Fuse, Apollo, Hunter, ContactOut, Datagma, and others). You MUST pass enrichType: email, phone, or all based on what the user asked for. It costs credits when it finds a new deliverable email or valid phone number (values the given contact already holds are not charged) and nothing when it finds none. Pass the identifiers nested under `input`: { url } for a LinkedIn URL, or a name (name, or firstName + lastName) with the company as `input.domain` (its website domain) or `input.company` (its name); to enrich an EXISTING saved contact, also pass its id as `input.contactId`. Returns `emails`, `phones`, `primaryEmail`, `primaryPhone`, and a `profile` object with `fullName`, `firstName`, `lastName`, `jobTitle`, `company`, `companyDomain`, `industry`, `city`, `state`, `country`, and `linkedinUrl` whenever profile data is available (null when not). It also saves the result to the user's contacts, which their team shares: without `input.contactId` the person is saved as a contact even when no email or phone is found (a saved contact with the same LinkedIn profile is updated instead). The emails and phones found are merged into the contact, which can change a re-found address's status and the primary email and drop stored work emails at a past employer's domain, and when a provider returns the person's profile, the contact's name, LinkedIn URL, job title, company and location are overwritten with it. When the enriched contact is on a list that an active dynamic campaign uses, an email or LinkedIn URL it gains can get the person enrolled in that campaign at its next refresh (every 30 minutes) and then sent its outreach. Adding the person to a list is a separate write; when doing that afterwards, carry ALL returned profile fields over (do not drop jobTitle/company/city/country). Optional inputs: `input.customContactId` is an older name for `input.contactId` (send one, not both); `input.location` is the person's location, used to help match them by name; `extendedEnrich` true keeps asking providers after the first one that finds a value and skips results cached from earlier lookups; `resumeAfterApiUsed` (enrichType email or phone, without `extendedEnrich`) takes the `apiUsed` of a returned email or phone and continues with the providers after it until one finds a new value; both are charged by the credit rule above. `listId` names the list the lookup was started from and only updates that list's last-activity time, without adding the person to the list; `source`, used only without `input.contactId`, records how the saved person was acquired as one of FuseAI's source labels (such as agents) and never replaces a label already set.
Inputs
enrichTypestringrequired- The type of enrichment to perform. Default to "email" — only pass "phone" or "all" when the user explicitly asked for phone numbers (phone enrichment costs several times more credits per contact).one of
emailphoneall inputobjectrequired- No description.
extendedEnrichboolean- No description.
listIdstring- No description.
resumeAfterApiUsedstring- No description.
sourcestring- No description.
fuseaimcp_skip_scheduled_emailSkip one lead's pending step (an email, auto-reply, LinkedIn action, call or to-do): it is never sent, any task for it is cancelled, and the skip cannot be undone.DestructiveSkip Scheduled Email
Skip one lead's pending step (an email, auto-reply, LinkedIn action, call or to-do): it is never sent, any task for it is cancelled, and the skip cannot be undone. If the lead's sequence is still running, it then advances: the next step is written and scheduled to go out to the lead, timed from the skipped step's date, so if that step was overdue the next one can go out at the next sending window. An AI-written next step costs credits; without enough credits the campaign is paused instead. Declining instead stops the whole sequence.
Inputs
scheduledEmailIdstringrequired- No description.
reasonstring- No description.
fuseaimcp_start_email_warmupStart (resume) managed deliverability / email warmup for a sending account: while it runs, the external warmup service sends warmup emails and replies from that mailbox, under its sender name, until warmup is paused.DestructiveStart Email Warmup
Start (resume) managed deliverability / email warmup for a sending account: while it runs, the external warmup service sends warmup emails and replies from that mailbox, under its sender name, until warmup is paused. `id` is the account's managedDeliverabilityCampaignId from the email-accounts list, passed AS A STRING — the field is numeric, so quote it (e.g. "18234"; an unquoted number is rejected). Only works for accounts whose warmup was already set up (managedDeliverabilityStatus paused/active); first-time warmup enabling happens in the FuseAI app.
Inputs
idstringrequired- No description.
fuseaimcp_start_linkedin_chatStart a LinkedIn chat from a connected account (accountId) with one or more people and send them a message.DestructiveStart LinkedIn Chat
Start a LinkedIn chat from a connected account (accountId) with one or more people and send them a message. Requires messageText plus either attendeesIds (LinkedIn provider ids) or one person's linkedinUrl, which is looked up live on LinkedIn without a profile-visit notification. If the message goes to one person the account already has a one-to-one chat with, LinkedIn adds it to that chat instead of opening a new one. Returns the chatId of the chat the message went to. The message reaches them right away and cannot be unsent: in a chat with the user, send only text they have seen and agreed to send. Without isInMail the message can reach only 1st-degree connections; isInMail true sends an InMail instead, which needs LinkedIn Premium, Sales Navigator or Recruiter, uses one of the account's LinkedIn InMail credits and counts against its daily, weekly and monthly InMail limits. subject is the InMail subject line the recipient sees; it is passed to LinkedIn whenever it is non-empty, even without isInMail, so set it only for an InMail and only to wording the user has seen and agreed to send. Each message counts against the account's daily and weekly LinkedIn message limits. Unless campaignId is set, sending also stops the pending campaign sequence steps and discards the pending AI reply drafts (with their approval cards and follow-up tasks) of the recipient (the first one, when there are several) if they are in a campaign of the account's owner or of a teammate the owner shares LinkedIn accounts with for campaigns. campaignId and campaignSequenceStepId are for FuseAI's own campaign sends: campaignId links the chat to that campaign, replacing any campaign it was linked to; campaignSequenceStepId is ignored without campaignId and is otherwise used only to tag the message it sends in FuseAI's stored copy of the chat with that sequence step, changing nothing in the campaign or the step; voiceMessagePreviousAttemptsAt has no effect here. If the lookup finds that the account's LinkedIn session has expired, FuseAI marks the account disconnected, which pauses its campaigns until it is reconnected, and notifies the owner.
Inputs
accountIdstringrequired- No description.
attendeesIdsstring- No description.
campaignIdstring- No description.
campaignSequenceStepIdstring- No description.
isInMailboolean- No description.
linkedinUrlstring- No description.
messageTextstring- No description.
subjectstring- No description.
voiceMessagePreviousAttemptsAtstring- No description.
fuseaimcp_sync_crm_ownersRe-read the connected CRM's users (nothing is written to the CRM) and refresh FuseAI's stored CRM owners: new owners are added, and stored owners' names and emails are overwritten with the CRM's.DestructiveSync CRM Owners
Re-read the connected CRM's users (nothing is written to the CRM) and refresh FuseAI's stored CRM owners: new owners are added, and stored owners' names and emails are overwritten with the CRM's. When a new owner is added, FuseAI also maps every owner with no team member mapped, including one an admin unmapped on purpose, to the team member with the same email or name; that mapping decides whose CRM records a member sees in FuseAI and which CRM owner FuseAI sets on contacts it pushes. Use when locally stored CRM-owner data looks stale.
Inputs
crmIdentifierstringrequired- The connected CRMone of
hubspotsalesforceattiozohopipedrivegohighlevelnotion
fuseaimcp_update_action_draftEdit a single action's draft (e.g.DestructiveUpdate Smart Action Draft
Edit a single action's draft (e.g. its message, subject or recipient) by id, before it is sent. This tool sends nothing, and no edited draft goes out on its own: an email send_message draft is sent only when someone explicitly sends the action, with send_action_email or the Send button in the FuseAI app; FuseAI does not send LinkedIn message or connection drafts from an action; and an approve_reply action's config only links to the campaign's AI reply draft (campaignId, messageId, channel), so editing it does not change that reply. Saving a config also marks the action's open task as started. `config` replaces the action's whole stored config instead of merging into it: start from the config get_action_detail returns and send it back complete with your changes (message under config.content.subject / config.content.body, recipient under config.recipient); any field you leave out is removed. Changing config.recipient changes who send_action_email will email. The only other editable field is isSystemGenerated, a label marking the action as created by FuseAI (true, the default) or by a person (false). Nothing in FuseAI acts on this label, so a call that changes only isSystemGenerated sends nothing and leaves the draft and the task unchanged.
Inputs
actionIdstringrequired- The action id (24-character Mongo id) — NOT the actionFeedId. Feed items surface it as `actionId`.
configobject- No description.
isSystemGeneratedboolean- No description.
fuseaimcp_update_action_feed_itemUpdate a Smart Actions feed item by id (e.g.DestructiveUpdate Smart Actions Feed Item
Update a Smart Actions feed item by id (e.g. change its status, snooze, pin or reassign it), overwriting the fields you pass. Takes the actionFeedId (feed card id), not the actionId. Accepts any subset of: status, category, title, description, priority, assignedUserId, snoozedUntil, isPinned, dueAt, completedAt, isSystemGenerated. The feed shows a priority derived from the item's points, so setting priority does not change what the feed shows. A status change also closes, reopens or cancels the item's linked task; it never sends, approves or declines a message (status approved on a reply-approval card does not approve the reply draft).
Inputs
actionFeedIdstringrequired- The Smart Actions feed card id (24-character Mongo id) — NOT the actionId.
assignedUserIdstring- No description.
categorystring- No description.one of
taskrecommendationreminderinsightalert completedAtstring- No description.
descriptionstring- No description.
dueAtstring- No description.
isPinnedboolean- No description.
isSystemGeneratedboolean- No description.
prioritystring- No description.one of
highmediumlow snoozedUntilstring- No description.
statusstring- No description.one of
activesnoozedapproveddismissedcompletedexpireddeleted titlestring- No description.
fuseaimcp_update_assistant_imessage_textsTurn the texts Dash, FuseAI's assistant, sends the user first over iMessage (channel "imessage" only; WhatsApp has none of these texts, so "whatsapp" is always refused: 422 PROACTIVE_TEXTS_NOT_AVAILABLE) on or off: urgent-task alerts (a text when a high or urgent task comes in, 8am-8pm local) and the weekday morning brief, and set the brief's local hour (0-23).DestructiveUpdate Assistant iMessage Texts
Turn the texts Dash, FuseAI's assistant, sends the user first over iMessage (channel "imessage" only; WhatsApp has none of these texts, so "whatsapp" is always refused: 422 PROACTIVE_TEXTS_NOT_AVAILABLE) on or off: urgent-task alerts (a text when a high or urgent task comes in, 8am-8pm local) and the weekday morning brief, and set the brief's local hour (0-23). Send only what changes; leave timezone out to keep the user's saved one. Turning alerts or the brief on also texts the user a confirmation about a minute later, unless they texted Dash in the five minutes before. Needs a connected number: 409 CHANNEL_LINK_NOT_ACTIVE means the user connects it in Settings first.
Inputs
channelstringrequired- The messaging channelone of
imessagewhatsapp alertsEnabledboolean- Text the user about high and urgent tasks as they arrive
briefEnabledboolean- Send the weekday morning brief
briefLocalHourinteger- Local hour the morning brief goes out
timezonestring- IANA timezone the alert hours and the brief follow (the user's browser); left out, the saved one stays
fuseaimcp_update_automationUpdate an automation by automationId: rename, change the summary, replace triggers/steps (same shapes as create_automation, validated against list_automation_nodes), set expiresAt, or change status.DestructiveUpdate Automation
Update an automation by automationId: rename, change the summary, replace triggers/steps (same shapes as create_automation, validated against list_automation_nodes), set expiresAt, or change status. Replacing triggers or steps overwrites the rule's current graph. Steps and triggers that choose their operation only when the rule runs (mcp_call, composio_call, agent_step and run_code steps, and composio_event triggers) are refused here and nothing is saved; rules that need them are built in the FuseAI app. Status "active" starts or resumes the rule and clears its failure counter; an active rule then runs on its own whenever a trigger fires, each run costs credits, and its steps may add contacts to campaigns (which then email and message them), post to Slack or run the other steps it holds, including steps added in the FuseAI app that act in the user's connected apps. "paused", "draft" or "expired" stops the rule, and its queued runs and runs waiting on a wait step are cancelled when they come due. Only the rule's owner may edit it; on a teammate's rule the team's owner and admins may only change its status (403 AUTOMATION_ACCESS_DENIED otherwise).
Inputs
automationIdstringrequired- No description.
compilerNotesarray- No description.
expiresAtstring- No description.
namestring- No description.
statusstring- No description.one of
activepauseddraftexpired stepsarray- No description.
summarystring- No description.
triggersarray- No description.
fuseaimcp_update_campaign_status_toolStarts, resumes, or pauses a campaign based on its current status.DestructiveUpdate Campaign Status
Starts, resumes, or pauses a campaign based on its current status. ACTIVATING STARTS SENDING REAL OUTREACH to prospects — never activate as an implicit part of another task; only on an explicit user request, and confirm with the user first. Activate: initializing (ready) uses first-time launch, which enrolls the contacts of the campaign's lists; stopped uses re-activate, which also rewrites all of the campaign's queued messages from its current steps, replacing hand edits, when any of them is over 60 days old. A running campaign charges credits for each AI-written or AI-personalized message it writes. Deactivate: only when the campaign is active. Pass campaignId when known, otherwise campaignName for fuzzy lookup.
Inputs
actionstringrequired- Use "activate" to start or resume outreach (draft → live, or stopped → active). Use "deactivate" to pause a running campaign (active → stopped).one of
activatedeactivate campaignIdstring- MongoDB id of the campaign. Prefer it over campaignName whenever the id is known (e.g. for a campaign just created in this conversation). Use with action activate or deactivate.
campaignNamestring- Campaign title as the user said it. Resolves by exact name then fuzzy match across initializing, active, and stopped campaigns. Use when no id is available.
fuseaimcp_update_campaign_toolUpdates a campaign's settings (rename via campaignName, set or change the calendar link via ctaLink, toggle auto-pilot/auto-reply via isAutoPilot, toggle the email unsubscribe footer via addUnsubscribeLink, toggle open/click tracking via openRateTrackingEnabledAt / clickTrackingEnabledAt, pin a tracking domain via trackingDomainId, change schedule via timezone / emailDays / emailStartHour / emailEndHour, set or clear per-campaign daily caps via dailyEmailLimit / dailyLinkedinConnectionLimit / dailyLinkedinMessageLimit, replace the campaign's co-owners via delegates, the first of whom takes over its dialer tasks; anyone left out loses co-owner access).DestructiveUpdate Campaign Settings
Updates a campaign's settings (rename via campaignName, set or change the calendar link via ctaLink, toggle auto-pilot/auto-reply via isAutoPilot, toggle the email unsubscribe footer via addUnsubscribeLink, toggle open/click tracking via openRateTrackingEnabledAt / clickTrackingEnabledAt, pin a tracking domain via trackingDomainId, change schedule via timezone / emailDays / emailStartHour / emailEndHour, set or clear per-campaign daily caps via dailyEmailLimit / dailyLinkedinConnectionLimit / dailyLinkedinMessageLimit, replace the campaign's co-owners via delegates, the first of whom takes over its dialer tasks; anyone left out loses co-owner access). On a running campaign the changes reach outreach already in flight: a changed ctaLink replaces the old link inside queued emails (clearing it leaves them unchanged), schedule, daily-cap and tracking settings apply when each queued or future message is sent, and with isAutoPilot on, AI-written replies to leads who respond go out without approval. Does NOT activate or deactivate the campaign and does NOT edit its sequence steps.
Inputs
campaignIdstringrequired- No description.
addUnsubscribeLinkboolean- When true, every email step appends a one-click unsubscribe footer (email channel only). When false, the footer is omitted.
campaignNamestring- New name for the campaign
clickTrackingEnabledAtstring- Rewrite links in outgoing emails to tracked redirects. Must be on for a custom tracking domain to appear in the email; otherwise links stay as the original URL. Pass true to turn on, false or null to turn off. An ISO 8601 timestamp is also accepted.
ctaLinkstring- The calendar/booking link (absolute http(s) URL) recipients follow to book a meeting; null or an empty string clears it.
dailyEmailLimitinteger- Maximum emails this campaign sends per calendar day in its timezone, across all its mailboxes and excluding replies. Mailbox limits and sharing with other campaigns still apply. Pass null to remove the cap.
dailyLinkedinConnectionLimitinteger- Maximum LinkedIn connection requests this campaign sends per calendar day in its timezone, across all its LinkedIn accounts. Account limits still apply. Pass null to remove the cap.
dailyLinkedinMessageLimitinteger- Maximum LinkedIn messages (InMails included) this campaign sends per calendar day in its timezone, across all its LinkedIn accounts. Account limits still apply. Pass null to remove the cap.
delegatesarray- Replacement list of teammate user ids (from list_team_members) who co-own the campaign. The FIRST id becomes the owner of its dialer (call) tasks: the campaign moves to that person's dialer page and its call batch is created for them instead of the creator; an already open call batch follows on its next refresh. Pass [] to hand the campaign back to its creator. Ids must be the team owner or active team members, else DELEGATE_NOT_TEAM_MEMBER.
emailDaysarray- Replacement set of weekdays on which the campaign may send. English names with a capital first letter: Monday ... Sunday.
emailEndHourstring- Local end of the daily sending window as 24-hour H:MM or HH:MM (e.g. 17:00).
emailStartHourstring- Local start of the daily sending window as 24-hour H:MM or HH:MM (e.g. 09:00).
isAutoPilotboolean- Enable (true) or disable (false) auto-pilot / auto-reply for the campaign
openRateTrackingEnabledAtstring- Enable the open-tracking pixel on outgoing emails. Pass true to turn on, false or null to turn off. An ISO 8601 timestamp is also accepted.
timezonestring- IANA time zone in which sending days and the daily window are evaluated, e.g. America/New_York. Invalid zones are rejected (the scheduler would otherwise silently fall back to UTC).
trackingDomainIdstring- Id of an ACTIVE team tracking domain from list_tracking_domains_tool. Null unsets the per-campaign pin so send-time click tracking uses the team's default ACTIVE domain if one exists. Click tracking must also be enabled or links are not rewritten.
fuseaimcp_update_campaign_visibilityReplace the team-wide campaign visibility, which decides who can see and act on teammates' campaigns: PRIVATE (only each campaign's owner and co-owners), ADMINS_ONLY (also team owners and admins) or ALL_MEMBERS (every team member).DestructiveUpdate Campaign Visibility
Replace the team-wide campaign visibility, which decides who can see and act on teammates' campaigns: PRIVATE (only each campaign's owner and co-owners), ADMINS_ONLY (also team owners and admins) or ALL_MEMBERS (every team member). Narrowing it removes other members' access at once. Team owners and admins only.
Inputs
visibilityModestringrequired- No description.one of
PRIVATEADMINS_ONLYALL_MEMBERS
fuseaimcp_update_contact_crm_ownerReassign one contact's owner on the external CRM, replacing the current owner, and record the new owner on FuseAI's link to that CRM record; `contactId` and `crmContactId` must be the same person, because the link is moved to `contactId`.DestructiveUpdate Contact CRM Owner
Reassign one contact's owner on the external CRM, replacing the current owner, and record the new owner on FuseAI's link to that CRM record; `contactId` and `crmContactId` must be the same person, because the link is moved to `contactId`. Not supported for Attio, or for Notion without a bound Owner property; for a Salesforce or Zoho lead pass `crmProfileType: "lead"` (the default is contact). Needs `newCrmOwnerId` (the new owner's CRM user id), the FuseAI `contactId`, the `crmContactId`, and `crmAccountId`.
Inputs
contactIdstringrequired- No description.
crmAccountIdstringrequired- No description.
crmContactIdstringrequired- No description.
crmIdentifierstringrequired- The connected CRMone of
hubspotsalesforceattiozohopipedrivegohighlevelnotion newCrmOwnerIdstringrequired- No description.
crmProfileTypestring- No description.one of
leadcontact isCustomContactboolean- No description.default
false
fuseaimcp_update_contact_noteEdit a contact note by noteId: replace its body.DestructiveUpdate Contact Note
Edit a contact note by noteId: replace its body. Only the note's author may edit it (canEdit from list_contact_notes); anyone else gets CONTACT_NOTE_NOT_AUTHOR. Returns the updated note.
Inputs
bodystringrequired- New plain-text body; replaces the old one.
noteIdstringrequired- The note id from list_contact_notes or the contact timeline.
fuseaimcp_update_contact_primary_email_or_phoneSet the calling user's primary email or primary phone for a contact saved in FuseAI, replacing the current one; teammates keep their own primaries.DestructiveSet Contact Primary Email/Phone
Set the calling user's primary email or primary phone for a contact saved in FuseAI, replacing the current one; teammates keep their own primaries. Pass the contactId plus primaryEmail and/or primaryPhone. A phone is saved in E.164 form, and setting one also rewrites the user's stored phone list in that form, merging duplicates. Each call also restores the contact, with that teammate's earlier data, for any teammate who had deleted it, and adds an empty record of it for teammates who never held it; a restored contact's emails are back too, so that teammate's active campaigns on lists holding the contact can start emailing the person again. Contacts already enrolled in a campaign keep the address they were enrolled with. The value must be one of the emails/phones ALREADY STORED in the user's own record of the contact; a value that record does not have fails with PRIMARY_EMAIL_NOT_IN_EMAILS / PRIMARY_PHONE_NOT_IN_PHONES instead of being added.
Inputs
contactIdstringrequired- The CRM contact id (24-character Mongo id)
primaryEmailstring- The new primary email address
primaryPhonestring- The new primary phone number
fuseaimcp_update_crm_settingsUpdate the team's CRM sync settings.DestructiveUpdate CRM Settings
Update the team's CRM sync settings. Only the fields sent change, and each replaces its stored value (send disabledContactFields, contactPushConditions and activitySyncSettings in full): profileType (lead|contact|null; lead only for CRMs with a lead object: Salesforce, Zoho), the contact and company auto push/pull flags, activitySyncEnabled, activitySyncSettings, activityPullEnabled, userForSyncId (the team member whose CRM connection runs the scheduled syncs; null clears it), and disabledContactFields (the fields left out of the sync; entries must be valid availableContactFields keys of the connected CRM). These settings drive FuseAI's scheduled syncs with the team's CRM: contact and company auto push create and update records in the CRM every 6 hours; activity sync, while contact auto push is on, logs emails, LinkedIn campaign messages and calls on CRM contacts and writes engagement fields every hour; the pull flags import CRM records into FuseAI. The other settings change what those runs write. contactPushConditions limits which contacts new to the CRM the scheduled push, push all and the activity sync create: up to 5 groups of { conditions: [{ field, operator: includes | excludes, values }] }, each field at most once per group; a contact is created when any group matches and a group matches when all its conditions do; [] clears them. Fields: primary_email_status, reply_outcome, lead_score (hot|warm|cold), campaign_engagement (contacted|in_active_campaign|never_engaged), phone_number (enriched|not_enriched), contact_type (fuse|custom), seniority, department, company_headcount (e.g. 51-200), lists (list ids), owner (member ids), icp_match (ICP ids, or any_icp alone, plus icpOwnerUserId: the member whose ICPs they are; get_push_condition_icp_options lists them), and free-text job_title, location, person_industry, company_name, company_industry, company_hq_country. The prospect-list fields match like the people list's filters; a contact without a value does not match an excludes condition on a free-text field. Contacts Fuse has already linked to a CRM record keep syncing; contacts picked by id and the members of real lists are always pushed, while all_enriched_contacts follows the conditions. Changing contactPushConditions is for team owners and admins only; confirm with the user first.
Inputs
crmIdentifierstringrequired- The connected CRMone of
hubspotsalesforceattiozohopipedrivegohighlevelnotion activityPullEnabledboolean- No description.
activitySyncEnabledboolean- No description.
activitySyncSettingsobject- No description.
autoPullEnabledboolean- No description.
autoPushEnabledboolean- No description.
companyAutoPullEnabledboolean- No description.
companyAutoPushEnabledboolean- No description.
contactPushConditionsarray- No description.
disabledContactFieldsarray- No description.
profileTypestring- No description.one of
leadcontact userForSyncIdstring- No description.
fuseaimcp_update_domain_visibility_settingsReplace the team-wide setting for who can see and manage the purchased domains/mailboxes: visibilityMode "PRIVATE" (each member only their own) | "ADMINS_ONLY" (owners and admins see and manage everyone's, members only their own) | "ALL_MEMBERS" (every member sees and manages everyone's).DestructiveUpdate Domain Visibility Settings
Replace the team-wide setting for who can see and manage the purchased domains/mailboxes: visibilityMode "PRIVATE" (each member only their own) | "ADMINS_ONLY" (owners and admins see and manage everyone's, members only their own) | "ALL_MEMBERS" (every member sees and manages everyone's). Managing covers deactivating, reactivating, changing domain forwarding and connecting mailboxes as sending accounts; owners and admins can also read, in the app, the sign-in credentials of every mailbox the setting shows them. A change grants or revokes this access for the whole team at once. Team owners and admins only.
Inputs
visibilityModestringrequired- No description.one of
PRIVATEADMINS_ONLYALL_MEMBERS
fuseaimcp_update_email_account_profileUpdate one email account's settings (by emailAddress): sender name, title, capacity (daily send limit, preserves today's usage), signature (HTML, sanitized), isAutoOppEnabled (a stored auto-opportunity setting that currently has no effect).DestructiveUpdate Email Account Profile
Update one email account's settings (by emailAddress): sender name, title, capacity (daily send limit, preserves today's usage), signature (HTML, sanitized), isAutoOppEnabled (a stored auto-opportunity setting that currently has no effect). At least one field required. The sender name, title and signature are what recipients see on campaign emails sent from this mailbox: the name is the From name, except on an Outlook mailbox's follow-ups sent as a reply in the same thread (the default for automatic email steps), which carry the name set in Microsoft; the three fill the {Sender …} placeholders. In the app's email account settings they are the mailbox's Name, Job title and Email Signature. New values replace the current ones for every campaign that sends from this mailbox, teammates' campaigns included, and the From name also changes on emails already queued; capacity caps the mailbox's daily campaign sends, so lowering it can stop them for the rest of the day.
Inputs
emailAddressstringrequired- No description.
capacityinteger- No description.
isAutoOppEnabledboolean- No description.
namestring- No description.
signaturestring- No description.
titlestring- No description.
fuseaimcp_update_email_flagsChange an email's flags in the user's connected Gmail, Outlook or IMAP mailbox as well as in FuseAI: read/unread, favorite, archive, spam, or delete/restore (setting archive, spam or delete to false moves the email back to the inbox).DestructiveUpdate Email Flags
Change an email's flags in the user's connected Gmail, Outlook or IMAP mailbox as well as in FuseAI: read/unread, favorite, archive, spam, or delete/restore (setting archive, spam or delete to false moves the email back to the inbox). Send any subset of the booleans for one emailId. isDeleted=true moves the email to trash — confirm with the user before using it.
Inputs
emailIdstringrequired- No description.
isArchivedboolean- No description.
isDeletedboolean- No description.
isFavoriteboolean- No description.
isReadboolean- No description.
isSpamboolean- No description.
fuseaimcp_update_email_tag_colorChange the color of an existing FuseAI inbox tag (stored in FuseAI only; nothing changes in the connected Gmail, Outlook or SMTP mailbox).DestructiveUpdate Email Tag Color
Change the color of an existing FuseAI inbox tag (stored in FuseAI only; nothing changes in the connected Gmail, Outlook or SMTP mailbox). tagName identifies the tag (tags are keyed by name); color is a 6-digit hex with leading '#' (e.g. #FF0000).
Inputs
colorstringrequired- No description.
tagNamestringrequired- No description.
fuseaimcp_update_enrichment_preferencesTurn personal-email priority on or off for the user's future contact enrichments (enrichmentPrioritizePersonalEmails: true for on, false for off), replacing the current setting; a call that omits enrichmentPrioritizePersonalEmails changes nothing.DestructiveUpdate Enrichment Preferences
Turn personal-email priority on or off for the user's future contact enrichments (enrichmentPrioritizePersonalEmails: true for on, false for off), replacing the current setting; a call that omits enrichmentPrioritizePersonalEmails changes nothing. When on, email enrichment looks for a personal email first and charges 100 credits for each personal email it finds, instead of the 20 or 50 credits a work email costs (5 either way when a provider key saved on the account finds it); when off, it looks for a work email. Only enrichments started after the change are affected: contacts already enriched, enrichments already running and campaign email lookups are not. This call only saves the setting: it runs no enrichment, contacts no data provider and charges no credits.
Inputs
enrichmentPrioritizePersonalEmailsboolean- No description.
fuseaimcp_update_icpUpdate an existing ICP by its id.DestructiveUpdate ICP
Update an existing ICP by its id. It REPLACES all of the ICP's criteria: every criterion left out is removed, so read the ICP's current criteria first and send back each one to keep along with the changes. Criteria the app no longer supports, such as gender or continents, cannot be sent and are removed if an older ICP still holds them. The ICP's description is kept when left out, and an empty string removes it. A new name that another of the user's ICPs already has (ignoring case) is refused. It never switches an ICP off or on. The new criteria take effect at once wherever this ICP is used: ICP-match conditions and Find contacts at company steps in the user's automations, CRM push conditions, and the ICP match on lists; lead scores, which count only the team owner's ICPs, use the new criteria when contacts are next rescored. Through those it can change which contacts the user's active automations act on (their steps can add contacts to campaigns that email and message them, post to Slack or act in the user's connected apps), which people Find contacts at company steps look up through an outside people-data provider, and which new contacts FuseAI creates in the team's connected CRM, such as HubSpot or Salesforce. Criteria come in three sections. People: the person (name, location, job title keywords and titles, seniority, the person's own industry, languages, skills, experience, education, LinkedIn URL, open to work, unemployed). Company: the person's current company (name, keywords its name or description must or must not carry, department, industry, headquarters location, size, website, revenue, funding, employee count). Past job: companies, titles, seniority and departments the person had before. Include/exclude criteria take { status: 'include' | 'exclude', value, label } entries. The person's location and the company's headquarters location are separate criteria, each at three levels: for the person, country, locationRegion (state) and personalLocation (city); for the company, hqCountry, jobCompanyLocationRegion (state) and location (city). A country is a lowercase full name such as "germany". A city is written "city, state, country", and a bare place name in a city criterion is refused. The industry criterion is the person's own industry; for the industries of the companies the ICP targets, use jobCompanyIndustry.
Inputs
idstringrequired- The ICP's id.
namestringrequired- The ICP name.
actualEmployeeRangeobject- Company: number of employees at the person's current company, as { min, max }, both required.
certificationstring- People: a certification the person holds, in lowercase, e.g. "pmp".
companyKeywordsarray- Company: words the company's name or description must carry, any of them, for the kind of company when no industry value names it, e.g. "fintech" or "agency". Each is a lowercase word or short phrase, matched as a whole word with its plural; a term with * or ? is refused. At most 25.
countryarray- People location, country level: countries the person lives in. Each value is a lowercase full country name, e.g. "germany" or "united states".
currentCompanyNamearray- Company: the person's current company, by name, e.g. "stripe".
currentCompanyWebsitearray- Company: the person's current company, by website domain, e.g. "stripe.com".
currentJobTitlearray- People: exact current job titles in lowercase, e.g. "vp of sales". Use keywords to match part of a title.
currentRevenuearray- Company: annual revenue band of the person's current company. Values: $0-$1M, $1M-$10M, $10M-$25M, $25M-$50M, $50M-$100M, $100M-$250M, $250M-$500M, $500M-$1B, $1B-$10B, $10B+.
degreearray- People: degrees the person holds, in lowercase, e.g. "bachelors", "masters" or "master of business administration". Use include entries only.
departmentarray- Company: the person's department at their current company. Values: advisory, analyst, creative, education, engineering, finance, fulfillment, health, hospitality, human_resources, legal, manufacturing, marketing, operations, partnerships, product, professional_service, public_service, research, sales, sales_engineering, support, trade, unemployed.
descriptionstring- A sentence or two on who the ICP targets. Left out, the current one stays; an empty string removes it.
excludedCompanyKeywordsarray- Company: words that remove a company whose name or description carries any of them, e.g. "agency". Same matching as companyKeywords. At most 25.
excludedKeywordsarray- People: words that remove a person whose current job title matches any of them, e.g. "intern". Same matching as keywords. At most 25 terms, and a term needs text other than the wildcards * and ?.
fullNamestring- People: the person's full name in lowercase, e.g. "jane doe".
fundingRaisedobject- Company: total funding the person's current company has raised, in US dollars, as { min, max }, both required.
hqCountryarray- Company location, country level: countries where the person's current company has its headquarters. Each value is a lowercase full country name, e.g. "germany" or "united states".
industryarray- People: the person's own industry (Person Industry in the app), in lowercase, e.g. "computer software". It is not the industry of their company: for the industries the ICP targets, use jobCompanyIndustry.
isOpenToWorkboolean- People: true keeps only people whose LinkedIn profile shows Open to Work. Leave it out otherwise.one of
true isUnemployedboolean- People: true keeps only people who hold no current job. Leave it out otherwise.one of
true jobCompanyIndustryarray- Company: the industry of the person's current company (Company Industry in the app). Use it for the industries the ICP targets. Same lowercase values as industry, e.g. "computer software".
jobCompanyLocationRegionarray- Company location, state level: states or regions where the person's current company has its headquarters, in lowercase, e.g. "texas".
keywordsarray- People: words to look for in the person's current job title, e.g. "growth". A bare word matches anywhere in the title; a term with * or ? must match the whole title, e.g. "manager*" for titles starting with manager. Matching ignores case.
languagesSpokenarray- People: languages the person speaks, as lowercase names, e.g. "german".
levelarray- People: seniority of the person's current job. Values: cxo, vp, director, manager, senior, entry, owner, partner, training, unpaid.
linkedinUrlstring- People: the person's LinkedIn profile URL, e.g. "linkedin.com/in/jane-doe".
locationarray- Company location, city level: cities where the person's current company has its headquarters. Write each as "city, state, country" in lowercase, e.g. "austin, texas, united states", or as "city, country" for a place with no state. A bare place name is refused: use hqCountry for a whole country and jobCompanyLocationRegion for a state.
locationRegionarray- People location, state level: states or regions the person lives in, in lowercase, e.g. "bavaria" or "california".
numbersOfMonthInCurrentRolenumber- People: number of months the person has been in their current role.
pastCompanyNamearray- Past job: companies the person worked at before, by name, e.g. "oracle".
pastCompanyWebsitearray- Past job: companies the person worked at before, by website domain, e.g. "oracle.com".
pastJobDepartmentarray- Past job: department of a job the person held before. Same values as department.
pastJobLevelsarray- Past job: seniority of a job the person held before. Same values as level.
pastJobTitlearray- Past job: exact job titles the person held before, in lowercase, e.g. "account executive".
personalLocationarray- People location, city level: cities the person lives in. Write each as "city, state, country" in lowercase, e.g. "munich, bavaria, germany", or as "city, country" for a place with no state. A bare place name is refused: use country for a whole country and locationRegion for a state.
schoolNamearray- People: schools the person attended, in lowercase, e.g. "stanford university".
sizearray- Company: headcount band of the person's current company. Values: 1-10, 11-50, 51-200, 201-500, 501-1000, 1001-5000, 5001-10000, 10001+.
skillsarray- People: skills on the person's profile, in lowercase, e.g. "python".
totalYearsOfExperienceobject- People: the person's total years of work experience, as { min, max }, both required.
fuseaimcp_update_inbox_visibility_preferencesSet whether the email/LinkedIn inboxes show only campaign-contact conversations: hideNonCampaignEmails for the email inbox and/or hideNonCampaignLinkedInMessages for the LinkedIn inbox (true hides non-campaign conversations, false shows them).DestructiveUpdate Inbox Visibility Preferences
Set whether the email/LinkedIn inboxes show only campaign-contact conversations: hideNonCampaignEmails for the email inbox and/or hideNonCampaignLinkedInMessages for the LinkedIn inbox (true hides non-campaign conversations, false shows them). Each setting passed replaces its current value; one left out keeps its value, and a call with neither changes nothing. This only filters what the user sees in FuseAI: hidden conversations are not deleted, nothing changes in the user's connected mailboxes or LinkedIn accounts, turning a setting off shows them again, and the same filters apply to list_emails and list_linkedin_chats while they are on.
Inputs
hideNonCampaignEmailsboolean- No description.
hideNonCampaignLinkedInMessagesboolean- No description.
fuseaimcp_update_knowledge_hub_profileUpdate an existing Knowledge Hub profile, identified by its profile id: its name, source URLs, extra info (optionally with a file attachment), and its learning, modelEffort and refreshEnabled settings.DestructiveUpdate Knowledge Hub profile
Update an existing Knowledge Hub profile, identified by its profile id: its name, source URLs, extra info (optionally with a file attachment), and its learning, modelEffort and refreshEnabled settings. If it is the team's profile, the change applies for everyone on the team. Source URLs passed are scraped from the web during the call, and pages built from changed sources are regenerated, replacing manual edits to them. A source URL or extra-info entry matching a stored one (by URL, _id or filename) replaces it, and a new attachment deletes the old file; entries cannot be removed here. Turning learning off removes the learned content. At modelEffort 'high', weekly refresh (re-scrapes the sources) and learning (reads the team's emails, LinkedIn messages and call transcripts stored in Fuse) each bill credits monthly while on; 'fast' is free.
Inputs
idstringrequired- The Knowledge Hub profile's _id.
additionalInfoarray- No description.
learningobject- No description.
modelEffortstring- No description.one of
fasthigh profileNamestring- No description.
refreshEnabledboolean- No description.
rootUrlsarray- No description.
fuseaimcp_update_linkedin_account_settingsUpdate one LinkedIn account's settings (by accountId), replacing the current values for everyone who uses the account: isActive (false holds its queued campaign LinkedIn steps and can stop a campaign left without a LinkedIn sender; true lets them send again), title and signature (filled into {Sender Job Title} and {Sender Signature} in campaign LinkedIn messages written from then on), dailyUsageLimits ({invitations, messages}, optional profileVisits, reactions, inmails) / weeklyUsageLimits ({invitations, messages}, optional inmails), which set how many invitations, messages and InMails the account sends per day and week (values above the account's tier maximum are rejected), and inmailMonthly ({budget}, null for no monthly cap).DestructiveUpdate LinkedIn Account Settings
Update one LinkedIn account's settings (by accountId), replacing the current values for everyone who uses the account: isActive (false holds its queued campaign LinkedIn steps and can stop a campaign left without a LinkedIn sender; true lets them send again), title and signature (filled into {Sender Job Title} and {Sender Signature} in campaign LinkedIn messages written from then on), dailyUsageLimits ({invitations, messages}, optional profileVisits, reactions, inmails) / weeklyUsageLimits ({invitations, messages}, optional inmails), which set how many invitations, messages and InMails the account sends per day and week (values above the account's tier maximum are rejected), and inmailMonthly ({budget}, null for no monthly cap).
Inputs
accountIdstringrequired- No description.
dailyUsageLimitsobject- No description.
inmailMonthlyobject- No description.
isActiveboolean- No description.
signaturestring- No description.
titlestring- No description.
weeklyUsageLimitsobject- No description.
fuseaimcp_update_linkedin_chat_readMark a LinkedIn chat as read on the connected LinkedIn account itself, not only in FuseAI (set isRead to true; marking a chat unread is not supported, so this cannot be undone here).DestructiveMark LinkedIn Chat as Read
Mark a LinkedIn chat as read on the connected LinkedIn account itself, not only in FuseAI (set isRead to true; marking a chat unread is not supported, so this cannot be undone here). It then re-syncs the chat from LinkedIn: a reply or hand-sent message FuseAI has not processed yet is handled as usual, which can stop that contact's campaign sequence, draft, schedule or discard an AI reply, send reply notifications and start the user's reply-triggered automations.
Inputs
accountIdstringrequired- No description.
chatIdstringrequired- No description.
isReadbooleanrequired- No description.one of
true
fuseaimcp_update_list_cell_toolSets custom-column cell values for rows in a list, replacing each listed cell's current value (smart-column AI values included); an empty string clears the cell.DestructiveUpdate List Cell
Sets custom-column cell values for rows in a list, replacing each listed cell's current value (smart-column AI values included); an empty string clears the cell. Campaigns that use the list read these cells when they prepare each contact's next message ({Column: <name>} tokens insert the value as written, and AI-written steps use every column as context), so an edit changes what those contacts are sent, and filling an empty or failed writer smart-column cell lets a held step send. Edits made with this tool do not trigger the user's 'Smart column updated' automations. Provide listId, columnId, and columnValues entries keyed by the row's entity — [{ contact: '<contactId>', value }] or [{ company: '<companyId>', value }] — plus optional isEdited (smart columns only: marks the change as a manual edit and keeps the AI's original value). Each entry names exactly one of contact, company or customContact (an older alias for contact). value is the cell's text, number, true/false or date; for a link column, value is the URL and label (required) its text. Optional reasoning (contact rows only, up to 1000 characters) is a note explaining the value: it is saved with the cell, get_smart_column_provenance_tool returns it when the cell has no AI research record, and campaigns do not read it.
Inputs
columnIdstringrequired- No description.
columnValuesarrayrequired- No description.
listIdstringrequired- No description.
isEditedboolean- No description.
fuseaimcp_update_scheduled_email_toolEdit a queued send's subject, body, send time or cc/bcc (each list replaces the current one).DestructiveUpdate Scheduled Email
Edit a queued send's subject, body, send time or cc/bcc (each list replaces the current one). The edited message goes to the lead and every cc/bcc address at its send time (a draft awaiting approval still needs approval), and later AI-written steps for this lead build on the new wording. Sent, stopped and task-owned sends cannot be edited. The edit is stored as a replacement row, so the response carries a NEW scheduledEmailId that later calls must use.
Inputs
scheduledEmailIdstringrequired- No description.
bccarray- No description.
bodystring- No description.
ccarray- No description.
scheduledDateUTCstring- No description.
subjectstring- No description.
fuseaimcp_update_sequence_step_toolUpdates a single sequence step BY ITS ID: its content (template, topic), wait days, sending mailboxes, cc, threading, channel, step type, execution mode, LinkedIn and task settings, a condition's window or an A/B split's weights.DestructiveUpdate Sequence Step
Updates a single sequence step BY ITS ID: its content (template, topic), wait days, sending mailboxes, cc, threading, channel, step type, execution mode, LinkedIn and task settings, a condition's window or an A/B split's weights. Each field passed replaces the stored value; omitted fields are kept. It also works on a running campaign: new content and cc reach contacts not yet scheduled for the step, a threading change also applies to its queued emails, a new wait reschedules the step's queued messages, and a new mailbox list moves its queued emails off the removed mailboxes onto the listed ones and rewrites them for the new sender, replacing any hand edits and charging credits for AI work not yet paid for. A new channel, step type, execution mode, topic, invitation note or voice note likewise reaches contacts not yet scheduled for the step; it does not convert messages already queued or tasks already open. sequenceStepId is the step's 24-hex MongoDB id. Fields: template is {subject, body, format ('editor' or 'html')} and cannot be set on an AI-written step. sequenceStepType 'manual' uses that template; 'ai_generated' or 'ai_reply' has AI write each contact's message when the contact is scheduled for the step, costing credits per message (the campaign pauses if credits run out). topic is what AI writes about on an AI-written step; changing it there also regenerates the step's title. channel can move the step to email, linkedin_connection, linkedin_message, linkedin_visit, linkedin_like, inmail, dialer or todo if the sequence's order rules still hold (for example, a LinkedIn connection before any LinkedIn message); no step becomes or stops being a condition (channel 'condition' is accepted only on a step that already is one, where it changes nothing), the channel cannot change once a campaign with a condition has started, and leaving linkedin_message drops the voice note. executionMode 'automatic' has FuseAI send or perform the step through the user's mailbox or LinkedIn account; 'manual' makes it a task for a person, and the contact waits until the task is done or skipped (dialer and todo steps are always manual, linkedin_visit and linkedin_like always automatic). invitationText (linkedin_connection, up to 200 characters; empty removes it) is the note sent with the connection request. voiceNoteScript (automatic linkedin_message steps, up to 100 words, about 45 seconds; empty or null removes it) adds a voice note in the cloned voice of the sending account's owner or the campaign owner; each one LinkedIn accepts costs credits. reactionType (linkedin_like: like, celebrate, support, love, insightful, funny, or auto to choose from the post), postRecencyDays (linkedin_like, 1-365: the reaction goes on the contact's newest post from that many days back, and the step is skipped if there is none) and inmailExhaustionBehavior (inmail: 'skip', the default, skips the InMail when the LinkedIn account has no InMail credits left; 'defer' waits once for the next monthly refill) are read at send time, so they also apply to queued messages. replyInSameThread (email and inmail): true replies in the contact's existing thread, false starts a new thread with the step's subject, null resets it to true; a step done by a person always starts a new thread. cc lists the addresses copied on the step's emails. On a step done by a person, autoSkipAfterDays (1-365; null waits forever) is how long an unstarted task stays open before it expires and the contact moves on, and taskInstructions and taskLinkUrl (http or https) are shown on the task; all three also update the step's open tasks, so a shorter autoSkipAfterDays can expire them. channelSequenceNumber is the step's place among steps of its channel (0 for the first), which AI-written steps use to write an opener or a follow-up; the backend recomputes it when steps are added, removed, reordered or change channel. On a condition step only daysBetween, its window of 1-30 days, can change; on an A/B split (conditionType 'ab_split') only splitArms can change, at any time and also on a running campaign: send both arms ('a' then 'b') with whole-percent weights adding up to 100 and optional names. On a started campaign, contacts the split already routed who were not yet sent anything in their arm then follow the new weights; contacts their arm already contacted, or who replied, bounced or unsubscribed, keep their arm. A step's position moves through reorder_sequence_steps_tool. To choose an email step's sending mailboxes, pass allowedInboxes built from list_email_accounts; it replaces the step's whole list. When editing email content, pass BOTH subject and body together (nested under template); the backend replaces the step's template, so sending only one would drop the other. Only these personalization tokens are substituted (single braces, exact names): {Contact Full Name}, {Contact First Name}, {Contact Last Name}, {Contact Company}, {Contact Job Title}, {Contact Industry}, {Sender Full Name}, {Sender First Name}, {Sender Last Name}, {Sender Job Title}, {Sender Signature}. The {Sender …} tokens take the name, title and signature of the mailbox or LinkedIn account that sends the step; the campaign's sender name and job title fill in only when that account has none. Two more token kinds are resolved per contact: {Column: <list column name>} inserts the contact's value from that column of the campaign's lists, and {{<instruction>}} has AI write that part for each contact from their data, which costs credits for each message that uses it ({{<instruction> :: <original text>}} rewrites existing words, keeping them as the fallback). Anything else is sent to the prospect literally.
Inputs
sequenceStepIdstringrequired- No description.
allowedInboxesarray- Email steps only: the user's mailboxes this step sends from; each contact is sent from one of them. Build every entry from list_email_accounts, using only accounts with isActive and isConnected true that are the user's own (isOwner) or have USE_IN_CAMPAIGN in sharedPermissions. Entries are not checked on save and the email is sent from inboxId as given. Replaces the step's whole list: omit to keep it, or pass [] to let the backend pick among all the user's usable mailboxes again.
autoSkipAfterDaysinteger- No description.
ccarray- No description.
channelstring- No description.one of
emaillinkedin_connectionlinkedin_messagelinkedin_visitlinkedin_likeinmaildialertodocondition channelSequenceNumberinteger- No description.
daysBetweeninteger- No description.
executionModestring- No description.
inmailExhaustionBehaviorstring- No description.
invitationTextstring- No description.
postRecencyDaysstring- No description.
reactionTypestring- No description.
replyInSameThreadboolean- No description.
sequenceStepTypestring- No description.one of
ai_generatedai_replymanual splitArmsarray- ab_split steps only: both arms, arm 'a' then arm 'b', each { key, label?, weight }. Weights are whole percents adding up to 100: each contact reaching the split is routed to an arm in these weights' proportions, by a fixed hash of the contact and the split, and stays in it. They can change at any time, also on a running campaign: contacts already routed but not yet sent anything in their arm then follow the new weights.
taskInstructionsstring- No description.
taskLinkUrlstring- No description.
templatestring- No description.
topicstring- No description.
voiceNoteScriptstring- No description.
fuseaimcp_update_sequence_toolEdits the sequence of an EXISTING DRAFT campaign, like the campaign editor: add a step, update an existing step (fill a manual campaign's empty placeholder steps, rewrite subject/body, change email threading — on one step or on every email step at once, ai-generated campaigns included — or change wait days), or delete a step.DestructiveUpdate Campaign Sequence
Edits the sequence of an EXISTING DRAFT campaign, like the campaign editor: add a step, update an existing step (fill a manual campaign's empty placeholder steps, rewrite subject/body, change email threading — on one step or on every email step at once, ai-generated campaigns included — or change wait days), or delete a step. Pick the action with the `action` argument. Steps are addressed by 1-based position along the main sequence and the campaign by id or fuzzy name; no step ids needed. Without campaignId, when campaignName fits more than one draft the call edits only one, without asking: a draft with exactly that name, or else one picked by fuzzy match; the result names the campaign it edited. Both the id and the name lookup match only campaigns that were never launched: a campaign that was launched, whether running, paused or stopped, is refused and left unchanged. It changes only the stored draft in FuseAI and sends nothing; what the subject, body and threading fields say about sending applies once the campaign is launched separately, for example with update_campaign_status_tool. After launch, {Column: <list column name>} tokens in a subject or body insert each contact's value from that column of the campaign's lists, and {{<instruction>}} tokens have AI write that part for each contact, which costs credits for each message that uses them. add_step and delete_step are not safe to retry: repeating an add_step can add a duplicate step, and because a delete moves every later step up one place, repeating a delete_step deletes whichever step moved into that position; check the steps with get_campaign_sequence_steps_tool before retrying. In a branched campaign the steps inside a condition's lanes are not counted: edit them by id with update_sequence_step_tool, and note that a step added at the end of a main sequence closed by a condition goes just before that condition, and that deleting a condition step also deletes both of its lanes and every step in them. It cannot change which mailboxes send a step; use update_sequence_step_tool (allowedInboxes) for that. Do NOT use for creating a brand-new campaign.
Inputs
actionstringrequired- What to do to the campaign's sequence. "add_step" = append/insert a brand-new step (requires channel). "update_step" = change an EXISTING step — fill a manual campaign's empty placeholder step, rewrite its subject/body, change its email threading, or change its wait days (requires position, except a threading-only change which may omit position to hit all email steps). "delete_step" = remove an existing step (requires position).one of
add_stepupdate_stepdelete_step bodystring- Message body. For add_step it is REQUIRED for email and linkedin_message and must be omitted for linkedin_connection / dialer. For update_step, omitting body leaves the current body unchanged; you cannot edit the body of an ai-generated step. === IF the step is "email": HTML ONLY === Output well-formed HTML. Never raw markdown, never plain text with newline characters. Conversion rules: - Wrap each paragraph in <p>...</p>. - Use an empty <p></p> ONLY where you intentionally want a visible blank line (e.g. before/after the greeting, before the signature, between major sections). Email clients collapse <p> margins, so an empty <p></p> is the way to render a real visual gap. Do NOT insert <p></p> between every paragraph by default — only where a blank line genuinely belongs. - Single newlines inside a paragraph become <br>. - **text** or __text__ becomes <b>text</b>; *text* or _text_ becomes <i>text</i>. - [label](https://url) becomes <a href="https://url">label</a>. - Lines starting with '- ' or '* ' become <ul><li>...</li></ul>; lines starting with '1. ' become <ol><li>...</li></ol>. - Lines starting with '# ' / '## ' / '### ' become <h1> / <h2> / <h3>. - Personalization: only these tokens are substituted (single braces, exact names): {Contact Full Name}, {Contact First Name}, {Contact Last Name}, {Contact Company}, {Contact Job Title}, {Contact Industry}, {Sender Full Name}, {Sender First Name}, {Sender Last Name}, {Sender Job Title}, {Sender Signature}. The {Sender …} tokens take the name, title and signature of the mailbox or LinkedIn account that sends the step; the campaign's sender name and job title fill in only when that account has none. Unknown tokens like {firstName} or {{companyDomain}} are sent to the prospect literally. Never emit raw asterisks, underscores, or markdown. - Email example: input 'Hi {Contact First Name},\n\nLoved your **post** on LinkedIn.\n\nBest,\nNico' → '<p>Hi {Contact First Name},</p><p></p><p>Loved your <b>post</b> on LinkedIn.</p><p></p><p>Best,<br>Nico</p>'. === IF the step is "linkedin_message": PLAIN TEXT ONLY === LinkedIn does not render HTML. The recipient sees whatever you send literally, tags and all. - Do NOT wrap in <p>, do NOT use <br>, do NOT use <b>/<i>/<a>/<ul>/<ol>/<h1> etc. Any HTML tag will appear as raw characters in the recipient's inbox. - Use real newline characters (\n) for line breaks. Use a blank line (\n\n) for paragraph spacing. - Do NOT use markdown either — no **bold**, no *italic*, no [links](url). LinkedIn renders all of those as literal characters too. Just write the words. - Personalization: only these tokens are substituted: {Contact Full Name}, {Contact First Name}, {Contact Last Name}, {Contact Company}, {Contact Job Title}, {Contact Industry}, {Sender Full Name}, {Sender First Name}, {Sender Last Name}, {Sender Job Title}, {Sender Signature}. The {Sender …} tokens take the name, title and signature of the mailbox or LinkedIn account that sends the step; the campaign's sender name and job title fill in only when that account has none. Unknown tokens are sent literally. - LinkedIn example: input 'Hey {Contact First Name}, loved your post on AI.\n\nWould you be open to a 15 min chat next week?' → 'Hey {Contact First Name}, loved your post on AI.\n\nWould you be open to a 15 min chat next week?' (sent as-is, no tags). === IF the step is "linkedin_connection" or "dialer": NO BODY === Do not pass a body. These steps are configured without message content.
campaignIdstring- The id of the campaign to modify (a 24-hex MongoDB id). ALWAYS pass this when editing a campaign you just created — it targets that exact campaign and avoids matching a different campaign with the same name. Omit it (and rely on campaignName) only when the user references an existing campaign you have no id for.
campaignNamestring- The name of the existing campaign to modify, resolved via fuzzy match. Required when campaignId is not provided; ignored when campaignId is present (the id is authoritative).
channelstring- REQUIRED for action "add_step": the channel of the new step — "email", "linkedin_message", "linkedin_connection" (one per sequence; must come before any LinkedIn messages), or "dialer" (multiple allowed). Ignored for update_step / delete_step (the existing step's own channel is used).one of
emaillinkedin_messagelinkedin_connectiondialer daysBetweeninteger- Number of days to wait before this step. For add_step it defaults to 1 when omitted. For update_step, omitting it leaves the current value unchanged; setting it requires a position.
positioninteger- 1-based position along the main sequence, counting ALL of its steps (not just emails); the steps inside a condition's lanes are not counted. For "add_step": optional — where to insert the new step; omitted = appended at the end. For "update_step": REQUIRED when changing subject/body/daysBetween; may be omitted ONLY for a threading-only change, which then applies to every email step in the campaign (non-email steps are left untouched, so it is safe on mixed sequences). For "delete_step": REQUIRED; the step to remove. Deleting a condition step also deletes both of its lanes and every step in them.
replyInSameThreadboolean- Email threading — only valid for email steps; omit (null) for every other channel. true (the backend default) = the email is sent as a reply inside the existing thread and inherits the thread's subject. false = the email starts a NEW thread using its own subject — provide a subject in that case. Set false ONLY when the user signals separate threads: they explicitly ask ("different threads", "don't send follow-ups as replies") or they supply a distinct subject for each email. An explicit instruction always wins over the subjects heuristic. On update_step you can set this with no position to flip ALL email steps at once — the only way to change threading on an ai-generated campaign (whose step CONTENT is not editable).
subjectstring- Email subject line — only for email steps (add_step or update_step). Plain text, no HTML/markdown. Send-time semantics: the FIRST email step starts the thread with this subject. A follow-up email with replyInSameThread true (or omitted) is sent as a reply inside the existing thread and INHERITS that thread's subject — its own subject is ignored. A follow-up with replyInSameThread false starts a NEW thread with this subject (if empty, it falls back to the first email step's subject). On update_step, omitting subject leaves the step's current subject unchanged.
fuseaimcp_update_signal_monitor_draftOverwrite a draft monitor's creationConfig so you can finish or adjust it before activating.DestructiveUpdate signal-monitor draft
Overwrite a draft monitor's creationConfig so you can finish or adjust it before activating. Replaces the whole saved config — include every field the draft should keep.
Inputs
creationConfigobjectrequired- The full edited creation config object to overwrite the draft with (same fields the draft's monitor type was originally created with).
documentIdstringrequired- The signal monitor's documentId.
fuseaimcp_update_signal_monitor_source_listsReplace the source prospect/company lists feeding a people-, companies- or job-changes-watcher monitor, by its documentId.DestructiveUpdate signal-monitor source lists
Replace the source prospect/company lists feeding a people-, companies- or job-changes-watcher monitor, by its documentId. A background job then stops watching the contacts or companies no longer in the lists and starts watching new ones at the external data provider. Newly watched entries cost credits (a setup charge per entry, or per check), and the update can be refused when the balance cannot cover the setup charge.
Inputs
documentIdstringrequired- The signal monitor's documentId.
listIdstring- New single source list id. Provide listId OR listIds, not both.
listIdsarray- New source list ids. Provide listId OR listIds, not both.
fuseaimcp_update_signal_monitor_statusSet an activated (non-draft) signal monitor's status to active or paused, by its documentId.DestructiveUpdate signal-monitor status
Set an activated (non-draft) signal monitor's status to active or paused, by its documentId. Pausing stops its watches at the external data provider, or its scheduled scrapes; resuming restarts them and their credit charges, so new matches are again added to its list, where they can update existing contacts, trigger the user's automations and be picked up by any campaign that refreshes from that list. An expired monitor can only be paused, a Google Maps search cannot be paused or resumed, and drafts are started with activate_signal_monitor.
Inputs
documentIdstringrequired- The signal monitor's document id.
statusstringrequired- Set the monitor active or paused.one of
activepaused
fuseaimcp_update_slack_settingsUpdate Slack integration settings — replace the channel that notifications post to, on the user's own Slack connection.DestructiveUpdate Slack Settings
Update Slack integration settings — replace the channel that notifications post to, on the user's own Slack connection. FuseAI's automatic Slack notifications, automation Slack steps and send_slack_message calls that name no channel then post there; a team owner's channel is also the default for teammates who post with the owner's Slack. The name is saved without being checked with Slack. A member without one posts with the team owner's, which only the owner changes (403 INSUFFICIENT_PERMISSION).
Inputs
channelstringrequired- The Slack channel notifications post to (e.g. #sales)
fuseaimcp_update_taskUpdate an open task by id: reassign it to a teammate via assignedToUserId (from list_team_members; the change sticks through later system refreshes), or change its title, instructions, link, dueAt, priority, pin or in-progress state.DestructiveUpdate Task
Update an open task by id: reassign it to a teammate via assignedToUserId (from list_team_members; the change sticks through later system refreshes), or change its title, instructions, link, dueAt, priority, pin or in-progress state. On a campaign step task whose step auto-skips after a set number of days, changing dueAt also changes when Fuse skips that step and moves the contact on to its next campaign step, which can email or message the contact; a to-do marked in progress is not auto-skipped. Group rows, such as a sequence step batch or a call batch, cannot be re-dated, and only call batches among them can be reassigned. Does NOT complete the task; use complete_task. Raising priority to high or urgent may notify the assignee by text if they turned on task alerts.
Inputs
taskIdstringrequired- The task id (24-character Mongo id) from list_tasks.
assignedToUserIdstring- Teammate user id (from list_team_members) to reassign the task to. Must be the team owner or an active member of the caller's team, else TASK_ASSIGNEE_NOT_ALLOWED. A manual reassignment sticks through later system refreshes. Group rows cannot be reassigned, except call batches.
dueAtstring- New ISO 8601 due date-time.
instructionsstring- New instructions; null or empty clears them.
isInProgressboolean- Mark a to-do in progress (true) or not (false); ignored for other task types.
isPinnedboolean- Pin (true) or unpin (false); pinned tasks float to the top of their bucket.
linkUrlstring- New http(s) link; null or empty clears it.
prioritystring- New priority.one of
lowmediumhighurgent titlestring- New title.
fuseaimcp_update_team_memberChange a joined team member's role and/or credits cap, replacing the current values.DestructiveUpdate Team Member
Change a joined team member's role and/or credits cap, replacing the current values. Pass at least one of role / creditsCap. Owner or admin only; the team owner and pending invitations cannot be updated. Passing role, even an unchanged one, signs the member out of their FuseAI sessions. creditsCap cannot exceed the team's credits for the current billing period.
Inputs
memberIdstringrequired- The team member's user id (a 24-hex id)
creditsCapinteger- No description.
rolestring- No description.one of
ADMINMEMBER
fuseaimcp_update_user_companyRename the user's own company or change its website (the company on the user's FuseAI account, not a saved or CRM company record).DestructiveUpdate User Company
Rename the user's own company or change its website (the company on the user's FuseAI account, not a saved or CRM company record). This is the company name the whole team sees in the app. The app shows every member the team owner's company, so only the owner's change shows for everyone; a member's change updates just their own profile. Pass companyName to rename the company, companyDomain to change its website, or both; a call with neither changes nothing. Changing companyDomain also clears the user's Home page recommendations and rebuilds them in a background job, which looks up the new domain and the user's email with an outside data provider, may fetch that website, and uses AI to write the new recommendations; no credits are charged.
Inputs
companyDomainstring- The company website domain
companyNamestring- No description.
fuseaimcp_update_website_allowed_domainsUpdate the team's website-tracking config (saved domains + tracked-page labels).DestructiveUpdate Website Tracking Config
Update the team's website-tracking config (saved domains + tracked-page labels). `domain` takes the team's website domains and replaces the whole saved-domain list (a removed domain drops out of the per-domain tag status, and its verification record is deleted at the next verification); `trackedPages` takes free-text page labels such as pricing, not URL paths, and when sent replaces the tracked-page list (omit it to keep the current one); the credit cap is not changed. Saved domains are used only for domain verification and the per-domain tag status. Tracked-page labels are not matched against URLs: the FuseAI app shows a per-page copy of the tag for each label, visits from a page carrying that copy are reported under its label, and an installed copy keeps reporting after its label is removed. Neither list is an allowlist: they do not limit which visits the tag records, which visitors are identified or which visits are charged credits, and this call does not fetch or check the sites. Confirm with the user before calling it. Only the team's owner and admins may (403 INSUFFICIENT_PERMISSION). If the team has no website-tracking setup yet, an admin's call creates it with these values, which registers the team for FuseAI's background visitor identification and website-intent credit charges: visitors to the team's website are then looked up hourly with Fiber, an outside people-data provider, and the people identified are added to the CRM's website-visitor lists, where each new person can trigger a Slack visitor alert and website-visitor automations. A call by the owner (or by a user on no team) in that state fails with 404 USER_ENTRY_NOT_FOUND.
Inputs
domainarrayrequired- The full set of allowed tracking domains (this OVERWRITES the current list).
trackedPagesarray- Optional free-text page labels, such as pricing, not URL paths; when sent, replaces the tracked-page list.
fuseaimcp_update_website_credits_limitSet the team's website-intent monthly credit cap.DestructiveSet Website Credits Limit
Set the team's website-intent monthly credit cap. It replaces the current cap, a spending limit on how many newly processed visits each billing period are charged credits and shown in visitor reports; lowering it, or setting 0, stops new charged visits once the cap is reached. It does not stop FuseAI's free background visitor identification, which can still make visits visible and add the people it identifies to the CRM's website-visitor lists, where each new person can trigger a Slack visitor alert and website-visitor automations. `amount` must be a whole number from 0 up to the plan's credit allowance (400 EXCEEDS_SUBSCRIPTION_LIMIT); null removes the custom cap, so the plan's default cap applies. If the team has no website-tracking setup yet, the call creates it first, which registers the team for FuseAI's background visitor identification and website-intent credit charges. Confirm with the user before calling it. Only the team's owner and admins may (403 INSUFFICIENT_PERMISSION).
Inputs
amountnumberrequired- The new monthly website-intent credit cap (null unsets the cap).
fuseaimcp_update_writing_styleUpdate a writing style by writingStyleId: rename it, replace the rules[] (tone, message length, formatting — one instruction per rule; the new list replaces every existing rule, learned ones included), switch style learning on or off and choose its sources (learning runs on the default style only), or set modelEffort.DestructiveUpdate Writing Style
Update a writing style by writingStyleId: rename it, replace the rules[] (tone, message length, formatting — one instruction per rule; the new list replaces every existing rule, learned ones included), switch style learning on or off and choose its sources (learning runs on the default style only), or set modelEffort. Campaigns that use this style (the default style covers campaigns with none selected) write every AI message and AI reply generated afterwards with these rules, including for contacts already in a sequence. While learning is on, a daily background job adds rules learned from the user's sent emails and LinkedIn messages stored in FuseAI; with modelEffort 'high' this costs 750 credits per enabled source every 30 days, and 'fast' costs none.
Inputs
writingStyleIdstringrequired- No description.
learningobject- No description.
modelEffortstring- No description.one of
fasthigh namestring- No description.
rulesarray- No description.
fuseaimcp_update_writing_style_from_textReplace all of a writing style's rules, learned ones included, with rules AI-extracted from pasted text (by writingStyleId).DestructiveUpdate Writing Style From Text
Replace all of a writing style's rules, learned ones included, with rules AI-extracted from pasted text (by writingStyleId). The extraction runs as a background job: the style keeps its current rules until it finishes, rule edits made meanwhile are overwritten, and only one extraction per style can run at a time. Check list_writing_styles until extractionStatus is completed or failed (failed leaves the rules unchanged and sets extractionError). Campaigns that use this style then write every AI message and AI reply generated afterwards with the new rules, including for contacts already in a sequence.
Inputs
textstringrequired- No description.
writingStyleIdstringrequired- No description.
fuseaimcp_upsert_company_by_domainCreate the canonical CRM company record for a website domain when none exists, or update it when it does.DestructiveUpsert Company By Domain
Create the canonical CRM company record for a website domain when none exists, or update it when it does. Pass `domain` plus at least one of `seed` (name, industry, size, address, linkedIn, headQuartersCountry; applied only when the company is created) or `fields` (allowed keys: fiberDetails, industry, description, address, headQuartersCountry, linkedIn, companyTwitterUrl, linkedInLogoUrl, size, yearFounded, companyType; applied on every call, overwriting the stored values of an existing record). Optional `nameFromEnrichment` replaces a placeholder name that still equals the domain. Returns { data: { created, _id, domain } } where data._id is the company id. This writes only the company record that all FuseAI accounts share for this domain, so a change shows for every account that has this company; it does not mark the company as saved for the user. The record is part of FuseAI's own company catalog: the call itself reaches no outside service and starts no job or automation, but running campaigns use these company details when writing AI messages to prospects at this company, in any account, so an update changes what those prospects are sent.
Inputs
domainstringrequired- No description.
fieldsobject- No description.
nameFromEnrichmentstring- No description.
seedobject- No description.
fuseaimcp_upsert_crm_contactFind or create a CRM contact for the user and return its id.DestructiveUpsert CRM Contact
Find or create a CRM contact for the user and return its id. Matches an existing contact by LinkedIn first (linkedInMemberId, LinkedIn's numeric member id, when given; then the linkedIn URL; then the person's older URLs in linkedInAliases), then by email, then by firstName + lastName at companyDomain; a match is attached to the user and reused (never duplicated), and one found by the member id or an older URL has its stored LinkedIn URL and member id replaced with the given ones (a shared contact changes for everyone who has it). When nothing matches, the contact is created (firstName required then; jobTitle, companyDomain and companyName seed the record; a given email is stored as the primary email), except that a teammate's contact with the same LinkedIn URL or member id can be reused instead and take the given name, job title and company (a teammate's custom contact found only by member id is not reused, so a second contact can be created). Pass listId (24-hex list id) to also add the contact to that list; a contact new to the list starts the user's automations triggered by additions to that list, and an active campaign that enrolls the list's new members can start messaging the person. Returns { contactId, created } where created is true when none of the lookups above matched (normally a new contact; a teammate's contact reused at the create step also answers true).
Inputs
companyDomainstring- No description.
companyNamestring- No description.
emailstring- No description.
firstNamestring- No description.
jobTitlestring- No description.
lastNamestring- No description.
linkedInstring- No description.
linkedInAliasesarray- No description.
linkedInMemberIdstring- No description.
listIdstring- No description.
fuseaimcp_wake_automation_executionEnd the wait an execution is parked on, now: its wait step's recheck runs again and the run goes on without waiting for the target (the wait's cancel rule still applies).DestructiveWake Automation Execution
End the wait an execution is parked on, now: its wait step's recheck runs again and the run goes on without waiting for the target (the wait's cancel rule still applies). A wait inside a forEach parks each item on its own: laneKey (an item's laneKey from get_automation_execution) resumes that one item; without it every parked item of the step resumes. Only a run with status waiting can be woken (EXECUTION_NOT_WAITING otherwise), and a laneKey must name an item still parked (EXECUTION_ITEM_NOT_WAITING otherwise). The run's later steps then go ahead, including any that send messages or call the user's connected apps; the recheck and those steps may cost credits. If the rule is no longer active, the run is cancelled instead of resumed.
Inputs
automationIdstringrequired- No description.
executionIdstringrequired- No description.
laneKeystring- No description.
No tools match.