Skip to content
Scalekit Docs
Talk to an Engineer Dashboard

Connect AI agents to the Wistia MCP server

Vendor MCP
Open markdown

The Wistia MCP connector routes your AI agent's tool calls to Wistia's own MCP server through Scalekit. Each user signs in to Wistia once, and Scalekit stores and refreshes their tokens, so your agent never handles credentials. It comes with 183 tools.

Tools
183
What they doRead · write · destructive
82 · 64 · 3782 read64 write37 destructive
Users sign in with
OAuth app
Your own Wistia MCP server app

Setup

  1. Install the SDK

    Terminal window
    npm install @scalekit-sdk/node dotenv
  2. Set your credentials

    Add your Scalekit credentials to your .env file. 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>
  3. Create the Wistia MCP connection

    In AgentKit > Connections, create a Wistia MCP connection and copy its redirect URI. The name you give it is the connection_name your code passes. See Configure connections.

  4. Register an OAuth app

    Wistia MCP server connections use your own OAuth app. Register one with Wistia MCP server and add the redirect URI you copied.

    Then enter the app's Client ID and Client Secret on the Wistia MCP connection.

  5. 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.actions
    const connector = 'wistiamcp'
    const identifier = 'user_123'
    // Generate an authorization link for the user
    const { link } = await actions.getAuthorizationLink({ connectionName: connector, identifier })
    console.log('Authorize Wistia MCP:', link)
    const rl = createInterface({ input: process.stdin, output: process.stdout })
    await rl.question('Press Enter after authorizing...')
    rl.close()
    // Make your first call
    const result = await actions.executeTool({
    connector,
    identifier,
    toolName: 'wistiamcp_check_credits',
    toolInput: {},
    })
    console.log(result)
    Terminal window
    npx tsx quickstart.mts

    Each user signs in once. See Authorize a user for the full flow and statuses.

Tools

Pass the exact name to execute_tool
Try in PlaygroundRequest a tool
  • wistiamcp_check_creditsCheck whether the account has enough Credits for a paid order of captions, dubs (localizations), text translations or extended audio descriptions, for one media or many, and get the per-item rate for each.Read-only

    check credits

    Check whether the account has enough Credits for a paid order of captions, dubs (localizations), text translations or extended audio descriptions, for one media or many, and get the per-item rate for each. Read-only: it reserves no Credits and places nothing, and each order still holds its own Credits when it runs. - Checking a create-bulk-purchase job or its actions before submitting them. - Checking a single dub, human-reviewed caption or translation order before placing it.

    Inputs

    actionsarray
    The items to order: each has resource_type, id (media hashed ID) and payload, as in create-bulk-purchase actions. A single dub is one localization action.
    jobobject
    A create-bulk-purchase job, instead of actions.
  • wistiamcp_display_resultsShow the user a card of the media, folders, subfolders, channels, or webinars that a search or list call returned.Read-only

    display results

    Show the user a card of the media, folders, subfolders, channels, or webinars that a search or list call returned. search and the list tools return data only and show the user nothing. When the user asked to see the records, call this with the `_mcp_tool_call_id` from that call's result: - "Find my guitar video": search, then display its result. - "Show me what's in the Launch folder": get-medias, then display its result. - Results the user picks from or reviews before a write. Skip it for lookups made only to get an ID, for counts, and when another tool shows the content itself, such as play-media. The card is built from the result that call returned, with its count and its link to the full results, so pass no record data. It shows every record that call returned; to show fewer, run the call again with narrower arguments. A search that matched nothing gets a card saying so. When you paged through one list (the same call with the same arguments apart from `page`), call this once, after the last page, with the ID of every page's call, and the pages show as one card. Results of different calls each need a call of their own. Only a result whose tool description names display-results can be displayed, and only by the caller who made the call. Only a host that renders MCP Apps shows the card. When the result has `card_shown: true`, do not repeat its rows in your response. Otherwise the user may see no card, so list the records in your response too.

    Inputs

    mcp_tool_call_idsarrayrequired
    The `_mcp_tool_call_id` in the result of the search or list call whose records to show. For pages of one list, the ID of each page's call.
  • wistiamcp_find_media_by_embed_locationFind the media embedded at a given URL.Read-only

    find media by embed location

    Find the media embedded at a given URL. Returns the hashed IDs of the account's media that recorded activity at that embed location during the date range, ranked by plays. The resulting hashed IDs can be passed to other endpoints, such as Show Account Top Content's `hashed_ids[]` filter, to fetch analytics for those media. The domain of `embed_url` is always matched exactly. Its path is matched exactly by default, or as a prefix with `path_match=prefix` (e.g. `/pricing` also matching `/pricing/plans`). A path that is empty or `/` is ignored, returning media across all paths on the domain. Embed location data is retained for 6 months. A `start_date` older than that is moved forward to 6 months ago (UTC); a range that ends before then returns a 422 error. When `start_date` and `end_date` are omitted, the full 6-month queryable window is used.

    Inputs

    embed_urlstringrequired
    The URL of the page to look up, e.g. `https://example.com/pricing`. The protocol is optional (https is assumed), so `example.com/pricing` also works.
    end_datestring
    End date for the analytics period in ISO 8601 format (YYYY-MM-DD). Exclusive — the range ends before the beginning of this date. Must be on or after `start_date` and no more than 730 days after it. Defaults to tomorrow, so today's activity is included.
    fieldsstring
    JSON Mask selector (Google partial-response syntax, * = wildcard) returning only the fields you name. Unknown fields are rejected. Fields: media_hashed_ids
    path_matchstring
    How to match the path of `embed_url` against embed locations. `exact` requires the path to match exactly; `prefix` matches any embed path starting with it.one of exactprefixdefault exact
    per_pageinteger
    Number of media hashed IDs to return (max 1000).default 100
    start_datestring
    Start date for the analytics period in ISO 8601 format (YYYY-MM-DD). Inclusive — the range starts at the beginning of this date. Must be no more than 730 days before `end_date`. A date more than 6 months ago is moved forward to 6 months ago. Defaults to 6 months ago, the start of the queryable window.
  • wistiamcp_find_transcript_matchesFind exact transcript wording across explicitly identified media without writing.Read-only

    find transcript matches

    Find exact transcript wording across explicitly identified media without writing. Use it to answer where wording occurs, or to prepare a targeted `edit-captions-text` call: it returns each track's current version and the start/end ranges that disambiguate an edit. A `resolved` match means the wording was located, not that a write will succeed — authorization, version conflicts, or non-editable boundaries can still reject it. Exact `matches` and fuzzy `suggestions` are separate; never treat a suggestion as the requested wording. Past 10 exact matches, call again with `occurrence` set to the next unseen one-based occurrence (`next_occurrence` when present). Omit `language_code` unless copying an exact tag from a prior result: a sole track is selected automatically, and several return `language_required` with exact IETF tags. Never infer `en` or `eng` from a language name. When reducing fields, always retain `track.version` so the edit can reuse them. Select through the envelope with `results(media_hashed_id,status,tracks(language),track(language,version),match_count,matches,next_occurrence)`.

    Inputs

    media_hashed_idsarrayrequired
    Explicit hashed IDs of media whose transcripts should be searched.
    target_textstringrequired
    Exact caption wording to locate.
    end_msinteger
    Optional end of an edit-compatible time range used to disambiguate the match. Omit to run to the end.
    fieldsstring
    JSON Mask selector (Google partial-response syntax, * = wildcard) returning only the fields you name. Unknown fields are rejected. Fields: requested_count,succeeded_count,failed_count,complete,results(media_hashed_id,status,tracks(id,language,english_name,native_name,version,is_draft),track(id,language,english_name,native_name,version,is_draft),match_status,match_count,matches(start_ms,end_ms,text,context),matches_truncated,next_occurrence,suggestions(start_ms,end_ms,text,context),suggestions_truncated,hint,recovery)
    language_codestring
    An exact IETF tag copied from a prior result, including extended tags such as zh-Hant. Omit on the first call; never infer en or eng from a language name.
    occurrenceinteger
    Optional one-based exact occurrence to return first, including occurrences after the first 10.
    start_msinteger
    Optional start of an edit-compatible time range used to disambiguate the match. Omit to start at the beginning.
  • wistiamcp_get_account_usageRead the account's plan, upload eligibility, and media, storage, seat, and bandwidth usage against limits.Read-only

    get account usage

    Read the account's plan, upload eligibility, and media, storage, seat, and bandwidth usage against limits. - Explaining why an upload or additional team member may exceed a limit. - Reporting remaining capacity before a bulk import or new seats. Use get-current-account for the account profile and get-credit-balance for Credits.

    Inputs

    fieldsstring
    JSON Mask selector (Google partial-response syntax, * = wildcard) returning only the fields you name. Unknown fields are rejected. Fields: plan(tier,name,is_standard,is_trial,trial_expires_at,trial_plan(tier,name)),can_upload,upload_blocked_reason,limits(primary_resource,media(count,flexible_limit,inflexible_limit),storage(used_bytes,flexible_limit_bytes,inflexible_limit_bytes),users(billable_count,account_billable_count,viewers_count,flexible_limit,inflexible_limit),bandwidth(used_bytes,limit_bytes,period,resets_at)),links(plans_url,usage_url,billing_url),role,verified_domain,can_invite_teammates,desktop_flags
  • wistiamcp_get_background_job_statusGet a Wistia background job's current status by its hashed id.Read-only

    get background job status

    Get a Wistia background job's current status by its hashed id. Call it for any tool whose result includes a background job status, to track the async operation it kicked off. In UI-capable hosts one call renders a live widget that messages you when the job ends: call it once, then wait rather than polling. In headless hosts each call is a snapshot, so poll until "finished" or "failed". Jobs carrying typed detail return it under `object.details`, keyed by `object.details.type`. A bulk operation (create-bulk-actions or create-bulk-purchase) returns type "bulk_operation" with `summary`, the total/succeeded/failed counts for the whole batch, and `results`, the per-action outcomes: `position`, `operation`, `resource_type`, `id` (the affected record, and the new one for a successful create), `status`, and `error` on failures. `results` is capped at 100 entries, failures first, with `results_truncated` set and `results_omitted` counting the rest; the full list is on GET /modern/background_job_status/{id}.

    Inputs

    background_job_status_hashed_idstringrequired
    The hashed id of the background job status to watch (the hashed_id returned by the operation that enqueued the job).
    app_only_propsobject
    Reserved for the host application. Agents must always pass this as an empty object ({}) and never put anything in it. The application embedding this MCP server may populate it with its own data.
  • wistiamcp_get_brand_kit_colorsGet the current account's brand colors: brand kit colors and each brand's solid colors as hex strings, plus any brand gradients.Read-only

    get brand kit colors

    Get the current account's brand colors: brand kit colors and each brand's solid colors as hex strings, plus any brand gradients.

    Inputs

    fieldsstring
    JSON Mask selector (Google partial-response syntax, * = wildcard) returning only the fields you name. Unknown fields are rejected. Fields: colors(name,value),brand_gradients(name,stops(color,position))
  • wistiamcp_get_brand_preloadRead the account's Brandfetch color, logo, domain, and brand-kit presence.Read-only

    get brand preload

    Read the account's Brandfetch color, logo, domain, and brand-kit presence. - Suggesting initial branding during signup. - Checking whether onboarding can reuse an existing brand kit.

    Inputs

    fieldsstring
    JSON Mask selector (Google partial-response syntax, * = wildcard) returning only the fields you name. Unknown fields are rejected. Fields: has_brand_kits,brandfetch_brand(domain,primary_color,logo,font,colors,logos(theme,type,tags,formats(src,format,background,height,width,size)),status)