The Browse AI connector lets your AI agent act in each user's Browse AI account. Each user connects their own Browse AI access token once, and Scalekit sends it with every call, so your agent never handles credentials. It comes with 23 tools.
- Tools
- 23
- What they doRead · write · destructive
- 12 · 9 · 212 read9 write2 destructive
- Users sign in with
- Access token
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 Browse AI connection
In AgentKit > Connections, create a Browse AI connection. The name you give it is the
connection_nameyour code passes. See Configure connections.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 = 'browseai'const identifier = 'user_123'// Generate an authorization link for the userconst { link } = await actions.getAuthorizationLink({ connectionName: connector, identifier })console.log('Authorize Browse AI:', 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: 'browseai_robot_drafts_list',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 = "browseai"identifier = "user_123"# Generate an authorization link for the userlink_response = actions.get_authorization_link(connection_name=connection_name,identifier=identifier,)print("Authorize Browse AI:", link_response.link)input("Press Enter after authorizing...")# Make your first callresult = actions.execute_tool(tool_input={},tool_name="browseai_robot_drafts_list",connection_name=connection_name,identifier=identifier,)print(result)Terminal window python quickstart.pyEach user opens the link once and enters their Browse AI credentials there. See Authorize a user for the full flow and statuses.
Tools
Pass the exact name toexecute_toolbrowseai_bulk_run_getGet one bulk run with a page of the tasks it ran.Read-onlyGet Bulk Run
Get one bulk run with a page of the tasks it ran. Returns the bulk run's id, title, status and counts, plus its tasks with status and captured data. Use this to follow a run started by browseai_bulk_run_create. Page through tasks with page.
Inputs
bulk_run_idstringrequired- Unique bulk run ID returned by browseai_bulk_run_create or browseai_bulk_runs_list. Example: 3c7d9e1f-2a4b-4c5d-8e6f-7a8b9c0d1e2f.
robot_idstringrequired- Unique robot ID. Get it from browseai_robots_list or from the robot's URL on the Browse AI dashboard. Example: f2d0a9a4-6b4e-4c3f-9d1a-2b7c8e5f1a3d.
pageinteger- Page of the bulk run's tasks, starting at 1. Example: 1.
browseai_bulk_runs_listList the bulk runs of one robot, with pagination.Read-onlyList Bulk Runs
List the bulk runs of one robot, with pagination. Returns bulk runs with their id, title, status, task counts and timestamps. Use this to find a bulk run id. Use browseai_bulk_run_get for its tasks.
Inputs
robot_idstringrequired- Unique robot ID. Get it from browseai_robots_list or from the robot's URL on the Browse AI dashboard. Example: f2d0a9a4-6b4e-4c3f-9d1a-2b7c8e5f1a3d.
pageinteger- Page number, starting at 1. Example: 1.
browseai_monitor_getGet one monitor on a robot.Read-onlyGet Monitor
Get one monitor on a robot. Returns the monitor's id, name, status, inputParameters, schedule and notification settings. Use this to inspect a monitor before updating it. Use browseai_monitors_list to find ids.
Inputs
monitor_idstringrequired- Unique monitor ID. Get it from browseai_monitors_list. Example: a8c3e1f2-4d5b-4a6c-8e7f-9b0a1c2d3e4f.
robot_idstringrequired- Unique robot ID. Get it from browseai_robots_list or from the robot's URL on the Browse AI dashboard. Example: f2d0a9a4-6b4e-4c3f-9d1a-2b7c8e5f1a3d.
browseai_monitors_listList the monitors on one robot.Read-onlyList Monitors
List the monitors on one robot. A monitor runs the robot on a schedule and reports changes. Returns each monitor's id, name, status, inputParameters, schedule and notification settings. Use this to find a monitor id. Use browseai_monitor_get for one monitor.
Inputs
robot_idstringrequired- Unique robot ID. Get it from browseai_robots_list or from the robot's URL on the Browse AI dashboard. Example: f2d0a9a4-6b4e-4c3f-9d1a-2b7c8e5f1a3d.
browseai_robot_draft_getGet one robot draft with its status, schema and the data its finished preview extracted.Read-onlyGet Robot Draft
Get one robot draft with its status, schema and the data its finished preview extracted. Returns the draft's status, schema, latestPreview (extracted data or why the preview failed) and timestamps. Poll this after browseai_robot_draft_create or browseai_robot_draft_update until the preview finishes.
Inputs
draft_idstringrequired- Unique robot draft ID returned by browseai_robot_draft_create or browseai_robot_drafts_list. Example: 5b9e2c1d-7a3f-4e8b-a1c6-0d2f3e4a5b6c.
browseai_robot_drafts_listList the robot drafts in your workspace, newest first, including saved drafts.Read-onlyList Robot Drafts
List the robot drafts in your workspace, newest first, including saved drafts. Returns drafts with their id, name, status, originUrl and schema, without latestPreview. Use this to find a draft id. Use browseai_robot_draft_get for a draft's preview data.
Inputs
pageinteger- Page number, starting at 1. Example: 1.
page_sizeinteger- Drafts per page. Browse AI defaults to 10. Example: 10.
browseai_robot_getGet one Browse AI robot by id.Read-onlyGet Robot
Get one Browse AI robot by id. Returns the robot's id, name, createdAt and inputParameters with their names, types and defaults. Use this to see which input_parameters a robot accepts before running it. Use browseai_robots_list to find ids.
Inputs
robot_idstringrequired- Unique robot ID. Get it from browseai_robots_list or from the robot's URL on the Browse AI dashboard. Example: f2d0a9a4-6b4e-4c3f-9d1a-2b7c8e5f1a3d.
browseai_robots_listList the robots in your Browse AI account.Read-onlyList Robots
List the robots in your Browse AI account. Returns each robot's id, name, createdAt and inputParameters (the inputs it accepts). Use this to find a robot id. Use browseai_robot_get for one robot's details.
Inputs
This tool takes no inputs.
browseai_status_getGet the current status of the Browse AI infrastructure and its task queue.Read-onlyGet System Status
Get the current status of the Browse AI infrastructure and its task queue. Returns the system status, including whether the task queue is operating normally. Use this to check for an outage before running robots or when tasks are slow.
Inputs
This tool takes no inputs.
browseai_task_getGet one robot task by id, with its status and the data it captured.Read-onlyGet Task
Get one robot task by id, with its status and the data it captured. Returns the task's status, inputParameters, capturedTexts, capturedScreenshots, capturedLists, timestamps and any error. Use this to poll a task started by browseai_task_run. A retried task stays in-progress until its last retry finishes.
Inputs
robot_idstringrequired- Unique robot ID. Get it from browseai_robots_list or from the robot's URL on the Browse AI dashboard. Example: f2d0a9a4-6b4e-4c3f-9d1a-2b7c8e5f1a3d.
task_idstringrequired- Unique task ID returned by browseai_task_run or browseai_tasks_list. Example: 9d8c7b6a-5e4f-4a3b-9c2d-1e0f2a3b4c5d.
browseai_tasks_listList the tasks (runs) of one robot, optionally filtered by status, bulk run or creation date, with pagination.Read-onlyList Robot Tasks
List the tasks (runs) of one robot, optionally filtered by status, bulk run or creation date, with pagination. Returns tasks with their id, status, inputParameters, captured texts, screenshots and lists, and timestamps. Use this to browse past runs. Use browseai_task_get for a single task by id.
Inputs
robot_idstringrequired- Unique robot ID. Get it from browseai_robots_list or from the robot's URL on the Browse AI dashboard. Example: f2d0a9a4-6b4e-4c3f-9d1a-2b7c8e5f1a3d.
from_dateinteger- Only tasks created at or after this time, as a Unix timestamp. Example: 1735689600.
pageinteger- Page number, starting at 1. Example: 1.
page_sizeinteger- Tasks per page. Browse AI defaults to 10. Example: 10.
robot_bulk_run_idstring- Only return tasks that ran in this bulk run. Example: 3c7d9e1f-2a4b-4c5d-8e6f-7a8b9c0d1e2f.
sortstring- Comma-separated fields to sort by. Ascending by default; prefix a field with - for descending. Example: -createdAt.
statusstring- Only return tasks with this status. Example: successful.one of
failedsuccessfulin-progress to_dateinteger- Only tasks created at or before this time, as a Unix timestamp. Example: 1738367999.
browseai_webhooks_listList the webhooks registered on one robot.Read-onlyList Webhooks
List the webhooks registered on one robot. Returns each webhook's id, hookUrl and eventType. Use this to find a webhook id before deleting it.
Inputs
robot_idstringrequired- Unique robot ID. Get it from browseai_robots_list or from the robot's URL on the Browse AI dashboard. Example: f2d0a9a4-6b4e-4c3f-9d1a-2b7c8e5f1a3d.
browseai_bulk_run_createRun a robot on many inputs at once: up to 50,000 tasks in one bulk run.WriteCreate Bulk Run
Run a robot on many inputs at once: up to 50,000 tasks in one bulk run. Returns the bulk run with its id, title, status and task counts. Use this for batches of URLs or inputs. Track it with browseai_bulk_run_get. Use browseai_task_run for one input.
Inputs
input_parametersarrayrequired- Array of input parameter objects, one per task. Each object overrides the robot's default inputs for that task, for example originUrl. Example: [{"originUrl": "https://www.ycombinator.com/companies/airbnb"}, {"originUrl": "https://www.ycombinator.com/companies/coinbase"}]
robot_idstringrequired- Unique robot ID. Get it from browseai_robots_list or from the robot's URL on the Browse AI dashboard. Example: f2d0a9a4-6b4e-4c3f-9d1a-2b7c8e5f1a3d.
titlestring- Title describing the bulk run, 1 to 200 characters. Example: Product pages batch 1.
browseai_monitor_createCreate a monitor that runs a robot on a recurring schedule and notifies you when captured text or screenshots change.WriteCreate Monitor
Create a monitor that runs a robot on a recurring schedule and notifies you when captured text or screenshots change. Returns the new monitor with its id, status and settings. Use this for ongoing change tracking. Use browseai_task_run for a single run.
Inputs
captured_screenshot_notification_thresholdnumberrequired- Screenshot change, in percent, above which the screenshot-changed email is sent. Example: 15.
input_parametersobjectrequired- Input parameters for each scheduled run, keyed by input name, for example originUrl. Example: {"originUrl": "https://www.ycombinator.com/companies/airbnb", "companies_limit": 10}
namestringrequired- Monitor name, 1 to 200 characters. Example: Competitor pricing page.
notify_on_captured_screenshot_changebooleanrequired- Email you when a captured screenshot changes by more than the threshold. Example: true.
notify_on_captured_text_changebooleanrequired- Email you when captured text changes. Example: true.
robot_idstringrequired- Unique robot ID. Get it from browseai_robots_list or from the robot's URL on the Browse AI dashboard. Example: f2d0a9a4-6b4e-4c3f-9d1a-2b7c8e5f1a3d.
schedulestring- Recurring schedule as an iCalendar RRULE string, for example hourly on weekdays. Example: FREQ=HOURLY;INTERVAL=1;BYWEEKDAY=MO,TU,WE,TH,FR.
browseai_monitor_updateUpdate a monitor's name, status, input parameters, schedule or notification settings.WriteUpdate Monitor
Update a monitor's name, status, input parameters, schedule or notification settings. Only the fields you send change. Returns the updated monitor. Use status paused to stop a monitor without deleting it, and active to resume it. Use browseai_monitor_delete to remove it.
- Idempotent
Inputs
monitor_idstringrequired- Unique monitor ID. Get it from browseai_monitors_list. Example: a8c3e1f2-4d5b-4a6c-8e7f-9b0a1c2d3e4f.
robot_idstringrequired- Unique robot ID. Get it from browseai_robots_list or from the robot's URL on the Browse AI dashboard. Example: f2d0a9a4-6b4e-4c3f-9d1a-2b7c8e5f1a3d.
captured_screenshot_notification_thresholdnumber- Screenshot change, in percent, above which the screenshot-changed email is sent. Example: 15.
input_parametersobject- Input parameters for each scheduled run, keyed by input name. Example: {"originUrl": "https://www.ycombinator.com/companies/airbnb", "companies_limit": 10}
namestring- New monitor name. Example: Competitor pricing page.
notify_on_captured_screenshot_changeboolean- Email you when a captured screenshot changes by more than the threshold. Example: true.
notify_on_captured_text_changeboolean- Email you when captured text changes. Example: true.
schedulestring- Recurring schedule as an iCalendar RRULE string, for example hourly on weekdays. Example: FREQ=HOURLY;INTERVAL=1;BYWEEKDAY=MO,TU,WE,TH,FR.
statusstring- active to run on schedule, paused to stop until set back to active. Example: paused.one of
activepaused
browseai_robot_draft_createTrain a new robot, or retrain an existing one, on a webpage from natural-language instructions and/or a JSON Schema of the data to extract.WriteCreate Robot Draft
Train a new robot, or retrain an existing one, on a webpage from natural-language instructions and/or a JSON Schema of the data to extract. A preview starts right away. Returns the draft with its id, status and schema; latestPreview is empty until the preview finishes. Poll browseai_robot_draft_get until status is ready, then call browseai_robot_draft_save to create the robot.
Inputs
origin_urlstringrequired- Webpage to train the robot on. After saving, the robot can run on other pages of the same site by passing a different originUrl. Example: https://www.ycombinator.com/companies?batch=Fall%202025&tags=AI.
countrystring- Country the robot browses from, as an uppercase ISO 3166-1 alpha-3 code. Defaults to the country of the URL's domain, or USA. Can't be changed later. Example: USA.
instructionsstring- Overall instructions for the robot, up to 5,000 characters. Send instructions, schema or both. Without schema, the columns to extract are worked out from these instructions. Example: Extract the AI companies in this batch: name, one-line description, batch, and industry.
namestring- Robot name, 1 to 255 characters and not a UUID. A name is suggested if omitted. Example: YC AI companies.
robot_idstring- Existing robot ID to retrain instead of creating a new robot. Saving the draft writes a new version of that robot. Example: f2d0a9a4-6b4e-4c3f-9d1a-2b7c8e5f1a3d.
schemaobject- The data to extract, as a JSON Schema object. Each key of properties is a column name and its description holds that column's instructions. Use an array property with items for lists (maxItems caps the rows). Send instructions, schema or both. Example: {"type": "object", "properties": {"Products": {"type": "array", "maxItems": 50, "items": {"type": "object", "properties": {"Name": {"type": "string"}, "Price...
browseai_robot_draft_saveSave a ready robot draft: create the robot (or a new version of the robot being retrained) and start its first task.WriteSave Robot Draft
Save a ready robot draft: create the robot (or a new version of the robot being retrained) and start its first task. Returns the saved draft with the robot id and the first task. Only drafts with status ready can be saved. The first task uses task credits.
Inputs
draft_idstringrequired- Unique robot draft ID returned by browseai_robot_draft_create or browseai_robot_drafts_list. Example: 5b9e2c1d-7a3f-4e8b-a1c6-0d2f3e4a5b6c.
browseai_robot_draft_updateChange a robot draft's origin URL, instructions, schema or name.WriteUpdate Robot Draft
Change a robot draft's origin URL, instructions, schema or name. Changing the URL, instructions or schema starts a new preview and cancels a running one. Returns the updated draft; latestPreview is empty until the new preview finishes. Poll browseai_robot_draft_get afterwards. Changing only the name does not start a preview.
- Idempotent
Inputs
draft_idstringrequired- Unique robot draft ID returned by browseai_robot_draft_create or browseai_robot_drafts_list. Example: 5b9e2c1d-7a3f-4e8b-a1c6-0d2f3e4a5b6c.
instructionsstring- New overall instructions. Sent without schema, they replace the draft's columns. An empty string clears them. Example: Only include B2B companies.
namestring- New robot name, 1 to 255 characters and not a UUID. Example: YC B2B AI companies.
origin_urlstring- Webpage the preview runs on. Example: https://www.ycombinator.com/companies?batch=Fall%202025.
schemaobject- The data to extract, as a JSON Schema object. Each key of properties is a column name and its description holds that column's instructions. Use an array property with items for lists (maxItems caps the rows). Send instructions, schema or both. Example: {"type": "object", "properties": {"Products": {"type": "array", "maxItems": 50, "items": {"type": "object", "properties": {"Name": {"type": "string"}, "Price...
browseai_task_runRun a robot once on demand, optionally overriding its input parameters such as originUrl.WriteRun Robot
Run a robot once on demand, optionally overriding its input parameters such as originUrl. Returns the new task with its id and status (usually in-progress); captured data is not ready yet. Poll browseai_task_get with the task id until status is successful or failed. Use browseai_bulk_run_create to run many inputs at once.
Inputs
robot_idstringrequired- Unique robot ID. Get it from browseai_robots_list or from the robot's URL on the Browse AI dashboard. Example: f2d0a9a4-6b4e-4c3f-9d1a-2b7c8e5f1a3d.
input_parametersobject- Input parameters that override the robot's defaults for this run, keyed by input name. Most robots take originUrl plus list limits such as <list>_limit. Values are strings, numbers or arrays of strings. Example: {"originUrl": "https://www.ycombinator.com/companies/airbnb", "companies_limit": 10}
record_videoboolean- Try to record a video of the run. Not guaranteed on heavy sites. Example: false.
browseai_webhook_createRegister a webhook URL that Browse AI calls when a robot's tasks finish or their captured data changes.WriteCreate Webhook
Register a webhook URL that Browse AI calls when a robot's tasks finish or their captured data changes. Returns the new webhook with its id, hookUrl and eventType. Use this to receive results without polling browseai_task_get.
Inputs
event_typestringrequired- Event that triggers the webhook. Example: taskFinishedSuccessfully.one of
taskCapturedDataChangedtaskFinishedtaskFinishedSuccessfullytaskFinishedWithErrortableExportFinishedSuccessfully hook_urlstringrequired- HTTPS URL that receives the webhook POST. Example: https://example.com/hooks/browseai.
robot_idstringrequired- Unique robot ID. Get it from browseai_robots_list or from the robot's URL on the Browse AI dashboard. Example: f2d0a9a4-6b4e-4c3f-9d1a-2b7c8e5f1a3d.
browseai_monitor_deletePermanently delete a monitor from a robot.DestructiveDelete Monitor
Permanently delete a monitor from a robot. The robot and its past tasks are kept. Returns a confirmation. Use browseai_monitor_update with status paused to stop a monitor temporarily instead.
- Idempotent
Inputs
monitor_idstringrequired- Unique monitor ID. Get it from browseai_monitors_list. Example: a8c3e1f2-4d5b-4a6c-8e7f-9b0a1c2d3e4f.
robot_idstringrequired- Unique robot ID. Get it from browseai_robots_list or from the robot's URL on the Browse AI dashboard. Example: f2d0a9a4-6b4e-4c3f-9d1a-2b7c8e5f1a3d.
browseai_webhook_deleteDelete a webhook from a robot so Browse AI stops calling its URL.DestructiveDelete Webhook
Delete a webhook from a robot so Browse AI stops calling its URL. Returns a confirmation. Use browseai_webhooks_list to find the webhook id.
- Idempotent
Inputs
robot_idstringrequired- Unique robot ID. Get it from browseai_robots_list or from the robot's URL on the Browse AI dashboard. Example: f2d0a9a4-6b4e-4c3f-9d1a-2b7c8e5f1a3d.
webhook_idstringrequired- Unique webhook ID from browseai_webhooks_list. Example: 7e6d5c4b-3a2f-4e1d-9c8b-7a6f5e4d3c2b.
No tools match.