Skip to content
Changelog

Report a bug

Browserless integration: operations for AI agents

Browserless connects to Type.com as an integration. This page lists the 106 operations documented for Claude and Codex in Type.com: 33 read, 62 write, 11 destructive.

About Browserless

Browserless API reference operations, subject to Type setup and account permissions.

Categories: Developer Tools & IT

How to connect Browserless to Type.com

Connect
Configure provider-specific API target and authentication setup.
Access
Organization connection: one shared account for the workspace.

To add the connection, follow the steps in Space connections.

Documented operations

106 operations. Your connection's permissions determine which operations are available. API references describe the vendor surface; they do not expand your connection's credentials or access.

106 operations listed.

Screenshots & PDFs

  • Write

    /screenshot

    Returns a screenshot binary from a `url` or `html` payload. "url" or "html" payload in your request. Many options exist including cookies, user-agents, setting timers and network mocks. **Note:** This is the preferred endpoint. The `/chromium/screenshot` route is equivalent.

    POST /screenshot

  • Write

    /pdf

    Returns a PDF binary from a `url` or `html` payload. "url" or "html" payload in your request. Many options exist for injecting cookies, request interceptors, user-agents and waiting for selectors, timers and more. **Note:** This is the preferred endpoint. The `/chromium/pdf` route is equivalent.

    POST /pdf

  • Write

    /edge/screenshot

    Returns a screenshot binary from a `url` or `html` payload using Edge. "url" or "html" payload in your request. Many options exist including cookies, user-agents, setting timers and network mocks.

    POST /edge/screenshot

  • Write

    /edge/pdf

    Returns a PDF binary from a `url` or `html` payload using Edge. "url" or "html" payload in your request. Many options exist for injecting cookies, request interceptors, user-agents and waiting for selectors, timers and more.

    POST /edge/pdf

Scraping & Content

  • Write

    /chromium/content

    Given a `url` or `html` field, loads the page and returns the fully rendered HTML. **Note:** Also available at `/content` as a convenience alias.

    POST /chromium/content

  • Write

    /chromium/scrape

    Returns text, HTML, and metadata from a given list of CSS selectors against the rendered DOM. Debugging information is available by sending in the appropriate flags in the "debugOpts" property. Responds with an array of JSON objects. **Note:** Also available at `/scrape` as a convenience alias.

    POST /chromium/scrape

  • Write

    /smart-scrape

    Intelligently scrapes a URL using cascading strategies (HTTP fetch, proxy, headless browser, etc).

    POST /smart-scrape

  • Write

    /search

    Search the web and optionally scrape result pages. Accepts a query, performs a web search via SearXNG, and optionally scrapes each result URL — returning structured, LLM-ready data (markdown, HTML, links, or screenshots).

    POST /search

  • Write

    /map

    Primarily discovers URLs from the site's sitemap, supplemented with link extraction from the page via the smart-scrape strategy pipeline (HTTP-first, browser fallback for JS-rendered pages). Use the search parameter to order results by relevance.

    POST /map

  • Write

    /edge/content

    Given a `url` or `html` field, loads the page in Edge and returns the fully rendered HTML.

    POST /edge/content

  • Write

    /edge/scrape

    Returns text, HTML, and metadata from a given list of CSS selectors against the rendered DOM (Edge). Debugging information is available by sending in the appropriate flags in the "debugOpts" property. Responds with an array of JSON objects.

    POST /edge/scrape

  • Write

    /chromium/performance

    Run lighthouse performance audits with a supplied "url" in your JSON payload. **Note:** Also available at `/performance` as a convenience alias.

    POST /chromium/performance

  • Write

    /edge/performance

    Run lighthouse performance audits with a supplied "url" in your JSON payload.

    POST /edge/performance

  • Write

    /scrape

    A JSON-based API that returns text, html, and meta-data from a given list of selectors. Debugging information is available by sending in the appropriate flags in the "debugOpts" property. Responds with an array of JSON objects. **Note:** This endpoint is also available at: `/chromium/scrape` for backwards compatibility.

    POST /scrape

Functions & Downloads

  • Write

    /chromium/function

    Executes a Puppeteer or Playwright script in the browser context and returns the result. Browserless sets up a blank page, injects your puppeteer code, and runs it. You can optionally load external libraries via the "import" module that are meant for browser usage. Values returned from the function are checked and an appropriate content-type and response is sent back to your HTTP call. **Note:** Also available at `/function` as a convenience alias.

    POST /chromium/function

  • Write

    /chromium/download

    Runs a Puppeteer or Playwright script and returns any files the browser downloaded during execution. the execution of puppeteer code, which is ran inside context of the browser. Browserless sets up a blank page, a fresh download directory, injects your puppeteer code, and then executes it. You can load external libraries via the "import" syntax, and import ESM-style modules that are written for execution inside of the browser. Once your script is finished, any downloaded files from Chromium are returned back with the appropriate content-type header. **Note:** Also available at `/download` as a convenience alias.

    POST /chromium/download

  • Write

    /edge/function

    Executes a Puppeteer or Playwright script in the Edge browser context and returns the result. Browserless sets up a blank page, injects your puppeteer code, and runs it. You can optionally load external libraries via the "import" module that are meant for browser usage. Values returned from the function are checked and an appropriate content-type and response is sent back to your HTTP call.

    POST /edge/function

  • Write

    /edge/download

    Runs a Puppeteer or Playwright script in Edge and returns any files the browser downloaded during execution. the execution of puppeteer code, which is ran inside context of the browser. Browserless sets up a blank page, a fresh download directory, injects your puppeteer code, and then executes it. You can load external libraries via the "import" syntax, and import ESM-style modules that are written for execution inside of the browser. Once your script is finished, any downloaded files from Chromium are returned back with the appropriate content-type header.

    POST /edge/download

  • Write

    /chromium/export

    Exports a webpage to a PDF or image format. This API is useful for generating reports, screenshots, or PDFs of web pages. **Note:** Also available at `/export` as a convenience alias.

    POST /chromium/export

Sessions & Connections

  • Read

    /json/version

    Returns a JSON payload that acts as a pass-through to the DevTools /json/version protocol in Chrome and Chromium.

    GET /json/version

  • Read

    /json/protocol

    Returns Protocol JSON meta-data that Chrome and Chromium come with.

    GET /json/protocol

  • Read

    /reconnect/*

    Reconnect to an existing Chromium or Chrome session with a library like puppeteer or others that work over chrome-devtools-protocol.

    GET /reconnect/*

  • Write

    /session

    Creates a new browser session with the specified parameters. The session can be used for persistent browser connections with well-defined lifetimes and reconnection semantics.

    POST /session

  • Write

    /json/new

    Returns a JSON payload that acts as a pass-through to the DevTools /json/new HTTP API in Chromium. Browserless mocks this payload so that remote clients can connect to the underlying "webSocketDebuggerUrl" which will cause Browserless to start the browser and proxy that request into a blank page.

    PUT /json/new

  • Write

    /session/bql/*

    > Executes BrowserQL queries against an existing session. BrowserQL is a GraphQL-based query language for browser automation that allows you to interact with web pages using a declarative syntax. Queries execute within the context of your existing session, maintaining browser state and session data across multiple operations.

    POST /session/bql/*

  • Write

    /session/connect/*

    Connect to an existing session with a library like puppeteer or playwright that work over chrome-devtools-protocol. See documentation for more details on how to start the session. The browser type (Chrome/Chromium, stealth/regular) is determined from the session metadata when it was created initially.

    GET /session/connect/*

  • Destructive

    /session/*

    Deletes a session and its associated data. This will immediately terminate any active connections to the session if "force" is applied. Query Parameters: - force (optional): If true, forces deletion even if the session has active connections

    DELETE /session/*

Browser Management

  • Read

    /active

    Returns a simple "204" HTTP code, with no response, indicating that the service itself is up and running. Useful for liveliness probes or other external checks.

    GET /active

  • Read

    /meta

    Returns a JSON payload of the current system versions, including the core API version.

    GET /meta

  • Write

    /browser/*/live

    Creates a live URL (view-only by default; pass interactable: true for a link a person can act through) for a currently-running browser, identified by browserId. Hosted authenticated Cloud plans, including Free, can use this route. Explicitly disabled and unlicensed deployments remain ineligible.

    POST /browser/*/live

  • Destructive

    /kill/*

    Kill running sessions based on BrowserId or TrackingId.

    GET /kill/+([0-9a-zA-Z-_])

  • Destructive

    /browser/*

    Terminates a browser instance by browserId. When browser ownership enforcement is enabled, the browser must belong to the authenticated user.

    DELETE /browser/*

Crawling

  • Read

    /crawl

    List all crawl jobs owned by the authenticated account (or, for keys not yet mapped to an account, the token). Returns an array of crawl summaries with status, progress counters, and timestamps. Supports cursor-based pagination via the `nextCursor` field.

    GET /crawl

  • Read

    /crawl/*

    Get the status of a crawl job and its paginated page results. Returns progress counters, per-page metadata with content URLs, and a `next` URL for offset-based pagination via the `skip` parameter.

    GET /crawl/*

  • Write

    /crawl

    Start an asynchronous crawl job that spiders a website and scrapes every discovered page. Returns a crawl ID and status URL for polling results via GET /crawl/{id}. Supports depth control, path filtering, sitemap strategies, and webhook notifications.

    POST /crawl

  • Destructive

    /crawl/*

    Cancel a running crawl job by its ID. Any pages already completed remain available for retrieval. Returns a 409 if the crawl is already in a terminal state.

    DELETE /crawl/*

Unblock & Stealth

  • Write

    /unblock

    Unblocks the provided URL from being blocked due to bot detection. Returns a payload of Cookies, HTML, a base64 encoded screenshot, and a "browserWSEndpoint" to allow connecting to the browser when specified in the JSON Payload. Only supports CDP or Puppeteer like libraries when connecting to the "browserWSEndpoint". **Note:** Also available at `/chromium/unblock` as a convenience alias.

    POST /unblock

  • Write

    /chrome/stealth

    Launch and connect to Stealthy Chrome with a library like puppeteer or others that work over chrome-devtools-protocol for scraping in a more stealth-like fashion.

    GET /chrome/stealth

  • Write

    /chromium/stealth

    Launch and connect to Stealthy Chromium with a library like puppeteer or others that work over chrome-devtools-protocol for scraping in a more stealth-like fashion.

    GET /chromium/stealth

  • Write

    /stealth

    Launch and connect to Stealthy Chromium with a library like puppeteer or others that work over chrome-devtools-protocol for scraping in a more stealth-like fashion.

    GET /stealth

Proxy

  • Read

    /proxy/cities

    Returns a list of available cities for proxy connections. This endpoint requires city proxying to be enabled on your plan. Query Parameters: - country (optional): Filter cities by two-letter country code (e.g., 'US', 'GB', 'DE') The response includes a list of available cities grouped by country code.

    GET /proxy/cities

Profiles

  • Read

    /profiles

    Lists authentication profiles owned by the requesting token. Paginated via "limit" (default 100, capped at 1000) and "offset" (default 0). Returns an empty array if the token has no profiles. **Note:** This endpoint is also available at: `/management/profiles` for backwards compatibility.

    GET /profiles

  • Read

    /profile/*

    Returns metadata for a specific authentication profile. The profile name is the URL-encoded path segment after "/profile/". Returns 404 if no profile with that name exists for this token. **Note:** This endpoint is also available at: `/management/profile/*` for backwards compatibility.

    GET /profile/*

  • Write

    /profile

    Launches a temporary browser session for creating an authenticated profile. Connect via the returned WebSocket URL, authenticate in the browser, then call Browserless.saveProfile via CDP to capture and persist the auth state. The profile is NOT saved automatically — you must explicitly call Browserless.saveProfile before disconnecting. The session expires after 10 minutes or your plan's session limit, whichever comes first. Free sessions are capped at two minutes; higher-tier plans retain the 10-minute maximum. **Note:** This endpoint is also available at: `/management/profile` for backwards compatibility.

    POST /profile

  • Write

    /profile/*

    Renames an authentication profile. The new name must be unique per token. The current name is the URL-encoded path segment after "/profile/". Returns 400 if the new name is already taken, 404 if the source profile is not found. **Note:** This endpoint is also available at: `/management/profile/*` for backwards compatibility.

    PUT /profile/*

  • Write

    /profile/upload

    Creates a new authentication profile from a pre-captured "state" payload (cookies plus per-origin localStorage and IndexedDB). The "name" must be unique per token. Counts of any dropped or truncated entries are surfaced under "diagnostics".

    POST /profile/upload

  • Destructive

    /profile

    Removes selected sites and their cookies, localStorage and IndexedDB from a saved profile. **Note:** This endpoint is also available at: `/management/profile` for backwards compatibility.

    DELETE /profile

  • Destructive

    /profile/*

    Deletes an authentication profile and its captured state. The profile name is the URL-encoded path segment after "/profile/". Returns 404 if no profile with that name exists for this token. **Note:** This endpoint is also available at: `/management/profile/*` for backwards compatibility.

    DELETE /profile/*

  • Destructive

    /profile/refresh

    Replaces the state of an existing authentication profile in place. The "name" must already exist for the requesting token. Counts of any dropped or truncated entries are surfaced under "diagnostics".

    POST /profile/refresh

Integrations

  • Read

    /integrations/onepassword

    Lists the 1Password integrations owned by the requesting token. Paginated via "limit" (default 100, capped at 1000) and "offset" (default 0). Records never include any token material. Pass includeVaults=true to return live vault titles for the first ten rows per response. Later rows and unavailable lookups return null. Paginate with limit=10 and offset to retrieve later titles. **Note:** This endpoint is also available at: `/management/integrations/onepassword` for backwards compatibility.

    GET /integrations/onepassword

  • Read

    /integrations/onepassword/*

    Returns metadata for a specific 1Password integration owned by the token. The id or label is the URL-encoded path segment after "/integrations/onepassword/". Returns 404 if no integration with that id or label exists for this token. The response never includes token material. Pass includeVaults=true to also return each integration's vault titles, looked up live from 1Password (null when unavailable). **Note:** This endpoint is also available at: `/management/integrations/onepassword/*` for backwards compatibility.

    GET /integrations/onepassword/*

  • Read

    /integrations/onepassword/*/audit

    Lists recent credential activity for a single 1Password integration owned by the token, newest first. One entry per Browserless.loadSecret or saveSecret attempt: outcome, error code, the credential reference, and non-secret target metadata. Secret values are never recorded or returned. Paginated via "limit" (default 100, capped at 1000) and "offset" (default 0). The URL-encoded path segment may be an integration id or label. Returns 404 if no matching integration exists for this token. **Note:** This endpoint is also available at: `/management/integrations/onepassword/*/audit` for backwards compatibility.

    GET /integrations/onepassword/*/audit

  • Read

    /integrations/onepassword/profile/*

    Poll an autologin task by id. Returns its state, the current live rationale, and (on success) the saved profile id plus a per-domain breakdown. Scoped to the requesting account (or token, for keys not yet mapped to an account) — returns 404 for unknown ids or another account's task.

    GET /integrations/onepassword/profile/*

  • Read

    /integrations/onepassword/autologin/form

    Returns the selectable references for building an autologin profile: the 1Password integration's login origins and, per origin, the field labels it fills (never values). The integrationId query accepts an id or label. Drives the create-profile form.

    GET /integrations/onepassword/autologin/form

  • Write

    /integrations/onepassword

    Registers a 1Password integration for the requesting token. The supplied service-account token is encrypted at rest and is never returned by this or any other endpoint. Returns the created integration without any token field. **Note:** This endpoint is also available at: `/management/integrations/onepassword` for backwards compatibility.

    POST /integrations/onepassword

  • Write

    /integrations/onepassword/profile

    Kicks off an autologin task: a managed agent creates a browser profile by logging into the integration's domains, then saves it. integrationId accepts an id or label. Returns a taskId to poll via GET /integrations/onepassword/profile/:taskId.

    POST /integrations/onepassword/profile

  • Write

    /integrations/script-generator

    Generates a Browserless automation script (BrowserQL / Playwright JS / Playwright Python / Puppeteer / BAP TypeScript) from a natural-language task. A managed agent drives a browser session; its ordered tool-call transcript is rendered into the requested script(s). Streams Server-Sent Events: status, script, complete, error.

    POST /integrations/script-generator

  • Destructive

    /integrations/onepassword/*

    Deletes a 1Password integration owned by the requesting token. The id or label is the URL-encoded path segment after "/integrations/onepassword/". Returns 404 if no matching integration exists for this token. Deleting an integration does not touch any authentication profile. **Note:** This endpoint is also available at: `/management/integrations/onepassword/*` for backwards compatibility.

    DELETE /integrations/onepassword/*

WebSocket APIs

  • Read

    /chromium/playwright

    Connect to Chromium with any playwright style library. **Note:** Also available at `/playwright` as a convenience alias.

    GET /chromium/playwright

  • Read

    /edge/playwright

    Connect to Chromium with any playwright style library.

    GET /edge/playwright

  • Read

    /firefox/playwright

    Connect to Firefox with any playwright-compliant library. **Note:** Also available at `/playwright/firefox` as a convenience alias.

    GET /firefox/playwright

  • Read

    /webkit/playwright

    Connect to Webkit with any playwright-compliant library. **Note:** Also available at `/playwright/webkit` as a convenience alias.

    GET /webkit/playwright

  • Read

    /live/*

    Websocket back-end that powers the live session experience. **Note:** Also available at `/chromium/live/*` as a convenience alias.

    GET /live/*

  • Read

    /chrome/live/*

    Websocket back-end that powers the live session experience.

    GET /chrome/live/*

  • Write

    /

    Launch and connect to Chromium with a library like puppeteer or others that work over chrome-devtools-protocol. **Note:** Also available at `/chromium` as a convenience alias.

    GET /

  • Write

    /chromium

    Launch and connect to Chromium with a library like puppeteer or others that work over chrome-devtools-protocol. **Note:** This endpoint is also available at: `/` for backwards compatibility.

    GET /chromium

  • Write

    /edge

    Launch and connect to Chromium with a library like puppeteer or others that work over chrome-devtools-protocol.

    GET /edge

Chrome (licensed build)

  • Read

    /chrome/playwright

    Connect to Chromium with any playwright style library.

    GET /chrome/playwright

  • Write

    /chrome/screenshot

    Returns a screenshot binary from a `url` or `html` payload (Chrome, legacy). "url" or "html" payload in your request. Many options exist including cookies, user-agents, setting timers and network mocks.

    POST /chrome/screenshot

  • Write

    /chrome/pdf

    Returns a PDF binary from a `url` or `html` payload (Chrome, legacy). "url" or "html" payload in your request. Many options exist for injecting cookies, request interceptors, user-agents and waiting for selectors, timers and more.

    POST /chrome/pdf

  • Write

    /chrome/content

    Given a `url` or `html` field, loads the page in Chrome and returns the fully rendered HTML.

    POST /chrome/content

  • Write

    /chrome/scrape

    Returns text, HTML, and metadata from a given list of CSS selectors against the rendered DOM (Chrome, legacy). selectors. Debugging information is available by sending in the appropriate flags in the "debugOpts" property. Responds with an array of JSON objects.

    POST /chrome/scrape

  • Write

    /chrome/performance

    Run lighthouse performance audits with a supplied "url" in your JSON payload.

    POST /chrome/performance

  • Write

    /chrome/function

    Executes a Puppeteer or Playwright script in the Chrome browser context and returns the result. Browserless sets up a blank page, injects your puppeteer code, and runs it. You can optionally load external libraries via the "import" module that are meant for browser usage. Values returned from the function are checked and an appropriate content-type and response is sent back to your HTTP call.

    POST /chrome/function

  • Write

    /chrome/download

    Runs a Puppeteer or Playwright script in Chrome and returns any files the browser downloaded during execution. the execution of puppeteer code, which is ran inside context of the browser. Browserless sets up a blank page, a fresh download directory, injects your puppeteer code, and then executes it. You can load external libraries via the "import" syntax, and import ESM-style modules that are written for execution inside of the browser. Once your script is finished, any downloaded files from Chromium are returned back with the appropriate content-type header.

    POST /chrome/download

  • Write

    /chrome/export

    Exports a webpage to a PDF or image format. This API is useful for generating reports, screenshots, or PDFs of web pages.

    POST /chrome/export

  • Write

    /chrome/unblock

    Unblocks the provided URL from being blocked due to bot detection. Returns a payload of Cookies, HTML, a base64 encoded screenshot, and a "browserWSEndpoint" to allow connecting to the browser when specified in the JSON Payload. Only supports CDP or Puppeteer like libraries when connecting to the "browserWSEndpoint".

    POST /chrome/unblock

  • Write

    /chrome

    Launch and connect to Chromium with a library like puppeteer or others that work over chrome-devtools-protocol.

    GET /chrome

  • Write

    /chrome/bql?(/*)

    Parses and executes BrowserQL requests, powered by the BrowserQL Editor or by other API integrations. See the BrowserQL Editor for more documentation on this API.

    POST /chrome/bql?(/*)

  • Write

    /chrome/bql?(/*)

    WebSocket endpoint for BrowserQL. Connect to establish a persistent browser session, then send multiple GraphQL operations as JSON messages. The browser and page state persist across all messages for the connection duration. Messages use the same JSON format as the HTTP POST endpoint: Send: { "query": "mutation { goto(url: \"..\") { status } }", "variables": {} } Receive: { "data": { "goto": { "status": 200 } } }

    GET /chrome/bql?(/*)

Agent

  • Read

    /agent/run

    List the caller's agent runs, newest first, with a short preview of each, using cursor pagination.

    GET /agent/run

  • Read

    /agent/run/*

    Poll an agent run: its status, the latest turn result and streamed step rationales, and the full turn transcript.

    GET /agent/run/*

  • Write

    /agent/run

    Start an asynchronous AI agent run: a managed agent drives a real browser against a natural-language task and returns a structured JSON result. Returns a run ID to poll via GET /agent/run/{id}.

    POST /agent/run

  • Write

    /agent/run/*/continue

    Continue a multi-turn agent run with a follow-up task. Not enabled in v1.

    POST /agent/run/*/continue

  • Write

    /chromium/agent

    WebSocket endpoint for agentic browser interaction. Connect to establish a persistent browser session, then send JSON-RPC messages to control the browser. Uses a simple JSON protocol instead of GraphQL: Send: { "id": 1, "method": "goto", "params": { "url": "https://example.com" } } Receive: { "id": 1, "result": { "status": 200, "url": "https://example.com" } } All BQL mutations are available as methods (goto, click, type, screenshot, etc.). Additionally supports a "snapshot" method that returns interactive page elements with CSS selectors, plus tab-management methods (getTabs, switchTab, createTab, closeTab) for multi-tab workflows.

    GET /chromium/agent

  • Destructive

    /agent/run/*

    Cancel an active agent run (including needs_human), aborting the in-flight turn, or delete a terminal one.

    DELETE /agent/run/*

Browser REST APIs

  • Read

    /batch-scrape/*

    Get batch scrape progress and paginated page results with content URLs.

    GET /batch-scrape/*

  • Read

    /scrapers

    Lists the scrapers and APIs available from the loaded scraper bundle, with the bundle version and each API's output schema. Call POST /scrapers/{scraper}/{api} to run one.

    GET /scrapers

  • Write

    /batch-scrape

    Start an async job over a list of URLs; poll GET /batch-scrape/{id}; optional webhooks.

    POST /batch-scrape

  • Write

    /scrapers/*/*

    Runs one API of a predefined scraper from the loaded scraper bundle: POST /scrapers/{scraper}/{api}, where {scraper} is the scraper name (an organisation slug such as zillow — never a hostname) and {api} is one of its APIs, both as listed by GET /scrapers. The body is the API's input ({ url?, params? }). A scraper that runs and reports a typed scrape error (blocked, not-found, invalid-input, upstream-changed, upstream-error) still returns HTTP 200 with ok:false. HTTP 4xx/5xx is reserved for host-level failures (unknown scraper/api, bundle unavailable, overloaded, job timeout, worker crash, job drained by a bundle hot swap, invalid result). Every ok:true result has been validated against the API's declared output schema.

    POST /scrapers/*/*

  • Destructive

    /batch-scrape/*

    Cancel a batch scrape job. Completed pages remain available; terminal jobs return 409.

    DELETE /batch-scrape/*

Management REST APIs

  • Read

    /browser/*/live/*

    Reports live URL status and completion reason until its browser closes.

    GET /browser/*/live/*

  • Read

    /chromium/agent/capabilities

    Returns the versioned capabilities declared by the Chromium agent route. Supply the intended route query parameters so authentication and plan checks can reject unsupported combinations before a browser is allocated.

    GET /chromium/agent/capabilities

  • Read

    /profile/*/download

    Downloads the captured authentication state for a specific profile. The profile name is the URL-encoded path segment before "/download". Returns 404 if no profile with that name exists for this token. **Note:** This endpoint is also available at: `/profile/.*/download` for backwards compatibility.

    GET /profile/*/download

  • Write

    /browser/*/pages

    Lists a running browser's open tabs. With mint=true, also returns a view-only, tab-pinned live URL per tab; a tab that already has a live URL returns it unchanged.

    GET /browser/*/pages

  • Destructive

    /browser/*/live/*

    Closes an active live URL owned by the specified browser. Returns 204 with no body.

    DELETE /browser/*/live/*

BQL Execution

  • Write

    /chromium/bql?(/*)

    Parses and executes BrowserQL requests, powered by the BrowserQL Editor or by other API integrations. See the BrowserQL Editor for more documentation on this API.

    POST /chromium/bql?(/*)

  • Write

    /chromium/bql?(/*)

    WebSocket endpoint for BrowserQL. Connect to establish a persistent browser session, then send multiple GraphQL operations as JSON messages. The browser and page state persist across all messages for the connection duration. Messages use the same JSON format as the HTTP POST endpoint: Send: { "query": "mutation { goto(url: \"..\") { status } }", "variables": {} } Receive: { "data": { "goto": { "status": 200 } } }

    GET /chromium/bql?(/*)

  • Write

    /stealth/bql?(/*)

    Parses and executes BrowserQL requests, powered by the BrowserQL Editor or by other API integrations. See the BrowserQL Editor for more documentation on this API.

    POST /stealth/bql?(/*)

  • Write

    /stealth/bql?(/*)

    WebSocket endpoint for BrowserQL. Connect to establish a persistent browser session, then send multiple GraphQL operations as JSON messages. The browser and page state persist across all messages for the connection duration. Messages use the same JSON format as the HTTP POST endpoint: Send: { "query": "mutation { goto(url: \"..\") { status } }", "variables": {} } Receive: { "data": { "goto": { "status": 200 } } }

    GET /stealth/bql?(/*)

CDP & DevTools Connections

  • Read

    /chromium/stealth/reconnect/*

    Reconnect to an existing stealth browser session (e.g. from /unblock) with CDP proxy features like solveCaptchas and stealth-mode fingerprint masking.

    GET /chromium/stealth/reconnect/*

  • Read

    /devtools/page/*

    Connect to an existing page in Chromium with a library like chrome-remote-interface or others that work the page websocketDebugger URL. You can get this unique URL by calling the /json/list API or by finding the page's unique ID from your library of choice.

    GET /devtools/page/*

  • Read

    /function/connect/*

    Internally used by the POST /function API to connect the underlying client-side code to. Not intended for direct use but documented for completeness and to distinguish between other reconnect style calls.

    GET /function/connect/*

  • Write

    /chromium/cli

    Interactive hosted-agent chat. Browserless runs the LLM server-side (no customer key) and drives a browser to complete the task. Send newline-JSON control frames ({"type":"start","prompt":"..."}) and receive a stream of data / tool_call / tool_result / usage / awaiting_input / end events.

    GET /chromium/cli

  • Write

    /devtools/browser/*

    Connect to an already-running Chromium process with a library like puppeteer, or others, that work over chrome-devtools-protocol. Chromium must already be launched in order to not return a 404.

    GET /devtools/browser/*

Platform / Skills

  • Read

    /skills

    Returns the browser-automation skills known for a site: tuned, site-specific recipes an agent should follow instead of planning its own steps. "domain" is required; "task" narrows the result to a single skill slug. Returns an empty array when no skill is known for the domain.

    GET /skills

See How connections work for access and permissions, or browse all integrations.

Reference date: October 29, 2026. Browserless API docs.