Integration

Anchor Browser integration

Anchor Browser runs cloud browsers for AI agents and automation. Start and end browser sessions (proxies, stealth, captcha solving, profiles, recording), drive them with mouse, keyboard, clipboard and navigation actions, take screenshots and PDFs, have Anchor's AI agent perform a web task from a prompt, fetch any page's rendered HTML through the Web Unlocker, generate and run saved tasks, and manage profiles, volumes, extensions, CA certificates, batch sessions, webhooks and managed-login applications; read recordings, downloads, usage and billing. Sessions and AI tasks consume Anchor credits. Left out: the legacy v1 task routes, the Agent Access sign-up flow, creating or updating identities and the dynamic login flow (they carry login credentials), identity tokens, creating 1Password integrations, webhook secret rotation, pre-signed volume uploads, the web task's secret_values (credentials handed to the agent) and downloading a recording (that route redirects to a pre-signed link, which this catalog does not follow; list_recordings lists each recording's link); CDP connection URLs, webhook signing secrets and stored session configurations are not returned. Authentication is an API key from app.anchorbrowser.io (API Access) sent in the anchor-api-key header.

Connect Anchor Browser How integrations work

Ask it in chat

Say what should happen.
The workflow builds itself.

When this happens, do that in Anchor Browser.Built from the 104 Anchor Browser tools below, on the connection you add once.

What you can do with Anchor Browser

Each of these is one operation on one endpoint, available both as a workflow step and as a tool an AI agent can call on its own. Nothing bundles several operations into one action, so what a run did stays readable afterwards.

Most endpoints run on the free plan. A few, mainly the ones that publish on your behalf, need a paid plan.

Batch session, 5 operations

cancel_batch_sessionCancel batch sessionPATCHCancels a pending or running batch (status=cancelled), which terminates every running session in it.
create_batch_sessionsCreate batch sessionsPOSTCreates up to 1,000 browser sessions with the same configuration in one call. They are provisioned in the background: follow them with get_batch_session. Every session consumes Anchor credits.
delete_batch_sessionsDelete batch sessionsDELETEDeletes every session of a batch: running ones are terminated and all their session data is removed.
get_batch_sessionGet batch sessionGETRetrieves detailed status information for a specific batch, including progress, individual session details, and any errors that occurred.
list_batch_sessionsList batch sessionsGETLists all batch session requests for the authenticated team, with optional filtering by status and pagination support.

File, 5 operations

delete_volume_filesDelete volume filesDELETEDelete one file (`path`) or everything under a folder (`prefix` with `recursive=true`).
list_agent_resourcesList agent resourcesGETList all resources that have been uploaded to the browser session for agent use. Returns resource metadata including name, size, type, and last modified timestamp.
list_volume_filesList volume filesGETList the contents of a folder in the volume (one level, like `ls`), or the whole tree with `recursive=true`. Files written by a running session appear here about a minute after the session stops writing to them.
upload_agent_resourceUpload agent resourcePOSTUpload files as agent resources to a browser session using multipart/form-data.
upload_volume_fileUpload volume filePOSTUploads a file (up to 10 MB) into a persistent volume at the given path, optionally overwriting an existing file, and returns its path, size and type.

Webhook, 5 operations

create_webhookCreate webhookPOSTRegisters an HTTPS endpoint that Anchor POSTs the chosen events to (task, session and other events). The one-time signing secret Anchor returns is not passed on by this tool. Up to 5 webhooks per project.
delete_webhookDelete webhookDELETESoft-delete a webhook. New events stop fanning out immediately. The events history (`list_webhook_deliveries`) is still queryable for the retention window even after delete.
get_webhookGet webhookGETFetch a single webhook config. The signing secret is never returned.
list_webhooksList webhooksGETList every active webhook for the authenticated project. The signing secret is never included.
update_webhookUpdate webhookPATCHPartial update, pass only the fields you want to change. Setting `enabled: false` immediately stops events from fanning out to this webhook (existing in-flight deliveries finish their attempts).

Application, 4 operations

create_applicationCreate applicationPOSTCreates a managed-login application for a website (its login page URL, name and description). Identities saved for it let sessions start already signed in.
delete_applicationDelete applicationDELETEDeletes a managed-login application, identified by its ID.
get_applicationGet applicationGETReturns one managed-login application by ID: name, URL, description, number of identities, supported login methods and timestamps.
list_applicationsList applicationsGETLists the managed-login applications of your team (a website Anchor can sign in to) with name, URL, description, number of identities and supported login methods. search filters by name.

Auth flow, 4 operations

create_auth_flowCreate auth flowPOSTCreates an authentication flow for an application: a name and the login methods it uses (username_password, authenticator, custom), plus custom field names.
delete_auth_flowDelete auth flowDELETEDeletes an authentication flow of an application.
list_auth_flowsList auth flowsGETLists the authentication flows of an application: name, description, whether it is the recommended one, login methods (username_password, authenticator, custom) and custom field names.
update_auth_flowUpdate auth flowPATCHUpdates an authentication flow of an application; pass only the fields to change (name, description, recommended flag, methods, custom fields).

Extension, 4 operations

delete_extensionDelete extensionDELETEDeletes an uploaded browser extension and removes it from storage.
get_extensionGet extensionGETReturns one uploaded browser extension by ID: name, manifest (name, version, permissions...) and timestamps.
list_extensionsList extensionsGETLists the browser extensions uploaded to your account, with ID, name, manifest and timestamps.
upload_extensionUpload extensionPOSTUpload a new browser extension as a ZIP file. The extension will be validated and stored for use in browser sessions.

Profile, 4 operations

create_profileCreate profilePOSTCreates a new profile from a browser session. A Profile stores cookies, local storage, and cache.
delete_profileDelete profileDELETEDeletes a stored browser profile by its name.
get_profileGet profileGETReturns one stored browser profile by name: description, source session, status and creation time.
list_profilesList profilesGETLists the stored browser profiles (saved cookies, local storage and cache) with name, description, source session, status and creation time.

Session, 4 operations

end_sessionEnd sessionDELETEEnds (terminates) a browser session, identified by its ID.
get_sessionGet sessionGETReturns one browser session by ID: status, duration, credits used, proxy bytes, AI steps, token usage, tags and creation time. Its stored configuration is not returned.
list_sessionsList sessionsGETLists past and current browser sessions page by page, with status, tags, duration, credits, proxy use and steps. Filter by status, tags, domains, dates, batch, profile, proxy or stealth; sort by a field.
start_sessionStart sessionPOSTStarts a cloud browser session and waits until it is ready, returning its ID and live view URL. Options cover proxy, recording, timeouts, profile, ad and popup blocking, captcha solving, viewport and tags. Consumes Anchor credits.

Status, 4 operations

get_async_session_statusGet async session statusGETReturns the progress of a session started with start_session_async: status (pending, processing, ready, completed, failed, cancelled), provisioning phase and, once up, the session ID and live view URL.
get_task_run_statusGet task run statusGETReturns the status of a task run (queued, running, success, failure, timeout, cancelled) with its result or error and the session it used.
get_web_task_statusGet web task statusGETReturns the state of a web task started by perform_web_task with async=true: status RUNNING, COMPLETED (with result) or FAILED (with error), and the session it ran in.
list_session_statusesList session statusesGETLists the browser sessions of the account with ID, status, tags and creation time, filtered by tags, domains, creation time, batch, profile, proxy or stealth use.

Volume, 4 operations

create_volumeCreate volumePOSTCreate a persistent volume for your team. The volume is usually `ready` immediately; if the response says `creating`, poll `get_volume` until it is. Each team can have up to 10 volumes.
delete_volumeDelete volumeDELETEDelete a volume and every file in it. Refused while a running session has the volume mounted; end those sessions first.
get_volumeGet volumeGETGet a volume by its ID. Deleted volumes return `404`; use `list_volumes (with include_deleted=true)` to see them.
list_volumesList volumesGETList the volumes of your team.

Certificate, 3 operations

delete_certificateDelete certificateDELETEDelete a CA certificate by name. The certificate will be removed from storage and can no longer be used in new sessions.
list_certificatesList certificatesGETLists the CA certificates uploaded for your team, with ID, name, description and upload time.
upload_certificateUpload certificatePOSTUpload a CA certificate to trust in browser sessions. Certificates are managed at the team level. Once uploaded, reference the certificate by name when creating sessions to trust services using that CA.

Identity, 3 operations

delete_identityDelete identityDELETEDeletes an identity (a saved login for an application).
get_identityGet identityGETReturns one identity by ID: name, source login URL, status (pending, validated or failed), metadata and timestamps. Its credentials are not returned.
list_application_identitiesList application identitiesGETLists the identities (saved logins) of an application with name, auth flow, status (pending, validated, failed) and timestamps. Filter by name or by metadata key/value pairs.

Tag, 3 operations

clear_session_tagsClear session tagsDELETEClears all tags from a browser session.
get_session_tagsGet session tagsGETRetrieves the tags associated with a specific browser session.
set_session_tagsSet session tagsPUTReplaces the tags of a browser session with the given list (existing tags are overwritten).

Clipboard, 2 operations

get_clipboardGet clipboardGETReturns the current text content of a browser session's clipboard.
set_clipboardSet clipboardPOSTSets the text content of a browser session's clipboard.

Download, 2 operations

get_volume_file_download_urlGet volume file download urlGETReturns a pre-signed download URL for one file of a volume, with its expiry, size and content type.
list_session_downloadsList session downloadsGETLists the files downloaded during a browser session, with their metadata.

Event, 2 operations

list_webhook_deliveriesList webhook deliveriesGETLists the delivery attempts of a webhook (pending, succeeded, failed or dead) with event type, attempt count and response status. Filter by status, event type or time; 7 days of history.
signal_eventSignal eventPOSTSignals an event with associated data, unblocking any clients waiting for this event. This enables coordination between different browser sessions, workflows, or external processes.

Integration, 2 operations

delete_integrationDelete integrationDELETEDeletes an integration and the credentials Anchor stored for it.
list_integrationsList integrationsGETLists your team's integrations (for example a 1Password integration) with ID, name, type, storage path and creation time; their credentials are not returned.

Pause, 2 operations

pause_agentPause agentPOSTPauses the AI agent for the specified browser session.
pause_recordingPause recordingPOSTPauses the video recording for the specified browser session.

Recording, 2 operations

delete_recordingDelete recordingDELETEDeletes a specific recording from a browser session. Use "primary" as the recording_id to delete the primary recording.
list_recordingsList recordingsGETLists the video recordings of a browser session with their URLs.

Resume, 2 operations

resume_agentResume agentPOSTResumes the AI agent for the specified browser session.
resume_recordingResume recordingPOSTResumes the video recording for the specified browser session.

Screenshot, 2 operations

screenshot_webpageScreenshot webpagePOSTCaptures a screenshot of a web page (or of a running session when sessionId is given) with viewport size, quality, wait time and full-page options, and returns the PNG as a stored file. Consumes Anchor credits.
take_session_screenshotTake session screenshotGETTakes a screenshot of a running browser session and returns the PNG as a stored file.

All, 1 operation

end_all_sessionsEnd all sessionsDELETEEnds every active browser session of the API key's account at once.

Async, 1 operation

start_session_asyncStart session asyncPOSTStarts a cloud browser session without waiting: returns a request_id at once while the browser is provisioned. Call get_async_session_status until status is ready to get the session ID and live view URL. Consumes Anchor credits.

Billing, 1 operation

get_billing_infoGet billing infoGETReturns the account's credit balance, credits used this period, included credits, billing period and tier; with from_date, to_date or granularity it adds a credits-used time series.

By name, 1 operation

get_volume_by_nameGet volume by nameGETGet a volume by its name.

Click, 1 operation

mouse_clickMouse clickPOSTClicks in a browser session, either at x/y coordinates or on the element matching a CSS selector (pass one of the two), with the left, middle or right button.

Copy, 1 operation

copy_selected_textCopy selected textPOSTCopies the text currently selected in a browser session to its clipboard and returns it.

Directory, 1 operation

create_volume_directoryCreate volume directoryPOSTCreate an empty folder in the volume. Folders are also created implicitly when a file is uploaded to a path inside them.

Double click, 1 operation

mouse_double_clickMouse double clickPOSTPerforms a double click at the specified coordinates

Down, 1 operation

mouse_downMouse downPOSTPerforms a mouse button down action at the specified coordinates

Drag and drop, 1 operation

drag_and_dropDrag and dropPOSTPerforms a drag and drop operation from start coordinates to end coordinates

Execute code, 1 operation

execute_codeExecute codePOSTRuns a TypeScript snippet (an async function body given page, browser, context, sessionId and anchorClient) against a running browser session and returns its JSON-encoded result.

Generate, 1 operation

generate_taskGenerate taskPOSTGenerates a reusable browser task from a natural-language prompt, with optional input and output parameters, identity and browser defaults. Returns the task ID; get_task_generation_status reports when it is ready.

Generation status, 1 operation

get_task_generation_statusGet task generation statusGETReturns the generation status of a task created by generate_task, with any error.

Goto, 1 operation

navigate_to_urlNavigate to urlPOSTNavigates a running browser session to the given URL.

History, 1 operation

get_session_usage_historyGet session usage historyGETReturns session usage analytics over a date range: per period (hour, day, week or month) the session count, proxy bytes, AI steps, session hours and credits used, plus totals.

Move, 1 operation

mouse_moveMouse movePOSTMoves the mouse cursor to the specified coordinates

Page, 1 operation

list_session_pagesList session pagesGETLists the open pages (tabs) of a browser session with ID, title, URL and a frontend URL to view the page.

Page pdf, 1 operation

render_page_pdfRender page pdfPOSTRenders a web page to PDF (in an existing session when sessionId is given) with paper size, margins, scale, orientation, header and footer options, and returns the PDF as a stored file. Consumes Anchor credits.

Paste, 1 operation

paste_textPaste textPOSTPastes the given text at the current cursor position of a browser session.

Path, 1 operation

list_volume_pathsList volume pathsGETFlat list of every path in the volume, optionally under a prefix. Folders appear with a trailing slash (for example `downloads/`).

Pending intervention, 1 operation

list_pending_interventionsList pending interventionsGETReturns all pending human-in-the-loop (HITL) requests across the team's active browser sessions. Each item includes the intervention request plus the session ID and tags for routing responses.

Perform web task, 1 operation

perform_web_taskPerform web taskPOSTHas Anchor's AI agent carry out a natural-language task in a browser on real websites, from a URL or a running session, and returns its result; with async=true, a workflow_id for get_web_task_status. Consumes Anchor credits.

Reauthenticate, 1 operation

reauthenticate_identityReauthenticate identityPOSTRe-authenticates an identity: validates its saved browser profile, signs in again if it is no longer signed in and saves the refreshed profile, in a temporary browser session that consumes Anchor credits.

Requested human intervention, 1 operation

list_session_interventionsList session interventionsGETRetrieves pending human-in-the-loop (HITL) requests for a specific active browser session. Requires `human_intervention: true` on the task that is running in the session.

Respond to human intervention, 1 operation

respond_to_interventionRespond to interventionPOSTAnswers a pending human-in-the-loop request of a session's AI agent, which then continues. Without requestId the most recent pending request is answered.

Retry, 1 operation

retry_batch_sessionsRetry batch sessionsPOSTRetries the failed sessions of a batch by provisioning new browsers for them; the batch goes back to processing. max_retries caps the attempts per session. The new browsers consume Anchor credits.

Run, 1 operation

run_taskRun taskPOSTRuns a saved task by ID with its input parameters, optionally in an existing session or with an identity. Returns the run ID and status; with sync=true it waits and returns the result. Each run consumes Anchor credits.

Scroll, 1 operation

scrollScrollPOSTPerforms a scroll action at the specified coordinates

Shortcut, 1 operation

press_keyboard_shortcutPress keyboard shortcutPOSTPresses a combination of keys at once in a browser session (for example Control and A), optionally holding them for a given time.

Stat, 1 operation

get_volume_file_infoGet volume file infoGETSize, type and modification time of one file.

Test, 1 operation

send_webhook_test_eventSend webhook test eventPOSTFire a synthetic event of the given type through the real signing + retry path. Useful during development for confirming your endpoint can verify the signature and is reachable from Anchor's egress IPs.

Type, 1 operation

type_textType textPOSTTypes the given text into the focused element of a browser session, with an optional delay between keystrokes.

Up, 1 operation

mouse_upMouse upPOSTPerforms a mouse button up action at the specified coordinates

Upload, 1 operation

upload_session_fileUpload session filePOSTUpload files directly to a browser session for use with web forms and file inputs. Files are saved to the session's uploads directory and can be referenced in CDP commands (for example `DOM.setFileInputFiles`).

Wait, 1 operation

wait_for_eventWait for eventPOSTWaits for a specific event to be signaled by another process, workflow, or session. This endpoint blocks until the event is signaled or the timeout is reached. Useful for coordinating between multiple browser sessions or workflows.

Webpage, 1 operation

fetch_webpageFetch webpagePOSTFetches the fully rendered page of any URL, solving CAPTCHA and bot challenges, without a browser session, and returns the body as text (HTML, JSON, XML or plain text). 0.005 credits per successful fetch.

Use Anchor Browser from Claude Code, Cursor or Codex

Connect Anchor Browser in LiveContext, create an API key in Settings, then add the LiveContext MCP server to your client. Its Anchor Browser operations become tools the client can call, next to every other integration you connected.

Claude Code

claude mcp add --transport http livecontext https://livecontext.ai/mcp --header "X-API-Key: YOUR_API_KEY"

Codex (~/.codex/config.toml)

[mcp_servers.livecontext]
url = "https://livecontext.ai/mcp"

[mcp_servers.livecontext.http_headers]
"X-API-Key" = "YOUR_API_KEY"

Cursor, Claude Desktop and other JSON configs

{
  "mcpServers": {
    "livecontext": {
      "type": "http",
      "url": "https://livecontext.ai/mcp",
      "headers": {
        "X-API-Key": "YOUR_API_KEY"
      }
    }
  }
}

Set up the MCP server step by step

How to connect Anchor Browser

Create your workspace
Sign up free. Nothing to install.
Connect Anchor Browser once
Add your Anchor Browser connection in Settings. Every tool from this integration reuses it, and secrets are encrypted at rest and visible only to you.
Describe the job in chat
The workflow builds itself in front of you and picks the endpoints it needs. Drag the steps yourself on the canvas whenever you would rather be explicit.