Browserbase connector for AI agents

Tools your browser agent reaches for on Browserbase, scoped per user.

CALL ANY TOOL

38 tools for cloud browsers in Browserbase: start and release sessions, fetch pages and search the web, create and run browser agents, invoke Functions, save contexts, and read session logs, replays, recordings, downloads, and usage.

browserbase_create_agent

Create agent

Create a reusable Browserbase custom agent with a name, a system prompt and a result schema. Returns the new agent's agentId, name, systemPrompt, resultSchema, createdAt and updatedAt. Use this to save instructions you run often, then pass the agentId to browserbase_run_agent. Use browserbase_update_agent to change an existing agent.

Parameters

Name

Type

Required

Description

name

string

Required

Name for the agent, 1 to 255 characters, shown in the dashboard and API responses. Example: Pricing checker

result_schema

object

Optional

JSON Schema that runs of this agent should return their result in. Runs can override it. Example: {"type": "object", "properties": {"price": {"type": "number"}}}

system_prompt

string

Optional

Instructions that steer the agent on every run that uses it. Example: You check competitor pricing pages and report prices in USD.

browserbase_create_context

Create context

browserbase_create_session

Create session

Start a new Browserbase cloud browser session, with optional saved context, proxies, stealth (Verified) mode, captcha solving, viewport, region, time limit and metadata. Returns the session's id, status, region, expiresAt, connectUrl (the Chrome DevTools Protocol WebSocket URL), seleniumRemoteUrl and signingKey. Use this to get a browser. Drive it by connecting Playwright, Puppeteer, Selenium or Stagehand to connectUrl from your own code; that WebSocket connection cannot be made through a tool call. Use browserbase_get_session_live_urls to watch it, and browserbase_release_session to end it.

Parameters

Name

Type

Required

Description

allowed_domains

array

Optional

Restrict top-level page navigation to these domains and their subdomains (example.com also allows www.example.com). Iframes and other in-page requests are not blocked. Leave empty for no restriction. Example: ["example.com", "github.com"]

block_ads

boolean

Optional

Set to true to block ads in the browser. Defaults to false.

captcha_image_selector

string

Optional

CSS selector of a custom captcha image, for captchas Browserbase does not detect on its own. Set together with captcha_input_selector. Example: #captcha-image

captcha_input_selector

string

Optional

CSS selector of the input field where the custom captcha answer goes. Set together with captcha_image_selector. Example: #captcha-input

context_id

string

Optional

ID of a saved context (from browserbase_create_context) whose cookies, local storage and logins the browser should load. Example: 2b7e4c1a-9d3f-4e58-b6a0-7c8d9e0f1a2b

extension_id

string

Optional

ID of a Chrome extension already uploaded to Browserbase, to load in the browser. Example: 4c3b2a19-0817-4f6e-9d5c-4b3a29180716

ignore_certificate_errors

boolean

Optional

Set to true to ignore TLS certificate errors, for example to reach hosts with self-signed or expired certificates. Defaults to false.

keep_alive

boolean

Optional

Set to true to keep the session running after the client disconnects, so you can reconnect later. You must then end it with browserbase_release_session. Available on paid plans (Hobby and above).

log_session

boolean

Optional

Set to false to turn off session logging (the CDP log browserbase_get_session_logs returns). Defaults to true.

string

Optional

Operating system the browser presents as in stealth mode: windows, mac, linux, mobile or tablet. One of: `windows`, `mac`, `linux`, `mobile`, `tablet`.

persist_context

boolean

Optional

Only used with context_id. Set to true to save the browser's cookies and storage back into the context when the session ends, so later sessions stay logged in. Defaults to false.

project_id

string

Optional

Project to run the session in. Leave empty to use the project that owns the API key. browserbase_list_projects lists your project IDs. Example: 8c9d0e1f-2a3b-4c5d-8e6f-7a8b9c0d1e2f

proxy_configs

array

Optional

List of proxy rules, applied in order, for finer control than use_proxy. Each rule has a type: browserbase (Browserbase's managed proxy, optionally with a geolocation that needs a two-letter country and can add a US state code or city), external (your own proxy server URL, with optional username and password), or none (connect directly). Any rule can also take a domainPattern so it only applies to matching domains. Example: [{"type": "browserbase", "geolocation": {"country": "US", "state": "NY", "city": "New York"}}, {"type": "none", "domainPattern": ".*\\.gov"}]

record_session

boolean

Optional

Set to false to turn off session recording, which powers replays and MP4 recording downloads. Defaults to true.

region

string

Optional

Region to run the browser in: us-west-2 (default), us-east-1, eu-central-1 or ap-southeast-1. One of: `us-west-2`, `us-east-1`, `eu-central-1`, `ap-southeast-1`.

solve_captchas

boolean

Optional

Set to false to turn off automatic captcha solving. Defaults to true.

timeout

integer

Optional

Seconds after which the session ends on its own, from 60 to 21600 (6 hours). Defaults to the project's defaultTimeout. Example: 600

use_proxy

boolean

Optional

Set to true to route the browser's traffic through Browserbase's managed residential proxy network with its default settings. Ignored when proxy_configs is set. Leave empty for no proxy.

user_metadata

object

Optional

Your own key-value data to attach to the session, which browserbase_list_sessions can filter on. Example: {"customer": "acme", "job": "price-check"}

verified

boolean

Optional

Set to true to use Verified Browser mode, Browserbase's stealth browser for sites with strong bot detection. Needs the Scale plan.

viewport_height

integer

Optional

Browser viewport height in pixels. Set together with viewport_width. Example: 1080

viewport_width

integer

Optional

Browser viewport width in pixels. Set together with viewport_height. Example: 1920

browserbase_delete_agent

Delete agent

browserbase_delete_context

Delete context

browserbase_delete_download

Download delete

browserbase_fetch_page

Fetch page

browserbase_get_agent

Get agent

browserbase_get_agent_run

Run get agent

browserbase_get_context

Get context

browserbase_get_download

Download get

browserbase_get_function

Get function

browserbase_get_function_invocation

Get function invocation

browserbase_get_function_invocation_logs

Get function invocation logs

browserbase_get_project

Get project

browserbase_get_session

Get session

browserbase_get_session_live_urls

Get session live urls

browserbase_get_session_logs

Get session logs

browserbase_invoke_function

Invoke function

browserbase_list_agent_run_messages

List agent run messages

browserbase_list_agent_runs

List agent runs

browserbase_list_agents

List agents

browserbase_list_projects

List projects

browserbase_list_session_downloads

List session downloads

browserbase_list_session_recording_downloads

List session recording downloads

browserbase_release_session

Release session

browserbase_resume_agent_run

Run resume agent

browserbase_run_agent

Run agent

Start a Browserbase browser agent that completes a natural-language task using web search and its own browser session, optionally as a saved custom agent and with a result schema. Returns the new run's runId, agentId, task, status (PENDING, RUNNING, COMPLETED, FAILED, STOPPED, TIMED_OUT or PAUSED), sessionId, result, cause, and created, started and ended times; the run starts as PENDING. Use this to hand a whole browsing task to Browserbase. Poll browserbase_get_agent_run for the result and use browserbase_list_agent_run_messages to follow its steps. Use browserbase_create_session instead to drive a browser yourself.

Parameters

Name

Type

Required

Description

task

string

Required

What the agent should do, in plain language. Example: Find the current price of the Browserbase Developer plan on browserbase.com/pricing

agent_id

string

Optional

ID of a custom agent (from browserbase_create_agent) whose system prompt and result schema to use. Leave empty for an ad-hoc run. Example: a1b2c3d4-e5f6-4a7b-8c9d-0e1f2a3b4c5d

context_id

string

Optional

ID of a saved context (from browserbase_create_context) for the agent's browser to load. Example: 2b7e4c1a-9d3f-4e58-b6a0-7c8d9e0f1a2b

pause_when

string

Optional

When the agent should pause and wait for your input, such as a verification code or an approval. The run then goes to PAUSED until you call browserbase_resume_agent_run. Example: Pause when the site asks for a one-time passcode

persist_context

boolean

Optional

Only used with context_id. Set to true to save the browser's cookies and storage back into the context when the run ends. Defaults to false.

proxy_configs

array

Optional

result_schema

object

Optional

JSON Schema the agent's result should follow when the run finishes. Overrides the agent's own resultSchema for this run. Example: {"type": "object", "properties": {"price": {"type": "number"}}}

use_proxy

boolean

Optional

Set to true to route the browser's traffic through Browserbase's managed residential proxy network with its default settings. Ignored when proxy_configs is set. Leave empty for no proxy.

variables

object

Optional

Named values the agent can use as %name% placeholders in the task, so secrets like passwords stay out of the task text. Each key maps to an object with a value and an optional description that tells the agent when to use it. Values are not stored. Example: {"password": {"value": "s3cret", "description": "Login password for the portal"}}

verified

boolean

Optional

Set to true to use Browserbase Verified (stealth) mode for the agent's browser. Needs the Scale plan.

browserbase_search_web

Search web

browserbase_update_agent

Update agent

For more tools, view docs.

Build your Agent

Same auth pattern across LangChain, OpenAI, Anthropic, and Google ADK.

import { ScalekitClient } from "@scalekit-sdk/node";
import { createReactAgent } from "@langchain/langgraph/prebuilt";

const sk = new ScalekitClient(env.SCALEKIT_ENV_URL, env.SCALEKIT_CLIENT_ID, env.SCALEKIT_CLIENT_SECRET);

// Browserbase tools scoped to this user
const { tools } = await sk.tools.listScopedTools("user_123", {
  filter: { connectionNames: ["browserbase"], toolNames: [
    "browserbase_run_agent",
    "browserbase_get_agent_run",
    "browserbase_list_agent_run_messages"] },
  pageSize: 100,
});

const agent = createReactAgent({ llm, tools });
await agent.invoke({ messages: [{ role: "user", content: "Find the pricing tiers on globex.com and summarize them" }] });

import OpenAI from "openai";
import { ScalekitClient } from "@scalekit-sdk/node";

const sk = new ScalekitClient(env.SCALEKIT_ENV_URL, env.SCALEKIT_CLIENT_ID, env.SCALEKIT_CLIENT_SECRET);
const openai = new OpenAI();

const { tools } = await sk.tools.listScopedTools("user_123", {
  filter: { connectionNames: ["browserbase"] }, pageSize: 100,
});

const res = await openai.chat.completions.create({
  model: "gpt-5",
  messages: [{ role: "user", content: "Find the pricing tiers on globex.com and summarize them" }],
  tools,
});

// Execute the tool call with the user's vaulted Browserbase API key
await sk.tools.executeTool(res.choices[0].message.tool_calls[0], "user_123");

import Anthropic from "@anthropic-ai/sdk";
import { ScalekitClient } from "@scalekit-sdk/node";

const sk = new ScalekitClient(env.SCALEKIT_ENV_URL, env.SCALEKIT_CLIENT_ID, env.SCALEKIT_CLIENT_SECRET);
const anthropic = new Anthropic();

const { tools } = await sk.tools.listScopedTools("user_123", {
  filter: { connectionNames: ["browserbase"] }, pageSize: 100,
});

const msg = await anthropic.messages.create({
  model: "claude-sonnet-5",
  max_tokens: 1024,
  messages: [{ role: "user", content: "Find the pricing tiers on globex.com and summarize them" }],
  tools,
});

// Tool call runs with the user's vaulted Browserbase API key
await sk.tools.executeTool(msg.content, "user_123");

import { Agent } from "@google/adk/agents";
import { ScalekitClient } from "@scalekit-sdk/node";

const sk = new ScalekitClient(env.SCALEKIT_ENV_URL, env.SCALEKIT_CLIENT_ID, env.SCALEKIT_CLIENT_SECRET);

const { tools } = await sk.tools.listScopedTools("user_123", {
  filter: { connectionNames: ["browserbase"] }, pageSize: 100,
});

const agent = new Agent({
  name: "browser_agent",
  model: "gemini-2.5-pro",
  instruction: "Browserbase tools scoped to this user",
  tools,
});

await agent.run("Find the pricing tiers on globex.com and summarize them");

Try these prompts

Copy any prompt into your agent. Each maps directly to a Browserbase tool. Click to copy, paste into your agent, done.

Sessions

Copy the prompt

Copied

Start a Browserbase session with my saved Acme login context.

Copy the prompt

Copied

Get the live view URL for the running session.

Copy the prompt

Copied

Release session abc123 now.

Agents and pages

Copy the prompt

Copied

Run a browser agent to find the pricing tiers on globex.com.

Copy the prompt

Copied

Fetch https://acme.com/pricing as markdown.

Copy the prompt

Copied

Search the web for recent posts about Browserbase Functions.

Debugging and usage

Copy the prompt

Copied

Show the replay of last night's failed session.

Copy the prompt

Copied

List the files that session downloaded.

Copy the prompt

Copied

How many browser minutes has our project used so far?

SEE HOW AUTH WORKS

Each user connects their own Browserbase API key once, and Scalekit sends it with every call. API keys stay vaulted, every call is scope checked, and every action is logged.

Authorize

Your user connects

Browserbase

once. We tie it to their identity and the meetings they approved — no shared bot account, no org-wide access

Who:

user ‘A’

when:

Once per user

access:

Limited to user

Store

Their

Browserbase

token lives in a vault scoped to them. User A's meetings are never reachable by an agent acting for user B, even on the same connection

vault:

encrypted

scope:

per-user

tokens:

auto-refreshed

Resolve

When your agent calls a

Browserbase

tool, we fetch the right token server-side. It never touches your agent, never appears in the LLM context, never shows up in your logs

speed:

~40ms

check:

before every call

seen by:

nobody

Audit

Every

Browserbase

tool call is logged — who triggered it, which meeting was fetched, what came back. 90 days of history, tied to the user who authorized it

history:

90 days

export:

SIEM-ready

logged:

every call

Test other agents

See the same per-user auth pattern across other developer and automation connectors.

GTM and RevOps Teams

Competitive intelligence briefing agent

Scans Gong calls for competitor mentions, matches each one to its Notion battlecard, and DMs every affected rep a single Slack digest per cycle. Every call runs as the PMM who owns the briefing, never a shared bot.

Engineering Teams

DevOps assistant agent

Polls GitHub for failing checks and stale PRs, opens Linear issues for the ones that need work, and posts a daily digest to Slack. It acts as the engineer, not a shared service account.

Why Scalekit

Secure your agent's access. Connectors ship in minutes

01.

Shared tokens break per-user analytics

A shared Browserbase API key looks fine in a demo. In production every session and agent run looks like one service account, and you cannot tell which user triggered it. Scalekit resolves the credential of the actual user who triggered the agent, never a shared bot.

// shared key
audit → bot_service_account

// scalekit
audit → user_abc ✓

02.

Authentication is not authorization

03.

Multi-tenancy is architectural

04.

Browserbase today. Ten connectors tomorrow.

“Our agents act across Salesforce, Gong, Google Drive, and more, on behalf of every customer. Scalekit behind the scenes meant we can keep adding tools without ever rebuilding how credentials or tool calling work.”

Venu Madhav Kattagoni

Head of Engineering / Von

FAQs

Frequently Asked Questions

Does the agent access Browserbase as the user or through a shared API key?

As the user. Each user connects their own Browserbase API key once, and Scalekit sends it with every call. Audit logs attribute every session, agent run, and Function invocation to that user, not a shared service account.

Where is the Browserbase API key stored?

In Scalekit's managed AES-256 token vault, namespaced per tenant. To rotate it, the user reconnects with a new API key. Revocation is a single dashboard action. Credentials never appear in prompts, logs, or LLM context.

Can I limit what the agent does in Browserbase?

Yes. Filter by tool name in listScopedTools to expose only what you want, for example the session and replay read tools without browserbase_delete_context, which permanently deletes the cookies, storage, and logins saved in a context. Scalekit also enforces scope checks before every API call.

What happens when a user revokes Browserbase access?

The connection is invalidated on the next tool call. Subsequent requests for that user fail closed with a clear error. Other users in the tenant remain unaffected. The event is logged for audit.

Can the agent run a browser task without driving the browser step by step?

Yes. browserbase_run_agent starts a Browserbase browser agent that completes a natural-language task, browserbase_get_agent_run tracks its status, and browserbase_list_agent_run_messages returns what it did. For a single page, browserbase_fetch_page fetches a URL without starting a browser.