browserbase-sessions

✓Verified·Scanned 2/18/2026

This skill manages Browserbase cloud browser sessions (create, persist auth, navigate, screenshot, record) and stores named contexts in ~/.browserbase/contexts.json. It requires BROWSERBASE_API_KEY/BROWSERBASE_PROJECT_ID, runs local commands like python3 {baseDir}/scripts/browserbase_manager.py setup, and contacts https://api.browserbase.com and wss://connect.browserbase.com.

from clawhub.ai·v2.0.0·48.6 KB·0 installs

Scanned from 1.0.0 at 587386e · Transparency log ↗

$ vett add clawhub.ai/jamesfincher/browserbase-sessions

Browserbase Sessions — OpenClaw Skill

A production-ready OpenClaw skill for creating and managing persistent Browserbase cloud browser sessions with authentication persistence, automatic captcha solving, session recording, and screenshot capture.

What it does

This skill gives the OpenClaw agent the ability to:

Create cloud browser sessions via Browserbase's infrastructure
Persist authentication across sessions using Contexts (cookies, local storage, session storage are saved and restored automatically)
Solve CAPTCHAs automatically — login flows and protected pages work without manual intervention (enabled by default)
Record sessions as video — every session is recorded for later download and sharing (enabled by default)
Take screenshots — viewport or full-page, during navigation or on demand
Reconnect to keep-alive sessions that survive disconnections
Automate browsing — navigate, execute JavaScript, extract page content, read cookies
Manage session lifecycle — list, inspect, and terminate sessions

Requirements

Python 3.9+
browserbase Python package
playwright Python package + Chromium browser
A Browserbase account with API key and project ID

Quick Start

1. Copy the skill to your OpenClaw skills directory

cp -r browserbase-sessions/ ~/.openclaw/workspace/skills/browserbase-sessions/

2. Install dependencies

cd ~/.openclaw/workspace/skills/browserbase-sessions/scripts
pip install -r requirements.txt
playwright install chromium

3. Set credentials

export BROWSERBASE_API_KEY="bb_live_your_key_here"
export BROWSERBASE_PROJECT_ID="your-project-uuid-here"

4. Run the setup test

python3 ~/.openclaw/workspace/skills/browserbase-sessions/scripts/browserbase_manager.py setup

This validates everything end-to-end: credentials, SDK, Playwright, API connection, and runs a live smoke test (creates a session, navigates to example.com, terminates).

Available Commands

Command	Description
`setup`	Validate credentials and run full smoke test
`create-context`	Create a persistent context (with optional `--name`)
`delete-context`	Delete a context
`list-contexts`	List saved named contexts
`create-session`	Create a browser session (captchas + recording ON by default)
`list-sessions`	List all sessions
`get-session`	Get session details
`terminate-session`	Terminate a running session
`navigate`	Navigate to a URL (with optional screenshot and text extraction)
`screenshot`	Take a screenshot of the current page
`execute-js`	Execute JavaScript in a session
`get-cookies`	Get cookies from a session
`get-recording`	Download session recording video
`get-logs`	Get session logs
`live-url`	Get live debug URL for a running session

Key Design Decisions

Captcha solving ON by default. Browserbase's captcha solver handles reCAPTCHA, hCaptcha, and other challenges automatically. This means login flows and protected pages work without manual intervention. Disable with --no-solve-captchas.

Recording ON by default. Every session is recorded as a video. After terminating a session, download the recording with get-recording to review or share. Disable with --no-record.

Named contexts. Instead of remembering UUIDs, name your contexts (--name github, --name slack) and reference them by name anywhere a context ID is expected.

Keep-alive for research. Set --keep-alive for long research sessions. The browser survives network disconnections and persists until explicitly terminated — up to 6 hours.

Architecture

This skill follows the AgentSkills open standard (agentskills.io):

SKILL.md — Instructions the agent reads on-demand
scripts/browserbase_manager.py — CLI tool the agent executes via bash
references/api-quick-ref.md — API reference for deeper details

License

MIT