bb-browser CLI
A CLI and MCP server that lets an agent drive your real Chrome with its logged-in sessions: 103 commands across 36 platforms plus browser automation
- 6,265 ★ on the repository
- Last commit
- MIT
Broad access
Why this level
- Acts as your logged-in sessions and can operate on sites as you
- Runs arbitrary JS in your browser and captures network traffic
- Avoiding anti-bot detection may violate service terms
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add cli/bb-browserDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- Acts as your logged-in sessions and can operate on sites as you
- Runs arbitrary JS in your browser and captures network traffic
- Avoiding anti-bot detection may violate service terms
This is third-party code. Review the repository files before installing.
What it does
The tool connects to your real Chrome over CDP and acts as the already-authenticated user, so an agent can reach pages that require a login. At its core is the site system: adapters turn site features into commands, for example Twitter search, a Reddit feed, a YouTube transcript, stock quotes, job listings or a Wikipedia summary. There is also general automation: open a page, capture the accessibility tree, click, fill a field, run JS, fetch with your cookies, capture network traffic and take screenshots. It runs as a plain CLI or as an MCP server over stdio, output can be filtered with jq and split across tabs. Adapters are community maintained in a separate repository, and an agent can reverse-engineer and add a new site itself.
Who it is for. For developers and agent builders who need web access through their own logged-in browser.
Good fit when
- You need to gather data across several platforms on one topic
- You need to automate actions on a site you are already logged into
- You need to turn a site feature into a command for an agent
Not a fit when
- The site has an official API that fits better
- Automated access would break the service terms or rate limits
- You need to run in CI without a real browser
Example request
Search Twitter and arxiv for recent work on retrieval augmented generation and put together a summaryLimitations
You need a running Chrome with a local daemon and extension, or a setup with OpenClaw. The tool acts as your own account and its sessions, so following each site's terms is on you. The author states access works without triggering anti-bot detection, which can conflict with a service's terms. Adapters are community maintained and break when sites change their markup or internal requests.
How to disable. Stop the daemon and remove the package: npm uninstall -g bb-browser. If you added it as MCP, remove the bb-browser entry from your agent's config.
FAQ
Do I need an API key?
No. The tool works through your browser and its active sessions, so no separate site keys are needed.
How do I add my own site?
Run bb-browser guide. The agent reads the guide, reverse-engineers the site's requests, writes an adapter, tests it and can submit it to the shared repository.
Discussion
No questions or reviews yet.
Related
CLI toolsA local browser automation platform: MCP server, CLI and skills so an agent can drive Chrome, fill forms and pull data from pages
An agent drives your real logged-in Chrome: a CLI, MCP and an extension with a separate tab group per session
An MCP server and Chrome extension give any agent 22 tools for your own logged-in browser, verifying the effect of every action.
A remote MCP server and Chrome extension: the agent reads data behind your logins and operates the sites you already use
An MCP bridge to control your browser from a coding agent: a Chrome extension and a set of tab and page automation tools
A fast Rust CLI for agent browser automation: accessibility snapshots with element refs, an MCP server and skills
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Repository stars | Main difference |
|---|---|---|---|
| bb-browser (this card) | no | 6,265 | |
| Browser Harness | no | 19,048 | a skill and CLI from Browser Use |
| twitter-cli | no | 3,023 | a Twitter/X CLI with an agent skill |
| Interceptor | yes | 625 | a CLI and MCP server that give an agent your signed-in browser |
README in short
The README lays out the idea: instead of asking sites for a machine interface, let the machine use the human one, a real browser with the login already done. It shows the site system with per-platform commands, general browser automation and MCP setup for Claude Code and Cursor. It also describes an OpenClaw mode that needs no extension or daemon. The daemon listens on a local address by default and the host can be changed. MIT licensed, with README versions in English and Chinese.