Get started
Browlix documentation
Run isolated browser profiles on your own Windows and Ubuntu machines and drive them from the web app, the API, MCP, the CLI or the TypeScript SDK.
Browlix keeps browser profiles — each with its own data, proxy and settings — in your workspace and opens them on machines you own that run the Browlix worker. Everything you can do in the web app is also available programmatically, through one operation catalog shared by every interface.
Start here
- Quickstart — from an API key to a Playwright-controlled browser and back, in seven calls.
- Where your code runs — the one rule to understand before you integrate: browser control is local to the machine running the session.
- Authentication — API keys, scopes and what a key can never do.
- For AI agents —
llms.txt, Markdown pages, MCP and the feedback channel.
Interfaces
| Interface | Use it for | Where |
|---|---|---|
| HTTP API v1 | Any language, servers, scripts | https://browlix.com/api/public/v1 |
| MCP server | AI agents (Claude Code, Codex, Cursor, VS Code…) | https://browlix.com/api/public/mcp |
| TypeScript SDK | Node ≥ 20, Bun, Deno, Workers | @browlix/sdk |
| CLI | Terminals and shell scripts | browlix |
| AdsPower-compatible Local API | Existing scripts written for AdsPower | On the machine |
All of them call the same operations with the same permissions, limits, idempotency and audit trail. The API reference is generated from that catalog, so it lists exactly what the server accepts.
Concepts
What Browlix does not promise
Browlix keeps each profile consistent with the machine it runs on and reports what the browser actually applied. It does not promise that a site will never detect automation or never restrict an account, and it must not be used to impersonate people or evade a platform's identity checks — see the acceptable use policy.