Frequently asked questions

Clawbrowser is a Chromium fork with native patches for browser fingerprint profiles and profile-bound residential/datacenter proxy routing. It exposes a standard Chrome DevTools Protocol (CDP) endpoint, making it compatible with Playwright, Puppeteer, and CDP-based automation tools.

Unlike regular Chromium, Clawbrowser changes fingerprint behavior inside the browser engine instead of relying on page-level JavaScript injection. Compared to traditional profile browsers, the public launcher is built for automation: it manages named sessions, prints a local CDP endpoint, and keeps residential/datacenter proxy credentials tied to the generated profile.

Clawbrowser patches surfaces such as Canvas, WebGL, AudioContext, navigator.* properties (userAgent, platform, languages, hardware concurrency, device memory), screen resolution, fonts, timezone, language, battery status, plugins, media devices, WebRTC-related behavior, and speech synthesis voices. Values are loaded from the generated profile so related surfaces stay consistent.

Normally, no for proxy-backed profiles. Residential or datacenter proxy credentials are bundled with the generated fingerprint profile via the Clawbrowser API. The browser reuses the cached fingerprint profile until that profile is regenerated.

Clawbrowser is designed to reduce CAPTCHA and anti-bot interruptions caused by mismatched browser fingerprint, proxy, locale, timezone, and geo signals. Generated profiles keep those signals aligned with residential or datacenter proxy routing, but no browser can honestly guarantee a universal CAPTCHA bypass across every website.

Sign up at app.clawbrowser.ai. After creating an account, your API key is available in the dashboard. Set CLAWBROWSER_API_KEY or let the launcher prompt once and save it to ~/.config/clawbrowser/config.json.

Yes. Start a managed session, read its endpoint, then connect your framework to that CDP URL: clawbrowser start --session work -- https://example.com clawbrowser endpoint --session work Playwright: browser = await chromium.connectOverCDP(endpoint) Puppeteer: browser = await puppeteer.connect({ browserURL: endpoint })

Clawbrowser is available on macOS (desktop app), Linux (container/headless runtime), and Windows. For VPS or CI deployments, use the Docker-backed launcher/container image: docker.io/clawbrowser/clawbrowser:latest.

Run the install script with the correct target: curl -fsSL https://raw.githubusercontent.com/clawbrowser/clawbrowser/main/scripts/install.sh | bash -s -- claude Targets: claude (Claude Code / Claude Desktop), codex, gemini, all (Cursor and others).

The launcher exits non-zero if the browser does not expose a ready CDP endpoint. For fingerprint-backed sessions, clawbrowser rotate --session <name> restarts the session and passes --regenerate to the browser. If you need country, city, or connection-type targeting, pass browser flags after -- when starting the session.

Yes. Launch separate managed sessions with different session names and ports, for example: clawbrowser start --session agent-us --port 9222 -- https://example.com and clawbrowser start --session agent-de --port 9223 -- https://example.com. Each session has its own CDP endpoint. For identity separation, use distinct fingerprint IDs after --.

Yes. For VPS, CI, or no-display environments, use the Docker-backed launcher/container image: docker.io/clawbrowser/clawbrowser:latest. Browser flags can still be passed after -- when needed.

Releases are published on GitHub: github.com/clawbrowser/clawbrowser/releases. By default, the install script installs the current stable release.