BrowserOxide MCP
A Rust browser engine without Chromium, with its own TLS fingerprint and V8; the MCP server gives agents pages and protection checks.
- 128 ★ on the repository
- Last commit
Broad access
Why this level
- Bypasses site anti-bot protection, which can breach their terms of use and the law.
- The `evaluate` tool runs JavaScript from third-party pages in a V8 engine.
Install
- Works in
Manual install
pip install browser-oxideThe README command installs the engine's Python bindings.
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- Bypasses site anti-bot protection, which can breach their terms of use and the law.
- The `evaluate` tool runs JavaScript from third-party pages in a V8 engine.
This is third-party code. Review the repository files before installing.
What it does
BrowserOxide is a browser engine written from scratch in Rust for scraping, archival and agents. It has JavaScript on V8, its own DOM and CSS, canvas, a BoringSSL TLS stack with a Chrome-matched fingerprint and configurable identity profiles (Chrome, Firefox, Pixel, iPhone Safari). There is no Chromium and no CDP driver underneath. The `browser-oxide-mcp` server gives an agent three tools: `fetch_page`, `evaluate` and `check_protection`. The last reports whether a URL sits behind protection such as Akamai, DataDome or Kasada and whether a real render got through. The same capabilities are available from Python through the `browser-oxide` package and through a CDP-compatible server for Puppeteer and Playwright.
Who it is for. Developers of scrapers and web-research agents who need a light browser with control over its fingerprint.
Good fit when
- You need to learn whether a URL is protected by Cloudflare, Akamai or DataDome and whether a real render got through.
- An agent collects public pages at volume, and Chrome is too heavy on memory.
- You test your own site or staging for anti-bot reactions with different browser profiles.
- A Python script needs a JavaScript-capable engine without installing Chromium.
Not a fit when
- The site forbids automated collection in its terms of use.
- The target is pages behind Kasada, which the project does not pass.
- You need Selenium or WebDriver, which are not supported.
Example request
Use check_protection to see whether https://example.com sits behind protection and whether a real render got through.MCP
- Transport
- stdio
- Authentication
- not required
Limitations
The project is research grade and pre-1.0, so the API is unstable. The README states MIT OR Apache-2.0 while GitHub shows NOASSERTION, so check the license files. In the author's own run on 126 protected sites a real render came through on 118. Kasada remains an open gap, the interactive Device-Check (Etsy) needs a human, and wildberries.ru did not pass. These numbers come from the author's benchmark and were not independently verified. Selenium and WebDriver are unsupported, and HTTP/3 is off by default. Apply the tool to your own sites and to data you have a right to, and follow service terms and the law.
How to disable. Remove the `browser-oxide` entry from the client's MCP configuration and delete the `browser-oxide-mcp` binary. Uninstall the Python package with `pip uninstall browser-oxide`.
FAQ
Which tools does the BrowserOxide MCP server have?
`fetch_page` loads a page, `evaluate` runs JavaScript on it and `check_protection` reports whether a URL sits behind protection such as Akamai, DataDome or Kasada.
Is Chromium required?
No. The engine is written from scratch in Rust, with neither Chromium nor a CDP driver underneath.
Does BrowserOxide pass Kasada?
No. Kasada remains an open gap and the project has no solver.
Does it work with Puppeteer and Playwright?
Yes, through the CDP-compatible WebSocket server `CdpServer::start_navigable`. Selenium and WebDriver are not supported.
Can I use BrowserOxide from Python?
Yes. The `browser-oxide` package installs with `pip install browser-oxide` and provides `Browser`, `Page`, `Profile` and `Verdict` classes; the GIL is released during navigation.
Discussion
No questions or reviews yet.
Related
MCP serversA Chromium with engine-level device fingerprint spoofing and an MCP server that lets an agent drive it through Playwright
Browserless.io MCP server: a cloud browser for the agent with page reads, site crawls, Lighthouse audits and click-driven sessions
A fast Rust CLI for agent browser automation: accessibility snapshots with element refs, an MCP server and skills
An MCP server that opens pages in headless Chromium through Playwright and hands the agent the main text as Markdown, even from JavaScript-heavy sites
macOS browser on WebKit and Rust: keyboard-first, with a built-in assistant and an MCP server an agent uses to drive the browser
Agent browser on direct CDP without Playwright: login profiles, a CLI, an MCP server and recording workflows into a skill
README in short
Covers the engine design, comparison with camoufox and nodriver, measurements on a corpus of 126 protected sites, usage from Rust and Python and the MCP server.