Brimp

Warning

Brimp is in alpha stage, API is not stable and subject to change.

Getting the html with curl_cffi is easy, but oftentimes, you need to render the page with JavaScript, and some websites rejects you because you are not using a browser.

Brimp is the browser runtime for agents and bots, it tries to add the missing JS and DOM layer for curl_cffi. So you can get not only static html pages, but also the dynamically rendered ones, and you can even interact with the page, a more browser-ish experience.

Under the hood, brimp combines JavaScriptCore from WebKit and curl-impersonate as the network stack. But, if you are familiar with requests or curl_cffi, it feels like the same simple request/response workflow with a JavaScript-rendered HTML result. Brimp is written in Rust, and offers both Python and Node.js bindings.

Documentation: docs.brimp.ai

Features

• Page rendered with DOM and JS, so you don't get the empty html placeholder.
• curl-impersonate as the network stack, providing perfect Ja3/TLS, http 2&3 fingerprints.
• The familiar curl_cffi API and experience from the same maintainer.
• Much faster than playwright with Chromium, on par with popular alternatives.
• Plugin system for you to fill the gaps between any website that's blocking you.
• Python and Nodejs bindings.
• Pre-compiled, so you don't have to compile on your machine.
• MIT licensed.

	chromium	camoufox	cloakbrowser	lightpanda	obscura	brimp
JS & DOM	✅	✅	✅	✅	✅	✅
http/3	✅	✅	✅	❌	❌	✅
CDP	✅	✅	✅	☑️	☑️	❌1
requests-like	❌	❌	❌	❌	❌	✅
screenshot	✅	✅	✅	❌	✅	❌
JS engine	V8	SpiderMonkey	V8	V8	V8	JSC
open source	✅	✅	❌	☑️2	✅	✅
ja3 fingerprints	☑️3	☑️3	☑️3	❌	✅	✅
fast?	🐢	🐢	🐢	🐇	🐇	🐇

Notes:

1. This is a deliberate decision.
2. Lightpanda is under the AGPL license.
3. Fingerprints are fixed to the browser version

For performance comparisons, see the benchmark.

Install

For humans

CLI:

curl -fsSL https://brimp.ai/install.sh | sh

You can also download the prebuilt binary tarballs from github release page.

Python:

pip install brimp

Node.js:

npm install @brimp/brimp

For agents

Paste this prompt:

Install this tool: https://github.com/lexiforest/brimp

Usage

CLI

The CLI runs the browser runtime in-process.

Check the runtime, extract a live page, evaluate JavaScript, or save the HTML:

# Run a JavaScript expression on a page
brimp fetch https://example.com --eval 'document.title'

# Extract markdown
brimp fetch https://example.com --output example.md

# Capture a profile from a real browser
brimp profile capture --output browser-profile.json

# Use that profile to visit a website
brimp fetch https://example.com --profile browser-profile.json

Python

The Python binding is synchronous and in-process. page.text is the original HTTP response text; Page.html() is the live DOM serialized after page scripts execute.

import brimp

with brimp.Session() as session:
    page = session.get("https://example.com")
    print(page.status_code)
    print(page.html())

Use a Session to retain cookies and connections and to access the current rendered document:

import brimp

with brimp.Session() as session:
    page = session.get("https://example.com/?q=browser")
    page.raise_for_status()
    heading = page.query_selector("h1")
    if heading is not None:
        print(heading.text_content())
        heading.dispose()

Node.js

import { createSession } from '@brimp/brimp';

const session = await createSession();
try {
  const page = await session.get('https://example.com');
  console.log(await page.evaluate('document.title'));
} finally {
  await session.close();
}

Development

Architecture

The implemented runtime supports:

• static HTML, JavaScript DOM mutation, selectors, inline styles, and non-rendering CSSOM;
• HTTP(S) navigation through a transport-neutral ResourceLoader;
• linked CSS, classic inline/external scripts, defer, and async;
• parser pausing around blocking scripts;
• events, timers, microtasks, Promise-based fetch, cookies, Location, and Navigator.

The canonical owner-thread automation API is exposed through:

• the brimp CLI for evaluation and page extraction;
• the Python and Node.js Session/Page bindings.

The CLI and bindings delegate browser operations to brimp-runtime in-process. The runtime uses brimp-lexbor for DOM, brimp-jsc for JavaScriptCore, and brimp-net for networking.

CI

The project is tested and built automatically on GitHub Actions. All PR must pass the automatic tests.

Testing

cargo build -p brimp-cli -p brimp-python -p brimp-node
cargo test --workspace
cargo clippy --workspace --all-targets -- -D warnings

License

MIT.