Lexmount Moli

Moli is an open-source headless browser built specifically for AI agents by Lexmount. Written in Rust and released under Apache-2.0 on GitHub (lexmount/moli), it describes itself as the "Best headless browser for AI agents. Lite, Fast, High-Compatibility."

The repo was created in August 2026 and, as of October 25, 2026, has about 20.5k stars and 553 forks; the latest release is v1.1.19 (2026-10-21). The same day @LexmountAI announced on X that Moli had passed 20,000 stars and that Moli is now available in the cloud as a hosted browser mode of Lexmount Cloud Browser.

Design: structure first, pixels on demand

Most browser-automation tasks need page structure, not a continuously rendered picture. Moli treats the native DOM and style state as the single source of truth:

• Extract HTML/Markdown, query the DOM, run JS, inspect network/storage: reads runtime state directly, with no layout or paint
• Read an element's box, hit-test coordinates, send coordinate input: runs one layout pass and keeps only the latest frozen layout tree
• Screenshot: rebuilds from the current DOM/style, renders a fresh frame, then discards paint state
• Screencast: emits no frame while state is clean; rebuilds only when something changed

Capabilities

• Complete web runtime: streaming HTML parsing, native DOM, V8 JavaScript, iframes and workers, CSS cascade, Fetch/XHR/WebSocket, cookies, WebCrypto, and storage (localStorage, IndexedDB, OPFS)
• Extraction-optimized outputs: the CLI produces HTML, Markdown, JSON and semantic text trees
• One automation binary: the same endpoint speaks CDP, WebDriver Classic and WebDriver BiDi — no separate ChromeDriver, geckodriver or browser install
• Visual features on demand: --layout enables full layout, hit-testing and screenshots; --resource also fetches images, fonts, audio/video
• Runs on Linux, macOS and Windows

Where it fits

Lexmount says this cost model suits crawling, browser-use agents, retrieval pipelines, evaluation environments and reinforcement-learning workloads.

Run it locally

# DOM-first automation server
moli serve
# enable real geometry, coordinate input and screenshots
moli serve --layout
# also fetch optional images, fonts, audio/video
moli serve --layout --resource

Playwright connects directly over CDP:

import { chromium } from "playwright";
const browser = await chromium.connectOverCDP("http://127.0.0.1:9222");

Use it in the cloud (Lexmount Cloud Browser)

In the Lexmount SDK, Moli is the browser mode with API value light:

from lexmount import Lexmount
with Lexmount() as client:
    with client.sessions.create(browser_mode="light") as session:
        print(session.id)

Note: hosted Moli does not support contexts, extensions, persistent replay or stored downloads. Use Chromium (normal) mode when you need those, and validate your target pages first.

Details: GitHub lexmount/moli and browser.lexmount.com.