Picture published by the creator on GitHub.
Moli: Extract web pages for AI agents with Moli
Added · Updated
Made by LexMount
View profile on GitHub (opens in a new tab)Let an AI agent fetch web pages as Markdown, search the web and automate browser tasks without installing Chrome.
Moli is a headless browser (a browser with no window, driven by programs) for developers building agents, crawlers and retrieval pipelines. Written in Rust, it runs JavaScript and reads page structure directly, and only works out layout or draws pixels when a task such as a screenshot needs them. You drive it from its CLI or through CDP and WebDriver, the standard protocols tools like Playwright use. Lexmount reports a median of 73 MiB of memory per page against 773 MiB for headless Chrome in its 192-URL crawl test.
What it does
Moli reads page structure and styles directly, running layout and rendering when an operation needs them.
How to use it
Install it with the one-line installer script on Linux or macOS or the PowerShell command on Windows, or ask your agent to install the skills from the repository. Then run moli fetch --dump markdown with a URL to get the page as Markdown. Add --layout with --dump screenshot or --dump pdf for visual output, or run moli serve and connect Playwright over CDP to automate pages from code.
Common use cases
- Turn a web page into text for a model: run moli fetch with --dump markdown or --dump semantic_tree_text on a URL and get Markdown or a compact page outline back.
- Click before you capture: pass a JavaScript expression with --eval, such as clicking a show-details button, and export the resulting page as Markdown.
- Save a page as an image or document: run moli fetch --layout with --dump screenshot_full or --dump pdf and get a full-page PNG or a PDF.
Can I use this?
Requirements Linux, macOS or Windows. Installation uses a shell script or PowerShell command from the project’s releases, and no separate Chrome, ChromeDriver or geckodriver install is needed. Playwright or another CDP or WebDriver client is optional for scripted automation. The source does not list hardware requirements.
Caveat Moli does not aim to match Chrome pixel for pixel or offer high-fidelity Canvas, WebGL or media playback, and not every Chrome screenshot or print mode exists. All benchmark figures are Lexmount’s own; in its crawl test about 54% of URLs produced useful pages.
Worth knowing When exporting after an interaction, only the supplied expression performs interactions. An exception stops the command without an export. Lexmount also runs Lexmount Browser, a managed cloud service built around Moli; the open-source browser works without it. Licensed under Apache License 2.0 OR MIT, at your option, unless a file states otherwise.
Why it's here
Not trending this week
Last seen on the GitHub trending list on 3 November 2026. It stays on BuildTube so you can still find it.
- 23,015 people have starred it on GitHub.
Numbers checked on 8 November 2026; not refreshed since.
Behind it
View setup guideBuildTube has not run or verified this project. Everything above is written from what the creator published.
Made this?
Quick questions
- What is Moli?
- Moli reads page structure and styles directly, running layout and rendering when an operation needs them.
- What can I use Moli for?
- Turn a web page into text for a model: run moli fetch with --dump markdown or --dump semantic_tree_text on a URL and get Markdown or a compact page outline back; Click before you capture: pass a JavaScript expression with --eval, such as clicking a show-details button, and export the resulting page as Markdown; Save a page as an image or document: run moli fetch --layout with --dump screenshot_full or --dump pdf and get a full-page PNG or a PDF
- Can I run Moli on my computer?
- The source does not list hardware requirements. The source lists direct installation for Linux and macOS, and PowerShell installation for Windows.
- What do I need to use Moli?
- Linux, macOS or Windows; use the CLI or a supported browser automation protocol.
- What license does Moli use?
- Licensed under Apache License 2.0 OR MIT, at your option, unless a file states otherwise.
More from this week
- agent-desktop Let AI agents read and click desktop apps
For developers building computer-use agents: agent-desktop lets an AI agent read and operate desktop apps on Mac, Windows or Linux from a command line.
- OpenBot Run AI coworkers with their own browser and files
Give an AI coworker a job, connect the tools it needs and watch it work.
- DeepTutor Study your own documents with a personal AI tutor
Read documents with page-by-page citations, practice daily and get video lessons from an AI tutor.
Explore by topic
Post this
Download share imageFor makers
Made this? Show it’s on BuildTube — add the badge to your README or embed the card on your site.
README badge (Markdown)
[](https://buildtube.app/p/gh-lexmount-moli/)Embed card (HTML)
<iframe src="https://buildtube.app/embed/gh-lexmount-moli/" width="560" height="240" style="border:0;max-width:100%" loading="lazy" title="Moli on BuildTube"></iframe>