explaingit

giantravens/mdbrowse

Analysis updated 2026-05-18

0PythonAudience · developerComplexity · 3/5Setup · moderate

TLDR

A command line tool that renders web pages through a real browser and converts them into clean, readable markdown for a terminal browser or AI agents.

Mindmap

mindmap
  root((mdbrowse))
    What it does
      Converts web pages to markdown
      Runs a terminal browser
      Serves pages to AI agents
    Tech stack
      Python
      Chromium
      Playwright
    Use cases
      Read articles without ads or clutter
      Archive pages as versioned snapshots
      Give an AI agent token-efficient web access
    Audience
      Developers
      AI agent builders

Code map

Detail Auto

An interactive map of this repo's files and how they connect — its source is parsed live in your browser. Click Visualize to build it.

filefunction / class

What do people build with it?

USE CASE 1

Read news articles and feeds in a distraction-free terminal browser.

USE CASE 2

Give an AI coding agent a token-efficient, structured view of web pages via the included MCP server.

USE CASE 3

Capture web pages as dated, citable markdown snapshots for research or open-source intel work.

USE CASE 4

Watch specific pages for real changes over time instead of manually rechecking them.

What is it built with?

PythonChromiumPlaywrightMCP

How does it compare?

giantravens/mdbrowse0xallam/my-recipe0xhassaan/nn-from-scratch
Stars00
LanguagePythonPythonPython
Last pushed2022-11-22
MaintenanceDormant
Setup difficultymoderatemoderatemoderate
Complexity3/52/54/5
Audiencedevelopergeneraldeveloper

Figures from each repo's GitHub metadata at analysis time.

How do you get it running?

Difficulty · moderate Time to first run · 30min

Needs a Chromium or Chrome browser engine plus a Python 3.11+ environment, first run auto-installs both via a bundled script.

In plain English

mdbrowse is a command line tool that turns any web page into clean markdown text, and gives you a terminal browser built on top of that conversion so you can read the web without leaving your terminal window. Its command is called mdb. Behind the scenes, mdb opens a real Chromium or Chrome browser in the background, so pages load with full JavaScript, cookies, and the same rendering a normal browser would give you. From that fully rendered page, mdb extracts the structure and content and produces clean, well organized markdown, stripping out ads, menus, and sidebar clutter while keeping links and page structure intact. The tool is aimed at two kinds of users at once. Humans get a calm, no-fluff way to read articles and feeds from a terminal, with keyboard controls similar to vim, tab navigation between links, and support for simple web forms like search boxes on Wikipedia or DuckDuckGo. It can also read your existing Safari cookies by default so logged-in or paywalled pages render as you normally see them, or you can pass a private flag to skip that. AI agents get a token-efficient view of the same pages, since raw HTML is expensive to feed into an LLM. The project states that in its own benchmark suite, mdb is the only extractor that keeps full fact recall, working links, and page structure at once, and it ships an MCP server called mdb-mcp so agents can fetch pages, search the web, and page through results. Captures made by mdb are meant to be deterministic and diffable: the same page state produces identical output, along with metadata recording where the page came from, when it was captured, and how confident the extraction was. This makes it useful for archiving pages as evidence, watching pages for meaningful changes over time rather than just noticing that something changed, and building a searchable local archive of everything you have captured. Installation on a Mac is done through Homebrew with a single command, or you can clone the repository and run its included launcher script, which sets up a local Python environment and installs the Chromium browser engine automatically on first run. The project requires Python 3.11 or newer if you set it up by hand instead. The README includes a detailed walkthrough for readers comfortable typing terminal commands but not necessarily debugging Python or virtual environments.

Copy-paste prompts

Prompt 1
Help me install mdbrowse with Homebrew and run my first page capture with mdb.
Prompt 2
Explain how to connect the mdb-mcp server to my AI agent so it can browse the web through mdbrowse.
Prompt 3
Show me how to use mdb watch to track a webpage for changes over time.
Prompt 4
Walk me through setting up mdbrowse from source using uv and Playwright instead of Homebrew.

Frequently asked questions

What is mdbrowse?

A command line tool that renders web pages through a real browser and converts them into clean, readable markdown for a terminal browser or AI agents.

What language is mdbrowse written in?

Mainly Python. The stack also includes Python, Chromium, Playwright.

How hard is mdbrowse to set up?

Setup difficulty is rated moderate, with roughly 30min to a first successful run.

Who is mdbrowse for?

Mainly developer.

Open on GitHub → Explain another repo

This repo across BitVibe Labs

Verify against the repo before relying on details.