The Problem
Web pages carry significant structural noise—navigation, ads, scripts, and deeply nested DOM—that dilutes content value when fed to LLMs or documentation pipelines. Converting a page to clean Markdown typically requires manual copy-paste into a converter or writing a scraping script, neither of which works well for one-off or interactive use.
What This Does
MD-This-Page is a browser extension that converts the current page to clean Markdown in one click. The core logic lives in background.ts (context menu and keyboard shortcut handling) and content.ts (page extraction), with the preview and export UI in tabs/markdown.tsx. It uses Mozilla's Readability library for content extraction and offers toggleable options: images, links, metadata, source URL, and a document structure map.
The extension targets LLM workflows specifically—the README argues Markdown is more token-efficient and structurally clearer than raw HTML. Output can be copied, downloaded as a .md file, or copied as a pre-formatted prompt. Built with React and Tailwind, styled via tabs/style.css and tailwind.config.js.
How To Use It
Setup — Install from the Chrome Web Store or Firefox Add-ons, or build from source. The repo uses pnpm (see pnpm-lock.yaml). Build commands are not documented in the README excerpt, so you'd run pnpm install then pnpm build (the exact script name isn't visible in the provided files).
Configuration — No environment variables or API keys required. The extension works entirely client-side. Options are set in the preview tab UI, not via config files.
Running it — After installation, right-click any page and select ".MD this page", or press Alt+M. The extension opens a dedicated tab showing the extracted Markdown with export controls.
Build from source (commands inferred from package.json and pnpm-lock.yaml) pnpm install pnpm build
Real-World Use
A researcher building a knowledge base for an LLM-powered Q&A system can use this to capture source material. Instead of writing per-site scrapers, the extension normalizes any article, blog post, or documentation page into consistent Markdown that feeds directly into a vector store or context window.
// Conceptual workflow: captured Markdown feeds an LLM pipeline const markdown = await captureCurrentPage(); // via extension const chunks = splitIntoChunks(markdown); await vectorStore.upsert(chunks);
Code Health & Issues
Med — No tests detected — No test files exist in the repo. The extraction and rendering logic in content.ts and tabs/markdown.tsx is untested, which is risky for a tool that must handle arbitrary, malformed HTML. Med — No LICENSE file — The repo is a fork of an MIT-licensed project (Ademking/MD-This-Page), but the LICENSE is absent here. Usage and redistribution rights are unclear. Low — No CI pipeline — .github/workflows/submit.yml exists but appears to handle extension submission, not automated testing or linting. No build verification on commit. Low — Minimal error handling visible — The README doesn't mention failure modes (e.g., pages with no readable content, paywalls, or dynamic-heavy SPAs). No evidence of robust fallback in the extracted code.
The codebase is small and focused; the main risks are the absence of tests and the missing license, not architectural problems.
The Bottom Line
A practical, well-scoped extension that solves a real pain point for LLM workflows. It's not novel—the original project has 1,300+ stars—but the implementation is clean and the feature set is appropriate. Use it if you regularly convert web content for LLM ingestion; skip it if you need guaranteed correctness on arbitrary pages, since the untested extraction logic will occasionally fail on complex sites.