I built this because my MCP clients kept needing web content as clean text for RAG and research — most URL tools return raw HTML or dump the whole page.

How it works: - Free API: http://167.233.135.161:9999/api/convert?url=... (10/day, no signup). Strips nav, ads, boilerplate. - MCP server (stdio + streamable HTTP): one tool, web2md_convert. Works with Claude Desktop, Cursor, Continue.dev, Windsurf. Published in the official MCP registry (io.github.astra-intelligence/web2md). - CLI + Dockerfile in the repo.

Honest scope: single-page conversion today (no site crawls, no PDFs yet). Feedback welcome — especially on output quality vs. readability, and whether people want batch/crawl next.

The repo: https://github.com/astra-intelligence/web2md-mcp