The value is that the page lands on disk as markdown with a dated filename, so a research session leaves behind something you can grep next month instead of a chat log. Worth knowing before you point it at anything sensitive: fetching goes through r.jina.ai rather than your own browser, so the target URL passes through a third-party reader.
A small Node MCP server that fetches a web page through the r.jina.ai reader service and writes the result to a markdown file in a download directory you choose. The directory setting persists between sessions.
- `download_markdown` — fetch a URL and save it as markdown, named from the URL plus the date
- `set_download_directory` — point it at a folder, checked for existence and write permission before it is saved
- `get_download_directory` — read back where files are currently going
- `list_downloaded_files` — see everything already saved
Node.js version 16 or higher and npm. Install from Smithery in one command, or clone and build. Configuration lives in a per-user config file, with the default download directory under your home folder on macOS and Linux and under Documents on Windows. Pages are fetched through r.jina.ai, so the URL you ask for is sent to that service.
One command — npx -y @smithery/cli install @dazeb/markdown-downloader --client claude