rss-mcp

RSS MCP Server

An MCP server that lets AI assistants subscribe to, fetch, search, and manage RSS feeds. Built with TypeScript and SQLite with full-text search.

Inspiration

Inspired by Andrej Karpathy's post on reclaiming your information diet:

"Finding myself going back to RSS/Atom feeds a lot more recently. There's a lot more higher quality longform and a lot less slop intended to provoke... We should bring back RSS - it's open, pervasive, hackable."

Quick start with curated feeds: Import the Most Popular Blogs of Hacker News 2025 OPML file to get 92 high-quality tech blogs:

import the OPML file from https://gist.github.com/emschwartz/e6d2bf860ccc367fe37ff953ba6de66b

Claude will automatically use the import_opml tool to add all feeds.

Features

• Subscribe to RSS/Atom feeds and fetch new posts
• Full-text search across titles, summaries, and content (FTS5)
• Extract clean article content from web pages (Readability)
• Import feeds in bulk from OPML files
• Track read/unread state
• Daily digest for compact summaries
• HackerNews popularity ranking for posts
• Conditional HTTP requests (ETag/Last-Modified) and per-feed rate limiting

Installation

As a Claude Code Plugin (Recommended)

/plugin marketplace add 0xQuinto/rss-mcp
/plugin install rss-mcp@0xquinto-rss-mcp

Via bunx

bunx @0xquinto/rss-mcp

Via npx

npx @0xquinto/rss-mcp

Tools

Tool	Description
list_feeds	List all subscribed feeds
add_feed	Subscribe to a feed by URL
remove_feed	Unsubscribe and delete all posts for a feed
import_opml	Bulk import feeds from an OPML file
refresh_feeds	Fetch latest posts (all feeds or a specific one)
get_posts	Query posts with filtering, pagination, and full-text search
get_post_content	Retrieve article content with pagination (max_length, offset)
get_daily_digest	Get compact digest of recent posts for synthesis
get_popular_posts	Rank recent posts by HackerNews engagement
mark_read	Mark posts as read
mark_unread	Mark posts as unread

Data

Posts are stored in ~/.rss-mcp/rss.db (SQLite). The database and FTS5 index are created automatically on first run.

License

MIT