We built a public MCP server that lets AI assistants like Claude and ChatGPT search feeds and browse the directory without signing in. Once signed in, you can also hand off managing subscriptions, folders, OPML, rules, saved searches, watched keywords, tags, and your read-later list. Neither one returns article text or article lists.
Each public page can now be fetched as Markdown — just headings and body text — by adding .md to the end of the URL or sending Accept: text/markdown. A summary of the whole site is available at /llms.txt (short version) and /llms-full.txt (full version).
Added a “/facts” page that lists things like pricing, limits, supported regions, and supported languages as plain facts, without marketing copy. The limits table on the help page now references the same numbers.
We allow crawlers from major AI services that fetch pages based on search or a user's question. For text and data that freefeed itself created, we also allow crawling for training purposes (this doesn't include other sites' content, such as feed names or article titles). Crawlers fetching for training purposes from the EU, EEA, UK, or Switzerland are still not accepted.
Added 8 guides covering things like RSS basics and how to move over from other services, and added FAQs to all 13 guides, including the 5 that already existed.
Enter the URL of a page that lists articles, and freefeed picks up the headlines and turns them into a feed you can subscribe to (updated hourly, up to 20 per person; it doesn't fetch the full text).