Absent News publishes a twice-daily tech digest: ranked links, analysis, clustering data. This page is about pulling that into your tools programmatically.
We run a public MCP server. Any MCP-compatible client can connect and query episodes, search stories, or browse the archive. No API key required.
Connect
If you use Claude Code:
claude mcp add --transport http absent-news https://mcp.absent.dev/mcp
Then ask naturally. “What’s in today’s absent news?”, “search absent news for Oracle”, that sort of thing.
For other MCP clients, point a Streamable HTTP transport at:
https://mcp.absent.dev/mcp
Tools
get_latest_episodeToday’s headlines, or full content including the blog post, cluster data, and social posts.
Pass format="full" for everything.
get_episodeFetch a specific episode by number.
e.g. ABS-0009
search_storiesKeyword search across all episode titles and story headlines.
list_episodesRecent episodes with date, edition, title, and link count.
How it works
The server is a Cloudflare Worker that reads from the same database as this site. Stateless, fast, no session management. Each request creates a fresh MCP context, queries Neon Postgres, and returns structured content.
The data is the same data that powers every page here. Ranked stories, cluster analysis, blog posts, social posts. Nothing is paywalled or filtered differently through the MCP.
