Fetch the latest articles from all configured RSS feeds and store them (duplicates are skipped). Use when the user wants to refresh the archive or asks about the newest articles.

Search stored articles by keyword (matches title or summary). Use when the user asks about a specific topic from the archive.

Return raw article data (title, URL, source, summary) from the last days days as a markdown list. Use when you (the host model) should write the digest yourself - this tool only provides the source material.

topic is an optional keyword pre-filter (whole-word match) - useful for large archives. Omit it to get everything and pick relevant articles yourself, which handles synonyms and related themes better.

Build a ready-made markdown digest of the most relevant articles from the last days days using the server's own LLM pipeline. Not implemented yet - use get_articles_for_digest instead.

List the feeds the digest currently aggregates, with the number of stored articles per source. Reports whether the app runs on bundled default feeds (feeds.txt is empty) or the user's own selection.

Validate and add a new RSS/Atom feed to feeds.txt. Use when the user wants to follow a new source. The feed is downloaded and parsed first - invalid or dead URLs are rejected, nothing is written.

While feeds.txt is empty the app runs on bundled default feeds. When adding the user's FIRST own feed, ask them once whether to keep the defaults too (then call this with keep_defaults=True) or start fresh with only their own sources. Do not ask again once feeds.txt has entries.

Remove a feed from feeds.txt. Use when the user no longer wants a source. Already stored articles from that source stay in the archive.