RSS Aggregator — User Guide | YoBench

What the RSS module does

The module turns scattered RSS/Atom feeds into manageable collections, each with its own AI provider and summarization prompt. Instead of scanning hundreds of headlines, you get auto-summarized articles in one place — and useful pieces can be pushed to RAG contexts for later retrieval inside AI Chat.

What you get:

• Topic-based collections — each with its own sources, AI provider and summarization prompt.
• On-the-fly AI summarization — for every fetched article the module calls the chosen LLM with your prompt and saves the summary.
• Automatic refresh — each collection is polled on its own interval (poll_minutes), no manual action required.
• Save to RAG — push an article (summary + full text) into a context with one click, ready for vector search.
• Full article text — stored and available independently of whether an AI provider is configured.

Collection parameters

When creating or editing a collection:

• Name — an arbitrary label (e.g. "Tech news" or "Regulatory updates").
• RSS/Atom sources — list of URLs, one per line. The same URL can be reused across collections with different prompts and providers.
• Polling interval (minutes) — poll_minutes. Default 60. Minimum 5 — to avoid hammering sources.
• AI provider (optional) — which provider from the central AI Chat registry to use. Without it, articles are saved without summaries.
• Processing prompt — the LLM instruction. Default: "Summarize this article in 2-3 sentences." Adapt to your topic (e.g. "Extract the key legal changes and highlight effective dates").
• Enabled — automatic refresh flag. Untick to pause polling without deleting the collection.

There are no module-wide settings — everything lives at the collection level.

What's stored in the database

• rss_collections — collection config (sources is a JSON array of URLs, provider_id, prompt, poll_minutes).
• rss_items — articles (title, article_url, summary, full_text, published_at, fetched_at).

There are no "read/unread" or "favorite" flags — the module is built around fast consumption via AI summaries.

Article actions

• Open — view the article: title, source, date, summary, and (if available) full text.
• Full text — toggle between summary and original.
• Context — push the article into a RAG context: a modal opens with a context picker, the article (summary or full text) is saved, indexed, and becomes available for AI queries.
• Clear — delete every article in the collection (with confirmation). Sources stay.

Workflow

1. Create a collection

1. Open the RSS module from the left sidebar.
2. On the Collections tab click + New collection.
3. Enter a name and a list of URLs (one per line).
4. Optionally pick an AI provider and adapt the prompt.
5. Set the polling interval (default 60 minutes).
6. Save.

2. Trigger fetching

Each collection has a Run button — polls every source immediately, without waiting for the next tick.

3. Read articles

The Content tab shows every article in the selected collection, sorted by publication date.

• Click an article to open the panel with the summary.
• The Full text button switches to the original text.
• Without an AI provider, only the full text is available.

4. Save useful items to RAG

Click Context on an article → pick a target context from the Context Manager & RAG module. The article (summary + full text) is saved to that context and becomes available for AI queries in AI Chat.

5. Manage collections

• Edit — change sources, prompt, provider or interval. Changes apply on the fly (scheduler is restarted).
• Untick "Enabled" — pause the collection.
• Clear — delete its articles (sources stay).
• Delete collection — remove together with its articles.

Next steps

• Connect AI providers — without them summarization does nothing.
• Configure the Context Manager so RSS articles flow into your knowledge base.
• For deep analytics of code projects, use Git Analytics.

Help and feedback

Found a bug or have a feature idea? Contact us via the feedback form.