Build an AI Automation with an RSS Trigger in n8n

Description and use

A persistent RSS-triggered automation summarizes new feed items with Claude, validates the JSON output in a Code node, filters by relevance ≥ 6, and posts to Slack. A separate Error Workflow alerts on any failure (feed down, API rate limit, bad JSON, Slack down). The first activation may emit a batch of existing items, so test with a small or low-traffic feed first.

Problem

Build a persistent, event-driven AI automation that triggers on new RSS feed items, summarizes them with an LLM, filters noise by relevance score, and posts to Slack — with robust error handling.

Outcome

A persistent RSS-triggered automation summarizes new feed items with Claude, validates the JSON output in a Code node, filters by relevance ≥ 6, and posts to Slack. A separate Error Workflow alerts on any failure (feed down, API rate limit, bad JSON, Slack down). The first activation may emit a batch of existing items, so test with a small or low-traffic feed first.

Workflow

• Add an RSS Feed Trigger node with your feed URL and poll interval of every 30 minutes (RSS has no webhooks, so polling is the trigger).
• Add a Basic LLM Chain node with an Anthropic Chat Model sub-node, attach credentials, pick a model from the dropdown, and set the prompt to summarize title, content snippet/content (capped at 6000 chars), and return JSON with summary, relevance (1–10), and tags.
• Set the system message to require ONLY a JSON object with no markdown or preamble.
• In the LLM node Settings, turn on Retry On Fail (3 tries, a few seconds between) to cover rate limits and temporary API errors.
• Add a Code node (JavaScript, Run Once for Each Item) to validate the raw LLM output: strip HTML, parse JSON, check shape (summary is string, relevance is number, tags is array); throw on bad output so failures are visible.
• Add an IF node gating on {{ $json.relevance }} >= 6 to filter out noise before it reaches Slack.
• Add a Slack Send Message node on the true branch posting title, link, relevance score, summary, and tags.
• Create a separate 'Alert on failure' workflow starting with an Error Trigger node, followed by a Slack or email node sending workflow name and error message.
• In the main workflow Settings, set Error Workflow to the Alert on failure workflow.

Evidence

Most n8n + AI tutorials end with a workflow that works once. This one is built to keep running: it triggers on an RSS feed, summarizes new items with an LLM, filters out noise, and posts to Slack.

Source

RSS Feed Trigger → Basic LLM Chain (Claude) → Code (validate JSON) → IF (relevance) → Slack

Source

Feed URL: your feed
Poll Times: every 30 minutes (RSS has no webhooks, so polling is the trigger)

Source

n8n remembers which items it has already emitted, so you get deduplication for free. Note that the very first activation can emit a batch of existing items, so test with a small or low-traffic feed first.

Source

The .slice(0, 6000) caps token cost per item. Some feeds ship full article HTML, which gets expensive fast.

Source

In the node’s Settings