RSS, Atom & Google News MCP

Bring current feed content into MCP-compatible agents without a browser or search API credential. This Apify standby Actor fetches RSS, RDF, and Atom XML, normalizes common feed fields, searches Google News RSS, and creates concise extractive feed summaries without using an LLM.

What you can do

• Parse a publisher, blog, podcast, changelog, or monitoring feed into consistent JSON.
• Search current Google News headlines with a free-form query.
• Retrieve topic headlines for world, business, technology, science, health, sports, or entertainment.
• Produce short feed-item previews for briefings and triage while retaining source links and dates.

The Actor reads feed metadata and item descriptions only. It does not crawl linked articles, bypass subscriptions, translate content, deduplicate across feeds, classify sentiment, or generate abstractive summaries.

Connect from an MCP client

Use this deployed Streamable HTTP endpoint:

https://reverberant-equality--mcp-rss-news.apify.actor/mcp

Authenticate at the Apify edge on every connection:

Authorization: Bearer $APIFY_TOKEN

Clients should accept application/json and text/event-stream. MCP operations use POST /mcp; direct GET /mcp and DELETE /mcp requests return HTTP 405.

Registered tools

fetch-rss-feed

• url — required valid URL for an RSS, RDF, or Atom feed.
• maxItems — optional number, default 20; values are truncated to an integer and clamped to 1–50.

Returns { title, description, link, items }. Each item contains title, link, pubDate, author, summary, and categories.

google-news-search

• query — required non-empty string.
• maxItems — optional number, default 20, clamped to 1–50.

Builds a US-English Google News RSS search and returns { query, title, description, link, items } using the normalized item fields above.

top-news

• topic — optional enum: world, business, technology, science, health, sports, or entertainment; default technology.
• maxItems — optional number, default 20, clamped to 1–50.

Returns { topic, title, description, link, items } from the corresponding US-English Google News topic feed.

summarize-feed-items

• url — required valid feed URL.
• maxItems — optional number, default 10, clamped to 1–50.

Returns { title, link, itemCount, items }; each item has title, link, pubDate, and summary. HTML is stripped and each summary is limited to 300 characters. If an item has no description, its title is used. No LLM or external summarization service is called.

All successful data is delivered as one MCP text content item whose text is formatted JSON.

JSON example

Arguments for top-news:

{
  "topic": "science",
  "maxItems": 2
}

Decoded text-result shape (headlines are illustrative):

{
  "topic": "science",
  "title": "Science - Latest",
  "description": "",
  "link": "https://news.google.com/",
  "items": [
    {
      "title": "Example science headline",
      "link": "https://news.google.com/rss/articles/example",
      "pubDate": "Tue, 18 Aug 2026 12:00:00 GMT",
      "author": "Example Publisher",
      "summary": "A short description supplied by the feed.",
      "categories": []
    }
  ]
}

Pricing

Every tool invocation charges one tool-call event at $0.005 USD. Charging happens before the feed request, so a later upstream timeout, malformed feed, or unsupported format can still consume the event. Invalid arguments rejected by the MCP schema do not enter the handler. Apify platform usage can be billed separately according to your plan.

Sources, limits, and error behavior

fetch-rss-feed and summarize-feed-items contact the URL you supply. Google News tools use public Google News RSS search and topic endpoints with fixed hl=en-US, gl=US, and ceid=US:en parameters. Google does not publish a stable RSS quota or availability SLA; feeds can throttle requests, change format, omit fields, reorder articles, or return tracking/aggregator links.

Every feed request has a 15-second timeout, follows at most five redirects, and accepts at most 5 MiB of response content. Only the first 50 items can be returned. RSS 2.x-style channels, RDF RSS, and Atom are recognized; other XML or malformed documents return an error. Namespace-heavy or unusual feeds may leave normalized fields empty. Dates are preserved as source strings rather than converted to a single timezone.

Tool failures return an MCP result with isError: true and text beginning Error in <tool-name>:. Typical causes include an invalid or unreachable URL, a non-success HTTP response, timeout, oversized content, malformed XML, or unsupported feed format. Invalid URL types, empty search queries, and unsupported topics are MCP schema errors. Unexpected HTTP transport failures return a JSON-RPC internal-server error.

Privacy and safe use

Supplied URLs and Google News queries are sent from Apify infrastructure to the relevant feed host. The source sees the Actor's network address and rss-news-mcp/1.0 user agent. Results are returned directly and are not written to an Apify dataset or key-value store by this implementation. Apify, Google, and feed publishers may retain operational metadata under their policies.

Only request feeds you are authorized to access. Do not place credentials or secrets in feed URLs, and do not use private/internal URLs; the server does not add authentication or promise isolation for user-supplied destinations.

Run locally

Node.js 20 or newer is required.

npm install
npm run start:dev          # http://localhost:3000/mcp
npm test
npm run build
npm run start:prod
apify validate-schema

Set APIFY_CONTAINER_PORT to use a port other than 3000.