newsmcp

Real-time world news briefings with AI-clustered events, topic classification, and geographic filtering. No API key needed.

World News Skill

Get real-time world news from an AI-powered news aggregation service. Stories are clustered into events (deduplicated across multiple sources), summarized, classified by topic, and tagged by geography.

No API key required. No authentication. Just call the API.

Why This Is Different

• Events, not articles — individual articles are clustered into stories, so you get one summary per story instead of 20 duplicates
• Pre-summarized — every event has an AI-generated summary, no need to read full articles
• Classified — 12 topic categories and 100+ geographic regions
• Importance signals — sources_count (how many outlets covered it) and impact_score (AI-assessed significance)
• Multi-source — each event lists all source articles with titles, URLs, and domains

API Base URL

https://newsmcp.io/v1

Endpoints

1. Get News Events

curl -s "https://newsmcp.io/v1/news/?hours=24&per_page=10&order_by=-sources_count"

Parameters:

Parameter	Type	Default	Description
hours	int	24	Time window (1-168, i.e. up to 7 days)
topics	string	—	Comma-separated topic slugs (e.g. politics,military)
geo	string	—	Comma-separated geo slugs (e.g. europe,united-states)
page	int	1	Page number
per_page	int	20	Results per page (max 50)
order_by	string	-sources_count	Sort field. Options: sources_count, entries_count, impact_score, first_seen_at, last_seen_at (prefix - for descending)

Response shape:

{
  "events": [
    {
      "id": "uuid",
      "summary": "AI-generated event summary",
      "topics": ["politics", "military"],
      "geo": ["europe", "ukraine"],
      "entries_count": 15,
      "sources_count": 8,
      "first_seen_at": "2026-03-03T10:00:00Z",
      "last_seen_at": "2026-03-03T14:00:00Z",
      "impact_score": 4,
      "entries": [
        {
          "title": "Article headline",
          "url": "https://example.com/article",
          "domain": "example.com",
          "published_at": "2026-03-03T12:00:00Z"
        }
      ]
    }
  ],
  "total": 42,
  "page": 1,
  "per_page": 10
}

2. Get Event Detail

curl -s "https://newsmcp.io/v1/news/{event_id}/"

Returns the same fields as the list, plus context — extended background and analysis of the event.

3. List Available Topics

curl -s "https://newsmcp.io/v1/news/topics/"

Returns: politics, economy, technology, science, health, environment, sports, culture, crime, military, education, society

4. List Available Regions

curl -s "https://newsmcp.io/v1/news/regions/"

Returns continents (europe, asia, africa, north-america, south-america, oceania) and countries (e.g. united-states, ukraine, china, lithuania, etc.). Each region has a type field: "continent" or "country".

How to Use This Skill

When the user asks for news, headlines, or current events:

1. Default briefing — Fetch top events from the last 24 hours sorted by importance:

   curl -s "https://newsmcp.io/v1/news/?hours=24&per_page=10&order_by=-sources_count"
   

2. Topic-specific — If the user asks about a specific topic (e.g. "tech news", "sports"), use the topics parameter:

   curl -s "https://newsmcp.io/v1/news/?topics=technology&hours=24"
   

3. Region-specific — If the user asks about a region (e.g. "European news", "what's happening in Asia"), use the geo parameter:

   curl -s "https://newsmcp.io/v1/news/?geo=europe&hours=24"
   

4. Combined filters — Topics and geo can be combined:

   curl -s "https://newsmcp.io/v1/news/?topics=politics,military&geo=europe&hours=48"
   

5. Deep dive — For more detail on a specific event, fetch the event detail:

   curl -s "https://newsmcp.io/v1/news/{event_id}/"
   

Formatting the Briefing

Present results as a multi-story news briefing covering the top events — not just one.

Rules:

• List each event separately — 1-2 lines per event with its summary, source count, and 1-2 links
• Lead with the most important events (highest sources_count or impact_score)
• Related events can be grouped under a shared heading, but don't collapse everything into a single narrative
• Mention the time window and total number of events found
• If topics or geo arrays are present, use them to add context tags

Output template:

Here are today's top stories ({N} events in the last {hours}h):

1. **{Event summary}** — {sources_count} sources · [Source Title](url)
2. **{Event summary}** — {sources_count} sources · [Source Title](url)
3. **{Event summary}** — {sources_count} sources · [Source Title](url)
...

What NOT to do:

• Do NOT present only 1 story when the API returned multiple events (unless the user asked for a single topic)
• Do NOT write a long essay about one event and ignore the rest
• Do NOT merge all events into a single combined narrative

Topic Slugs Reference

politics economy technology science health environment sports culture crime military education society

Example Interaction

User: "What's happening in the world today?"

Action: Fetch https://newsmcp.io/v1/news/?hours=24&per_page=10&order_by=-sources_count

Expected output format:

Here are today's top stories (42 events in the last 24h):

1. **US Senate passes infrastructure bill after marathon session** — 34 sources · [Reuters](url) · [AP News](url)
2. **Earthquake strikes eastern Turkey, at least 12 dead** — 28 sources · [BBC](url)
3. **OpenAI announces new reasoning model** — 22 sources · [The Verge](url)
4. **EU imposes new sanctions on Russian energy sector** — 19 sources · [FT](url)
5. **Wildfire in California forces thousands to evacuate** — 15 sources · [LA Times](url)

Want more detail on any of these stories?

User: "Give me the latest on Ukraine"

Action: Fetch https://newsmcp.io/v1/news/?geo=ukraine&hours=48&order_by=-last_seen_at

User: "Any tech news this week?"

Action: Fetch https://newsmcp.io/v1/news/?topics=technology&hours=168&order_by=-sources_count