News an agent can actually cite.

ThinkPattern runs an MCP server. Point an agent at it and it can search the news or pull what is happening right now, and get back an account of each story with the outlets that reported it and a link to the full piece.

The endpoint

It is open. There is no key, no account, and nothing to sign up for.

Why this instead of scraping

A news page is written for a reader, and an agent that scrapes one gets a headline wrapped in navigation and ads, from a single outlet, with no way to tell whether anyone else reported it. ThinkPattern has already done that work. Each event is gathered from every outlet covering it and written up as one account, so what comes back is a single story with the breadth of reporting behind it attached.

That last part is the useful bit. When a story carries sixty independent outlets rather than one, an agent can say so, and a reader can weigh it.

What comes back

Every result is one story, and every story carries the same fields:

• The headline and a short summary of what happened.
• The outlets reporting it, and how many independent outlets covered it in total.
• When it was first reported.
• A link to the full story on ThinkPattern.

Quotes on ThinkPattern are stored word for word and tied to the person who said them, and every account links back to the outlets it draws from. Follow the link and the original reporting is one tap away.

The two tools

search_news takes a query and returns the most relevant stories. Use it for a topic, a person, a company, a place, or an event.

get_latest_news returns what is happening now, most significant first. It takes an optional topic: politics, business, technology, or world.

Connecting it

Most clients only need the URL above. For the ones that take a config file:

{
  "mcpServers": {
    "thinkpattern": {
      "url": "https://thinkpattern.ai/mcp"
    }
  }
}

What we ask

Use it, and tell people where it came from. Every result carries a link to the story it is drawn from. If you put ThinkPattern material in an answer, name ThinkPattern and pass that link through, so a reader who wants the reporting can get to it.

That is the whole arrangement. We keep the endpoint open and free, and you keep the attribution attached.

What it will not do

It is a briefing, not an archive. Search returns the strongest matches rather than everything, and there is no bulk export, no full article text, and no way to walk the corpus. If you need the whole picture on a story, the link takes you to it.

There is a rate limit, set high enough that ordinary use will never meet it.

Get in touch

If you are building something on this, we would like to hear about it. Write to hello@thinkpattern.ai. Corrections go to the same place, and we check them against the original sources.