plainradar

What's happening in
today?

  1. 01 Enter any topic and press Enter.
  2. 02 We scan sources across the web live for matching coverage.
  3. 03 Browse your curated stories and enjoy.

Welcome back.

Account

Your account.

Each manual search or scheduled topic tracker uses 1 credit. Agents use the same credits.

Trackers

Each tracker runs on its schedule. A topic tracker update uses 1 credit. Site trackers are free; with noise filters, each update uses 1 credit.

Agent tokens

Each token lets one MCP agent search as you. Revoke a token to cut that agent off.

Password

Two-factor login

Ask for a code from an authenticator app (Google Authenticator, 1Password, Authy, …) every time you log in.

Sessions

Logged in somewhere you don't recognise? This logs out every device, this one too.

Delete account

Deletes your account, tokens and sessions for good.

New tracker.

Either way, everything it finds also shows up in My trackers.

Add sites.

To “” ·

Type a topic above and press Explore. We search our index of news sites and show the best matches. Uses 1 credit.

    Track this topic.

    For “”. You only hear from us when something new turns up.

    Either way, everything it finds also shows up in My trackers.

    Webhook.

    For “”. We POST the new stories as JSON to this URL after each run that finds something.

    Checking the signature

    Every request carries Plainradar-Signature: t=<unix time>,v1=<hex>, where hex is HMAC-SHA256 of t + "." + raw body with your secret. Recompute it, compare in constant time, and reject old timestamps (over 5 minutes). Also sent: Plainradar-Event (saved_search.new_stories or saved_search.test) and Plainradar-Delivery (a unique id). Answer with any 2xx; redirects aren't followed.

    Google Sheet.

    For “”. Each run that finds something adds the new stories as rows.

    Share the sheet with as Editor, then paste its link. New stories are added as rows; we write a header into an empty first row and never overwrite anything.

    Copy all stories adds every story this tracker shows to the sheet, also the ones already in it.

    Filter out the noise.

    Say what you don't want (or only want), in plain words, one per line. Every story is checked against them and the ones that don't fit are hidden; you can still show them.

    Filters are included: they use no extra credits. Tracking this topic keeps them.

    Get in touch.

    Questions, ideas or something not working? Write to us and we'll reply by email.

    Topics

    Pick a topic.

    Model Context Protocol

    Let your AI read the news.

    Connect any MCP-capable agent. It gets one tool, search_news, which runs the same live search you see here and returns the top verified stories with sources.

    In your MCP client, add a remote (HTTP) server with this endpoint and header.

    Apps with sign-in, such as Claude, need only the endpoint: they open a page where you log in and allow them. Other clients need the header too.

    Endpoint
    Header

    If your client only runs local servers, use this command instead (needs Node.js):

    Tool

    search_news(query: string, days?: 1–7, language?: string, filters?: string[], limit?: 1–100)

    Returns every verified story from the last 24 hours. Each story has its headline, link, source, publication time, summary, image, how many sources covered it, and the result of the relevance check. A page has the 30 most covered stories unless you set limit (up to 100).

    • days (2–7): adds a second page with older stories. With days: 5, the first page has the last 24 hours and the second page has the best stories from the 4 days before that.
    • language: sources are English by default. Use a language name (e.g. "French") to read only sources in that language, or "all" to read every language.
    • filters (up to 5, e.g. "Skip opinion pieces and listicles"): drops stories that don't fit, like the noise filters on this site. Filters are included in the search's credit.