AI AGENT RADAR PUBLIC EXPERIMENT

Discover. Verify. Produce.

LOCAL
GitHub ↗
Article · Open experiment

AI AGENT
RADAR

Scan public sources. Cut the noise. Keep only what actually moved in the AI-agent ecosystem.

Keyless Scout · first-party + HN/arXiv · no telemetry
  • ON DEMANDYou press Scan. Nothing runs 24/7.
  • SPECIALIZED ROLESScout · Researcher · Skeptic · Analyst
  • FIRST-PARTY FIRSTOfficial blogs before recaps
  • NOTHING NEW = SILENCENo recycled board. No paid filler.
  • OPEN ARTIFACTSEvery count is a file you can audit

TASK QUEUE

  • Waiting for a scan

PROGRESS

0%

READY TO DISCOVER

LATEST OUTPUT

SCOUT

— → final

Grok: skipped (keyless scout)

raw
new
changed
rejected
final

LATEST RUN PERSISTED ARTIFACT

Waiting for a scan

SOURCE ↗
Press Scan. The board only fills with a real run.

HISTORICAL FEED

Previous coverage stays history — never NEW again.

THE RULE Nothing new → nothing invented.

If a rescan finds no material change, the correct output is silence. No recycled signal. No filler. No paid analysis just to make the run look busy.

THE MODEL Open core. User-owned compute.

The public Scan is keyless Scout. Optional Grok analysis uses your own xAI key — it is never required to operate the first brick.