Incident network
One bad agent,
every platform warned.
When an AI agent scams, spams or takes payment and doesn't deliver, the business it hurt reports it to Neptuv. We review the report, hear the owner out, and publish a short finding that lowers the agent's score for every platform that checks it.
Fair to both sides
A report network is only useful if it can't be weaponised. These rules are built in.
Reports come from paying Pro and Platform workspaces that signed in with a wallet — not anonymous accounts. Each one has a public-to-us accuracy record.
The reporter's text and evidence are only seen by Neptuv's reviewers. Nothing is published, and the score doesn't change, until a report is confirmed.
The agent's owner is told a report exists and can reply. Confirmed findings are always shown next to the owner's reply.
A confirmed report adds one risk flag (−25 points) and is logged as a correction. Owners running several flagged agents lose trust across all of them.
What can be reported
- Fraud or theft
fraud - Took payment, didn't deliver
non_delivery - Impersonation
impersonation - Spam or abusive traffic
abuse - Data misuse or leak
data - Security issue
security - Other
other
Latest confirmed findings
No confirmed findings yet.
Platforms: block bad agents before they reach you
Neptuv's /gate and /verify already include confirmed reports. Set max_open_reports in your policy to also hold back agents while reports are under review.