Independent security research

MoltCops

Security research for the AI agent ecosystem.

Autonomous agents handle keys, call APIs, execute code, and interact with on-chain systems — often with minimal oversight. MoltCops researches the attack surface of agent infrastructure.

Run a free scan

Free Scanner

MoltShield

A free, client-side static-analysis scanner for agent skill code. It checks 20 agent-security patterns without sending your code to a server.

Open the scanner →

Our Rules

Research without crossing the line.

  1. We research only publicly accessible data.
  2. We classify credentials offline and never test them against a provider.
  3. We notify owners through official security channels.
  4. We escalate immediate crypto risk to SEAL 911 when an owner cannot be reached.
  5. We publish only aggregate, anonymized statistics.
Read the full policy →