Discover your actual agents

Find the ghost, shadow, and undeclared AI already running in your environment — then map it to Colorado SB 26-189.

Discover your actual agents

beta

An 8-layer detector for ghost / shadow AI— local & air-gapped inference, stealth remote-access, and undeclared cloud AI. Paste a config, pick a folder (names only), or run the read-only local scanner. Matching runs in your browser; only matched fingerprint IDs are sent — never your raw config or file contents.

Not sure what to paste? Try a sample

This environment should be air-gapped (flag any egress as a breach)We already have a declared agent registry

Deepest reach: run the read-only local scanner on the machine itself, then paste its report above — agentscan.sh (macOS/Linux) · agentscan.ps1 (Windows). It reads process/port/file/dependency/env-var names only — never contents or secret values — and uploads nothing.

Observe/Detect tier (EU AI Act / SOC 2 / NIST AI RMF / ISO 42001). Discovered agents start at trust tier T0 with no attestation — full Protect (quarantine / circuit-breaking) requires an endpoint agent.

Want a guided governance-readiness score instead?