Review AI-agent
access before rollout.
For security and platform teams reviewing AI-agent access before production: Skillfile maps tool grants from config, flags risky operations, and records approved behavior as signed .skill files.
Stamp org/default-agent-policy@4 d9f2a1b8
Signed .skill files — reviewable capability boundariesThe governed capability loop
Scan agent configs, expose unmanaged access, draft a signed .skill manifest, review risky operations, then sync and audit the result.
.skill manifestWhat an audit shows
A concrete readout of current agent access: what is allowed, what needs human approval, and what should stay blocked.
Production Incident
Investigation
Diagnose production incidents with read-only access to logs, metrics, and dashboards. Sensitive operations require human approval.
Risk breakdown
Proof for the people who have to trust it
Security, platform, and leadership get the same evidence trail: who approved access, what changed, where it synced, and what stayed blocked.
Start with one capability audit
Bring one repo or agent profile. We map tool access, flag risky grants, and return a reviewable .skill path forward — no secrets required.