earned-autonomy
Pass
Audited by Gen Agent Trust Hub on Jul 29, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill establishes a structured autonomy rubric (L0 to L3) that prioritizes human review for any task involving external communications or 'worded items'.
- [SAFE]: The instructions mandate strict safety guardrails, including channel volume limits, suppression lists, and a 'kill switch' to immediately cease automated operations, which are effective controls against automated abuse.
- [SAFE]: The skill includes evidence-based promotion criteria that require agents to demonstrate judgment (via 'logged refusals') before gaining autonomy, effectively mitigating risks associated with blind obedience.
- [SAFE]: No indicators of data exfiltration, hardcoded credentials, remote code execution, or obfuscation were found in the skill files or metadata.
Audit Metadata