harness-experiment
Pass
Audited by Gen Agent Trust Hub on Oct 30, 2026
Risk Level: SAFENO_CODE
Full Analysis
- [SAFE]: The skill serves as a documentation-based framework for performance evaluation. It instructs the agent or user on how to state hypotheses, measure metrics, and analyze results.
- [NO_CODE]: There are no scripts, binaries, shell commands, or external package dependencies defined in the skill content.
- [DATA_EXPOSURE_AND_EXFILTRATION]: The skill does not access sensitive local file paths or attempt to transmit data to external domains.
- [PROMPT_INJECTION]: The instructions do not contain patterns intended to bypass safety guidelines or override core system behavior.
Audit Metadata