agent-memory-testing
Pass
Audited by Gen Agent Trust Hub on Oct 14, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill implements strong internal guardrails. In
SKILL.mdandprompts/agent-memory-testing.md, it explicitly instructs the agent not to fabricate results, bypass safety checks, or upgrade static plans to "real execution" proof. - [SAFE]: Data handling instructions prioritize privacy, specifying the use of de-identified (masked) data, mocks, and isolated environments when dealing with sensitive information.
- [SAFE]: The included scripts (
evals/fixtures/check_domain_decisions.py) are deterministic validation tools used for automated testing of the skill's own output format. They do not perform network operations or access sensitive system files. - [SAFE]: No prompt injection, obfuscation, or persistence mechanisms were found. The skill operates within the provided context and relies on local reference files for domain calibration.
Audit Metadata