testability-analysis
Pass
Audited by Gen Agent Trust Hub on Oct 31, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill consists entirely of instructional markdown and configuration files with no executable scripts (Python, JavaScript, or shell). It does not request any tool permissions in its metadata or frontmatter.
- [SAFE]: The instructions in
SKILL.mdandprompts/testability-analysis.mdcontain security-positive constraints, explicitly forbidding the agent from proposing unsafe backdoors or bypassing production security controls (authorization and tenant isolation) during analysis. - [SAFE]: The provided evaluation suite includes specific test cases (e.g.,
evals/cases/architecture-unsafe-seam-boundary.yaml) designed to verify that the agent correctly rejects malicious or insecure requests from users. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted external data such as requirements and logs. While it lacks explicit boundary markers for this data, the risk is mitigated by the lack of exploitable tools and the strong internal safety instructions provided to the model.
Audit Metadata