test-reliability
Pass
Audited by Gen Agent Trust Hub on Sep 10, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill provides a structured framework for diagnosing and managing unstable tests. It uses an evidence-based approach requiring specific rerun results before classifying issues.
- [INDIRECT_PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface because it processes external data from CI logs and rerun records.
- Ingestion points: Test execution logs, CI records, and triage tables.
- Boundary markers: The skill does not define explicit prompt delimiters; however, it mandates the use of strict evidence matrices (M1-M6) and non-stationary signal checks (N1-N3), providing a logical validation layer for ingested content.
- Capability inventory: Capabilities are limited to reading project logs and writing local markdown reports and metadata in the .qa/ directory; no high-risk tools like network access or arbitrary code execution are requested.
- Sanitization: The skill emphasizes data-driven classification over natural language interpretation, effectively mitigating the impact of instructions potentially embedded in logs.
Audit Metadata