assess-findings
Pass
Audited by Gen Agent Trust Hub on Nov 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of untrusted data from external sources such as pull request descriptions, issue tracker comments, and discussion threads to form judgments and recommendations.
- Ingestion points: External pull request bodies, GitHub issue comments, and discussion threads are read to recover requirements and decisions (SKILL.md, references/tracker-decisions.md).
- Boundary markers: The instructions include explicit directives to treat supplied labels and diagnoses as 'claims to examine against their evidence, not established premises', providing a conceptual boundary (SKILL.md).
- Capability inventory: The skill manages local file records (read/write), performs Git operations (stage, commit), and executes repository-specific query commands (references/working-records.md).
- Sanitization: The methodology relies on the agent's reasoning to distinguish evidence from untrusted claims; no explicit automated sanitization or schema validation for the external text is implemented.
Audit Metadata