s-qa
Pass
Audited by Gen Agent Trust Hub on Oct 5, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external candidates and has capabilities to write reports and execute commands, creating a surface for instructions embedded in test data to influence agent behavior.
- Ingestion points: The skill ingests deliverables under review and candidate evidence in SKILL.md.
- Boundary markers: The instructions recommend a fresh context for independent acceptance but do not specify technical delimiters for untrusted input.
- Capability inventory: The skill is authorized to create directories, write report files, and execute verifier-chosen probes or commands as per SKILL.md.
- Sanitization: No specific data sanitization or escaping mechanisms are described for handling candidate-provided content.
Audit Metadata