qa-thinking
Pass
Audited by Gen Agent Trust Hub on Sep 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data including feature specifications, implementations, and PR content, which creates a surface for indirect prompt injection if that data contains malicious instructions.
- Ingestion points: Feature specifications, code implementations, and PR data provided in the context (SKILL.md).
- Boundary markers: The instructions lack explicit delimiters or specific warnings to ignore embedded commands within the processed input.
- Capability inventory: The skill is primarily analytical, focusing on risk identification and questioning, and directs the agent to other specialized skills for further task execution.
- Sanitization: There are no explicit instructions to sanitize or validate external content before the agent analyzes it.
Audit Metadata