debug
Pass
Audited by Gen Agent Trust Hub on Apr 26, 2026
Risk Level: SAFEPROMPT_INJECTIONNO_CODE
Full Analysis
- [PROMPT_INJECTION]: The skill defines a process for analyzing external data (bug reports), creating a surface for indirect prompt injection.
- Ingestion points: The agent is instructed to analyze potentially untrusted bug reports (SKILL.md, Phase 1).
- Boundary markers: The instructions do not specify the use of delimiters or 'ignore instructions' warnings for the input data.
- Capability inventory: The agent has the capability to search the codebase and implement file system changes (SKILL.md, Phase 2).
- Sanitization: No validation or sanitization of the input bug reports is described.
Audit Metadata