bd-fix
Pass
Audited by Gen Agent Trust Hub on Apr 22, 2026
Risk Level: SAFEPROMPT_INJECTIONNO_CODE
Full Analysis
- [PROMPT_INJECTION]: The skill is designed to process external "issue files" and user-provided instructions. This creates an attack surface for indirect prompt injection, where an attacker could place malicious instructions within a bug report to influence the code the agent writes or the commits it performs.
- Ingestion points: "issue file" and "user instructions".
- Capability inventory: writing production code (file system writes) and creating atomic commits (command execution via git).
- Boundary markers: Not specified in the instructions.
- Sanitization: Not specified in the instructions.
- [NO_CODE]: The skill contains only natural language instructions in markdown format and does not include any executable scripts, binaries, or configuration files that run code directly.
Audit Metadata