automated-fix
Pass
Audited by Gen Agent Trust Hub on Jun 23, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill presents an indirect prompt injection surface by processing untrusted issue data.
- Ingestion points: The issue context variable (including reproduction steps and description) entering via SKILL.md.
- Boundary markers: Absent; no delimiters are used to separate untrusted content from instructions.
- Capability inventory: Filesystem modification to apply fixes and execution of shell commands/scripts for issue reproduction.
- Sanitization: Absent; the content is processed without validation or escaping.
Audit Metadata