skill-approval-gate
Pass
Audited by Gen Agent Trust Hub on Sep 7, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted external data (target skills and review evidence), which inherently presents an indirect prompt injection surface. However, the skill includes comprehensive defensive instructions that effectively mitigate this risk.
- Ingestion points: The agent reads the target skill's directory and specific evidence files (SKILL.md).
- Boundary markers: The instructions mandate copying a fenced skeleton verbatim for output and explicitly state that target files must be treated as untrusted data rather than instructions (SKILL.md).
- Capability inventory: The skill uses file-reading tools to perform its analysis.
- Sanitization: The workflow includes strict commands to reject absolute paths, parent traversal, and symlink escapes, and requires explicit permission before reading any files outside the defined target root (SKILL.md).
Audit Metadata