curate
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided insights and rationale to generate and save YAML documentation, creating a potential surface for indirect injection.
- Ingestion points: Interactive user prompts for 'concept' and 'why' fields defined in SKILL.md.
- Boundary markers: No explicit delimiters or sanitization instructions are provided for the YAML content generation.
- Capability inventory: The skill performs file-write operations to the
.gauntlet/annotations/directory. - Sanitization: No specific filtering or validation of user input is specified before writing to the filesystem.
- [SAFE]: The identified file-system access is restricted to a specific local path and is central to the skill's stated purpose of knowledge curation. No unauthorized command execution, data exfiltration, or external dependencies are present.
Audit Metadata