retention-exception-mgmt
Pass
Audited by Gen Agent Trust Hub on Sep 19, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill is entirely instructional and administrative in nature. The included Python script (scripts/process.py) implements internal state management and validation for retention intervals without external dependencies, network connectivity, or file system interaction.
- [INDIRECT_PROMPT_INJECTION]: 1. Ingestion points: User-provided text for justifications and assessments within the markdown templates (SKILL.md, assets/template.md). 2. Boundary markers: No explicit delimiters or instructions are used to wall off user-provided content. 3. Capability inventory: The skill does not have any tools for network communication, subprocess execution, or file writing. 4. Sanitization: No input filtering is present. While the skill ingests untrusted data, the absence of high-risk tools makes this surface non-exploitable.
Audit Metadata