rules-distill
Pass
Audited by Gen Agent Trust Hub on Jul 22, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [SAFE]: The skill operates within the agent's expected configuration directories and requires explicit user confirmation for all file system modifications, preventing automated or unauthorized changes.
- [PROMPT_INJECTION]: The skill processes content from other installed skills, creating a potential surface for indirect prompt injection if those files contain adversarial instructions.
- Ingestion points: Skill definition files located at ~/.claude/skills//SKILL.md and ~/.claude/skills/learned/.md.
- Boundary markers: No explicit boundary markers or instruction-isolation warnings are mentioned for the ingested skill content.
- Capability inventory: The skill possesses the ability to read and write to the rules directory (~/.claude/rules/) and execute the 'date' command for logging.
- Sanitization: The risk is mitigated by a mandatory human-in-the-loop gate requiring the user to approve each distilled rule draft before it is applied.
Audit Metadata