personal-ui-taste
Pass
Audited by Gen Agent Trust Hub on Nov 2, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill implements a self-evolution protocol (evolution.md) that directs the agent to modify its own instruction files based on user feedback and project reviews.
- Ingestion points: External user feedback and UI review sessions are captured and processed into permanent updates within the skill's directory (SKILL.md, patterns.md, evolution.md, scenarios/*.md).
- Boundary markers: The protocol lacks explicit instructions to sanitize or delimit the incoming feedback to prevent the inclusion of malicious instructions or payload-carrying text.
- Capability inventory: The agent is granted write access to its own skill files and is expected to perform file system operations to maintain the 'Personal UI Taste' library.
- Sanitization: There are no mechanisms described for filtering or validating the feedback before it is committed to the skill's permanent instructions.
Audit Metadata