remember
Pass
Audited by Gen Agent Trust Hub on Aug 4, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill creates a persistence vector for indirect prompt injection by storing unvetted conversation content for future retrieval.
- Ingestion points: Captures rules and preferences from user arguments or the preceding conversation history (SKILL.md, Step 1).
- Boundary markers: Content is appended to
.ui-craft/brief.mdwithout clear delimiters or instructions for future sessions to ignore embedded commands within these constraints. - Capability inventory: The skill performs file write operations to
.ui-craft/brief.mdto store the captured data. - Sanitization: The instructions do not specify any sanitization, filtering, or escaping of the user-provided content before storage, allowing potentially malicious instructions to be persisted into the project's long-term memory.
Audit Metadata