muse-capture
Pass
Audited by Gen Agent Trust Hub on Oct 5, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external data including build outputs and test logs, which can be manipulated by an attacker to include hidden instructions meant for the AI agent.
- Ingestion points: SKILL.md Step 1 directs the agent to verify evidence using 'real build outputs, test runs, or command logs'.
- Boundary markers: The instructions do not specify the use of delimiters or 'ignore' instructions for the external content being analyzed.
- Capability inventory: The skill has the capability to write and modify a knowledge base using the
memory_captureandmemory_supersedetools. - Sanitization: Although the skill includes instructions to redact credentials (via the 'Vibeguard' protocol), it does not explicitly address the filtering of natural language instructions that may be embedded in logs.
Audit Metadata