nav-brief
Pass
Audited by Gen Agent Trust Hub on Sep 13, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill operates as a safety and clarification layer. It analyzes user prompts for ambiguity using a local Python script (ambiguity_scorer.py) and requires explicit human confirmation via a structured template before the agent proceeds with task implementation. This pattern follows security best practices for AI autonomy by maintaining a human-in-the-loop for non-trivial operations.
- [INDIRECT_PROMPT_INJECTION]: While the skill ingests and analyzes user prompts, it does so to identify ambiguity and halt execution until a brief is confirmed. This process effectively mitigates risks associated with vague or malicious instructions by introducing a mandatory verification step.
- Ingestion points: User input prompts are processed by the ambiguity_scorer.py script.
- Boundary markers: The skill explicitly instructs the agent to wait for a 'Confirm / edit?' response from the user.
- Capability inventory: The skill itself only performs text rendering and logic checks; it restricts the agent's capability to modify files until confirmation is received.
- Sanitization: The analysis logic uses a deterministic regex-based scoring system rather than executing the prompt content.
Audit Metadata