pathfinding

Pass

Audited by Gen Agent Trust Hub on Jun 20, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill provides a structured methodology for clarifying user requirements and brainstorming ideas. It uses internal confidence levels to guide the interaction and ensures artifacts are only delivered when requirements are sufficiently understood.\n- [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface by design, as it ingests user-provided information to generate implementation plans and code artifacts. This is consistent with its primary function as a requirement gathering assistant.\n
  • Ingestion points: User responses to clarifying questions and initial prompts (SKILL.md).\n
  • Boundary markers: The framework uses conversational stages and turn-based interactions but does not implement explicit delimiters for input sanitization.\n
  • Capability inventory: The skill generates documentation, code snippets, and architectural plans, and can invoke a 'skeptic' sub-agent using the Task tool (SKILL.md).\n
  • Sanitization: No explicit input filtering or validation logic is present in the skill's instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 20, 2026, 06:52 AM
Security Audit — agent-trust-hub — pathfinding