The Agent Skills Directory

[SAFE]: No malicious patterns, unauthorized network access, or command executions were detected. The skill primarily consists of natural language instructions for requirement management.
[INDIRECT_PROMPT_INJECTION]: The skill manages a surface for indirect prompt injection. 1. Ingestion points: User-provided requirements in SKILL.md. 2. Boundary markers: The 'Scope Confirmation' section acts as a structural boundary for the agent's understanding. 3. Capability inventory: No subprocesses, file-writes, or network operations are defined in the skill files. 4. Sanitization: Mitigation is achieved via a mandatory 'Hearing Protocol' which requires explicit human-in-the-loop confirmation before requirements are promoted to buildable scope.

requirement-convergence