The Agent Skills Directory

[SAFE]: No malicious patterns or security risks were identified in the skill. The instructions define a rigorous, intent-first clarification loop that utilizes local tools for state management (omx state) and user input (omx question). The skill follows safe practices for handling repository context, including the use of a summary gate for oversized inputs to prevent prompt instability. It adheres to a policy of gathering evidence from existing repository documentation and code without executing unverified remote content.

deep-interview