The Agent Skills Directory

[SAFE]: The skill is entirely instructional, defining a deterministic scoring system (rubric) for human or agent use. No dangerous operations such as file system writes, network requests, or credential access were identified.
[SAFE]: No external dependencies are downloaded or executed. The mentioned .cavekit/config.json and cavekit-router.cjs are described as internal project components for configuration and routing based on the skill's output.
[SAFE]: While the skill processes external task descriptions (potential for indirect prompt injection), the impact is limited to resource allocation (model tier and token budget), which is the intended functional purpose of the tool.

complexity-detection