The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local Python scripts located in ~/.claude/scripts/ (e.g., feature-state.py, learning-db.py) to manage feature lifecycles and record developer learnings. These are local platform-specific utilities.
[DATA_EXPOSURE]: The skill accesses the ~/.claude/ directory to manage internal state. This is standard behavior for the associated agent platform's state management.
[PROMPT_INJECTION]: The skill processes local artifacts such as design documents and validation reports to generate pull request descriptions. This represents a standard surface for processing project data without signs of malicious intent.

feature-release