The Agent Skills Directory

[COMMAND_EXECUTION]: The verification phase involves executing standard development tools such as ruff for linting, pytest for testing, and mypy or ty for type checking via the uv run command. These actions are within the expected scope of a development-oriented skill.
[DATA_EXPOSURE]: The skill's discovery phase instructs the agent to identify project structure elements like pyproject.toml, src/, and tests/. This is standard project context gathering and does not involve accessing sensitive system files or credentials.
[PROMPT_INJECTION]: The skill ingests user input through the $ARGUMENTS variable, which is encapsulated within <feature_description> tags. This provides a clear boundary for the model to distinguish between instructions and user-provided data, reducing the risk of prompt injection.

python3-add-feature