The Agent Skills Directory

[SAFE]: The skill is a legitimate utility for scaffolding and validating AI agent skills. It establishes a structured workflow for skill authoring and maintenance.
[COMMAND_EXECUTION]: The skill instructions and test scripts (e.g., scripts/tests/test_init_skill.py and scripts/tests/run.py) use subprocess.run to execute local Python and Node.js scripts. This is standard and expected behavior for a development tool suite used for initialization and testing.
[DYNAMIC_EXECUTION]: The skill generates new skill directories and resolver scripts based on predefined templates. The resolve.py template includes robust path-containment checks to ensure that resolved instruction paths remain within the designated project root, preventing directory traversal attacks.
[INDIRECT_PROMPT_INJECTION]: The skill processes project configuration files (config.yaml) and user-provided inputs to generate files. The risk is minimized by the use of yaml.safe_load (or a restricted custom parser) and the validation logic in scripts/quick_validate.mjs which prevents hardcoded project-specific branches in reusable skill code.
Ingestion points: Reads project-level config.yaml and .md profiles via resolve.py (template).
Boundary markers: The generated output is structured using Markdown headers.
Capability inventory: File system write access for scaffolding and subprocess execution for running tests.
Sanitization: Includes name normalization and path validation to ensure targets are within allowed directories.

skillcraft