The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes a utility script scripts/check.mjs that executes the git command via execFileSync. It uses git rev-parse --show-toplevel to determine the repository root and git grep to identify requirement IDs (e.g., ACC-01) within project files. This functionality is used to generate reports on whether requirements are covered in the codebase.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted local project data, including the requirements.yaml file and the source code found via git.
Ingestion points: The agent reads the user's requirements.yaml and all repository files via git grep.
Boundary markers: The script does not utilize explicit boundary markers in its output to differentiate project data from agent instructions.
Capability inventory: The skill has the ability to read local files and execute the git command-line tool.
Sanitization: Data is validated against strict regular expression patterns (PREFIX-NN) before being processed, which limits the potential for instruction injection.
[EXTERNAL_DOWNLOADS]: The documentation links to requirements.template.yaml and requirements.schema.json hosted on the author's GitHub repository. These are provided as developer aids for project initialization and IDE support.

requirements-yaml