The Agent Skills Directory

[SAFE]: No security issues or malicious patterns were detected. The skill's functionality is limited to local file analysis and generating a summary report.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from the draft files.
Ingestion points: Reads output/DRAFT.md and outline/outline.yml to perform quality checks (scripts/run.py).
Boundary markers: No specific delimiters or "ignore" instructions are used around the processed data.
Capability inventory: The skill is restricted to file system read/write operations and does not possess network or shell execution capabilities.
Sanitization: The script uses regex to identify quality issues and does not interpret or execute the content of the draft as instructions.
[DYNAMIC_EXECUTION]: The script uses computed paths to manage internal module imports.
Evidence: The script dynamically modifies sys.path by calculating the repository root relative to its own file path (scripts/run.py).
Context: This is a standard pattern for accessing vendor-provided utility functions in the tooling package and does not involve untrusted code.

pipeline-auditor