The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes git and gh commands via Python's subprocess module in scripts/collect.py. These commands are used to list files, retrieve commit history, and clone repositories. Arguments are passed as lists to subprocess.run without shell=True, which is a secure implementation that prevents shell injection.
[EXTERNAL_DOWNLOADS]: The skill fetches repository metadata, directory trees, and file contents from GitHub via the official gh CLI and GitHub API. These operations target a well-known service and are limited to the scope defined by the user's authentication and command-line arguments.
[PROMPT_INJECTION]: The generated HTML report contains a vulnerability surface for indirect prompt injection (Category 8) due to lack of output sanitization.
Ingestion points: Repository names, branch names, and file paths are ingested from local file systems or the GitHub API by scripts/collect.py and stored in coverage-data.json (File: scripts/collect.py).
Boundary markers: None implemented for the reporting UI.
Capability inventory: The skill uses subprocess to execute CLI tools and generates a local HTML artifact (render.py).
Sanitization: The report renderer (scripts/render.py) uses innerHTML to display repository names (e.g., in the findings and repo panels). If a repository has a malicious name containing HTML or script tags, it could execute in the browser context when the user opens the local report.

context-coverage