The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides and executes a Python script (scripts/pr-watch.py) which invokes the gh (GitHub CLI) tool using subprocess.run to poll Pull Request status and CI checks.
Evidence: scripts/pr-watch.py line 92 uses subprocess.run(["gh", *args], ...) with list-based arguments, which is a standard safety measure to prevent command injection.
[PROMPT_INJECTION]: The skill creates a surface for indirect prompt injection because it is designed to ingest and act upon external data from GitHub comments and reviews.
Ingestion points: scripts/pr-watch.py fetches untrusted text from GitHub PR reviews and comments (lines 145, 156, and 166).
Boundary markers: The skill does not implement explicit delimiters or "ignore embedded instructions" warnings for this external content before it is processed by the agent.
Capability inventory: The agent is instructed to "address BOTREVIEWs automatically" and respond to human feedback, providing a mechanism for external content to influence repository actions.
Sanitization: No sanitization or escaping of the external text is performed before interpolation into the prompt context.

git