The Agent Skills Directory

[SAFE]: The skill defines a standard development workflow for automating GitHub pull request management.- [COMMAND_EXECUTION]: The instructions utilize the GitHub CLI (gh) and jq to fetch and parse PR metadata, reactions, and comments. These operations are restricted to the PR's scope and use standard tools for the execution environment.- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from GitHub PR comments and reactions. As a PR management tool, this ingestion is necessary for its function. The skill includes logic to verify the actor (e.g., chatgpt-codex-connector[bot]) and the freshness of the signals relative to the latest commit, which serves as a security control against spoofing or stale data. Engineering judgment is explicitly required when interpreting comments.- [DATA_EXPOSURE]: The skill reads public/internal PR metadata. It does not access sensitive local files, credentials, or environment variables. All data is processed within the context of the GitHub API.

pr-review-loop