The Agent Skills Directory

[COMMAND_EXECUTION]: Utilizes the GitHub CLI (gh) for fetching pull request data and posting comments via the GitHub API. These operations are restricted to the repository context and represent the core intended functionality of the skill.
[DATA_EXFILTRATION]: Reads findings from local JSON state files located within the ~/.claude/ directory. This data is processed and subsequently transmitted to GitHub as inline or conversation comments on the relevant pull request.
[SAFE]: Employs a 'human-in-the-loop' safety pattern by explicitly requiring a single confirmation after a batch preview of all drafted comments, preventing unauthorized or accidental automated posting of content.

pr-comment