babysit-pr

Works through code review comments one by one, backing every accept or reject with evidence.

Ecosystem: Agent Skills
Skill category: Coding
Creator / organization: Original creator not yet identified

What it is

babysit-pr comes from .agents, Serghei Iakovlev's personal collection of AI agent tools. After a developer opens a pull request (PR), teammates leave review comments. This Skill has the agent handle them in six steps: collect the comments, either pasted in or pulled from GitHub; verify any comment about library or framework behavior against official docs through Context7; sort each comment into one of seven categories, such as valid and actionable, valid but deferred, subjective style, incorrect, or outdated; apply the fixes and open tickets for deferred work; confirm that it has not replied to the reviewer; and finally post a report in the chat listing every decision and its evidence. It never replies to reviewers on your behalf and never commits or pushes code on its own.

Who it's for:developers who handle a lot of code review and need to decide which suggestions to take. If you don't write code, it won't be useful to you.

Before you start: you need an authenticated GitHub CLI (gh), Python 3, git and the Context7 documentation service. The author describes the whole collection as permissively licensed for others to adapt; the project page does not name the specific license.

Based on the project's public documentation. Not tested by AI Inspiration World.