메뉴
Resolves the PR for an issue's currently checked-out branch, then monitors it for merge/close/approval/new owner comments, blocking until one of those occurs. Used by auto-fix-all. Usage: /auto-monitor-issue-pr <id>
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Monitors a given PR for merge/close/approval/new owner comments, blocking until one of those occurs, then reports the outcome. Tracks each owner comment's open/addressed lifecycle with :eyes:/:+1: reactions on the comment itself, but leaves deciding how to address a comment to the caller. Usage: /auto-monitor-pr <pr_number>
Autonomously runs the full pipeline (new issue → plan → fix → monitor) for a queue of issue IDs, one at a time, forever — waiting for new IDs to be pushed onto the queue whenever it runs dry — with no user interaction except when a PR is closed without merging. Usage: /auto-fix-all <id1> <id2> ...
Autonomously implements a planned issue with no user interaction. Discovers the specialist agents involved in the plan, dispatches them in parallel, reviews and re-dispatches until correct, then opens or marks ready a pull request. Usage: /auto-fix-issue <id> or /auto-fix-issue #<id>
Discusses and refines an existing GitHub issue through an iterative dialogue, optionally spawning specialist agents to deepen understanding, then saves a structured markdown issue file and, on confirmation, kicks off planning on a committed branch. Usage: /discuss-issue #19
Configures CLAUDE.md and .github/copilot-instructions.md for a project, creating or updating them to point to a shared AGENTS.md file. Handles multiple scenarios depending on which files already exist. Usage: /init-claude
Creates a implementation plan for a given issue. Reads the issue file, analyzes the codebase, asks clarifying questions, and writes a structured plan in the plans folder. Usage: /plan-issue 99 or /plan-issue #99
You are the coordinator. Delegate this skill's work to the architect agent — do not perform the steps yourself.
Spawn:
Agent(subagent_type: "architect", prompt: "Read steps/run.md (resolved relative to the
auto-monitor-issue-prskill folder) and follow it. ARGUMENTS: ")
Wait for the agent to finish, then relay its final report to the caller verbatim — do not summarize, reinterpret, or decide what to do about any comment it reports.
