github-issue-resolve

Implements GitHub issues end-to-end from investigation through worktree-based development to pull request creation.

1|Updated Aug 24, 2026
One-click install
npx skills add https://github.com/mjun0812/skills --skill github-issue-resolve-mjun0812
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: github-issue-resolve
Source: https://github.com/mjun0812/skills/tree/main/skills/github/github-issue-resolve
Command: npx skills add https://github.com/mjun0812/skills --skill github-issue-resolve-mjun0812

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Turning a GitHub issue into a merged-ready pull request normally requires manual investigation, branch setup, implementation, review, and PR authoring. This Skill automates that entire pipeline from a single issue number while keeping your main working tree untouched. ## Core Features & Use Cases - End-to-end issue resolution: Reads the issue, extracts acceptance criteria, builds a task queue, implements each task in an isolated git worktree, and creates a PR with Closes #N. - Subagent implementer/reviewer loop: Each task is implemented by one subagent and adversarially reviewed by another, with structured status/verdict parsing and a maximum of two rejection rounds per task. - Safe isolation and cleanup: All work happens in a dedicated worktree under .tmp/, with automatic worktree and branch cleanup after completion or abort. - Use Case: You receive a bug report as issue #456. Invoke the Skill with the issue number, and it investigates the codebase, implements a verified fix in a worktree, commits via the git-commit skill, and opens a pull request via the github-pr-create skill. ## Quick Start Resolve GitHub issue #123 by investigating it, implementing the fix in a worktree, and creating a pull request.

Frequently Asked Questions about github-issue-resolve

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I automatically implement a GitHub issue and create a PR?

Invoke the Skill with the issue number, such as #123. It reads the issue body and comments, builds a task queue, implements each task in an isolated git worktree with implementer and reviewer subagents, then commits and opens a pull request containing a Closes reference.

How does the implementer and reviewer subagent loop work?

Each task is handled by a fresh implementer subagent that returns a structured Status Report, then a reviewer subagent adversarially checks the actual diff and test results. Rejected work is sent back with remediation notes, up to two rounds per task before aborting.

Does it modify my current working tree or branch?

No. All implementation happens in a dedicated git worktree created under the repository's .tmp directory on a new branch derived from the default branch. After completion or abort, the worktree and local branch are removed, except when PR creation fails.

What happens if the issue lacks enough information to implement?

The Skill stops during the investigation phase and asks the user targeted questions, listing each missing piece of information with options or recommendations. It does not guess at specifications or proceed with assumed requirements.

Can I preview the plan without making any changes?

Yes. Passing the --dry-run flag stops after the investigation phase, presenting only the implementation plan and task breakdown. No worktree is created and no code changes are made.