working-on-an-issue

Implements GitHub issues end-to-end with scoped recon, verification, and a pull request.

5|Updated Nov 9, 2025
One-click install
npx skills add https://github.com/britt/agent-skills --skill working-on-an-issue-britt
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: working-on-an-issue
Source: https://github.com/britt/agent-skills/tree/main/skills/working-on-an-issue
Command: npx skills add https://github.com/britt/agent-skills --skill working-on-an-issue-britt

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Implementing a GitHub issue well requires reading the full discussion, scoping the change, planning verification, and delivering reviewable work — doing this ad hoc leads to scope creep, undocumented assumptions, and unverified code. This Skill provides a hands-off workflow where the pull request is the single approval gate. ## Core Features & Use Cases - Hands-off implementation pipeline: Understand the issue, update GitHub Projects status, run scoped recon, plan verification and implementation, implement, verify, and open a PR — with no mid-flow approval gates. - Documented assumptions: Ambiguities are resolved with the narrowest interpretation and recorded in the plan file and PR description, never left silent. - Verification before PR: Acceptance-test verification must pass before the pull request is opened; failures trigger systematic debugging rather than symptom patching. - Use Case: A developer pastes a GitHub issue URL and asks for it to be implemented. The Skill reads the issue and comments, moves it to In Progress on its project boards, locates the affected files, writes plans to docs/plans/, implements with issue-referenced commits, runs the verification plan, and opens a PR with Fixes #<number>. ## Quick Start Implement GitHub issue number 123 from this repository using the working-on-an-issue workflow and open a pull request when verification passes.

Frequently Asked Questions about working-on-an-issue

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

FAQPage Schema
How do I implement a GitHub issue automatically with an AI agent?▼

Provide the issue URL or number and the workflow reads the issue and comments, runs scoped recon to locate the change, writes verification and implementation plans, implements with issue-referenced commits, verifies, and opens a pull request for review.

How to scope code exploration when fixing a GitHub issue?▼

Start from files, symbols, and error messages named in the issue, then expand at most one hop to direct callers and their tests. Stop once you know which files change, which pattern to follow, and which tests cover the area.

Can the workflow update GitHub Projects status automatically?▼

Yes, it queries the issue's Projects v2 membership via the GitHub GraphQL API and sets each project item's status field to an in-progress option using gh project item-edit. If the token lacks project scope, it records the failure and continues.

When should I not use an autonomous issue implementation workflow?▼

Avoid it when the issue is epic-sized or bundles independent changes (decompose it first), when it is too vague to restate as testable acceptance criteria (elicit requirements first), or when you only need triage and labeling.

What happens if verification fails before opening a pull request?▼

The workflow uses systematic debugging to find the root cause rather than patching symptoms, then fixes, re-runs verification, and updates the log. A pull request is never opened with failing verification.