メニュー
Read a GitHub issue with all comments and linked content to add as working context. Use when the user provides a GitHub issue URL or reference.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
This skill should be used when the user asks to "onboard", "analyze this repo", "understand this codebase", "familiarize yourself", "what is this project", or needs to deeply understand a codebase's architecture, patterns, and conventions before starting work.
Analyze the current repo deeply and suggest agentic workflows tailored to its architecture, patterns, and pain points. Use when user calls /aw-planner.
Generate a comprehensive manual testing guide for completed work, with note sections for user feedback that the agent reviews afterward. Use when the user wants to manually test changes before shipping.
Review PR comments, validate which are actionable and correct, and propose fixes for valid ones. Use when user wants to address feedback on a pull request.
Convert Markdown text to Zendesk-compatible HTML. Use when user needs to format content for Zendesk tickets, macros, triggers, or automations.
Create a branch from unstaged changes, commit with conventional message, and open a PR. Use when user wants to quickly ship current work.
| name | read-issue |
| description | Read a GitHub issue with all comments and linked content to add as working context. Use when the user provides a GitHub issue URL or reference. |
Fetch a GitHub issue and all related content to build full context for a task.
The user provides a GitHub issue reference in one of these forms:
https://github.com/owner/repo/issues/123owner/repo#123#123Extract owner, repo, and issue number from the input. Use gh to fetch the issue:
gh issue view <number> --repo <owner/repo> --json title,body,state,labels,assignees,milestone,createdAt,author,url
gh api repos/<owner>/<repo>/issues/<number>/comments --paginate
Extract each comment's user.login, created_at, and body.
Scan the issue body and all comment bodies for:
https://github.com/<owner>/<repo>/issues/<n> or /pull/<n>#<n> or <owner>/<repo>#<n>https:// links (docs, wikis, designs, etc.)Deduplicate the collected links.
For each linked GitHub issue or PR, fetch a summary:
gh issue view <number> --repo <owner/repo> --json title,body,state,labels,url
gh pr view <number> --repo <owner/repo> --json title,body,state,url,files
Keep these summaries concise — title, state, and body are the priority.
Use WebFetch to retrieve content from non-GitHub URLs. Extract the key information relevant to the issue context. Skip URLs that are:
.png, .jpg, .gif, .mp4, etc.)Output everything as a structured context summary:
## Issue: <title> (<state>)
**URL**: <url>
**Author**: <author> | **Created**: <date>
**Labels**: <labels> | **Milestone**: <milestone>
### Description
<body>
### Comments (<count>)
**<user>** (<date>):
<comment body>
---
### Linked Issues/PRs
#### <title> (<state>) — <url>
<body summary>
---
### External References
#### <url>
<fetched content summary>
gh CLI — never construct API URLs for raw curl calls--paginate to ensure none are missedgh auth fails or a repo is inaccessible, report it clearly and continue with what's available