analyse-issue

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Analyse Issue

分析GitHub Issue

Overview

概述

Fetch the issue details, verify repo alignment, and inspect the local codebase to confirm or refute the reported problem. Provide an evidence-backed analysis with clear next steps.

获取Issue详情，验证仓库是否匹配，并检查本地代码库以确认或反驳报告的问题。提供有证据支持的分析，并给出明确的后续步骤。

Workflow

工作流程

1) Parse the issue input

1) 解析Issue输入

If input is a GitHub issue URL, extract owner/repo and issue number.
If input is just a number, assume the current repo unless the user specifies another repo.
If the provider is not GitHub or is unclear, ask for clarification before proceeding.

如果输入是GitHub Issue URL，提取仓库所有者/仓库名和Issue编号。
如果输入只是一个数字，则默认使用当前仓库，除非用户指定了其他仓库。
如果来源不是GitHub或不明确，请在继续前向用户确认。

2) Verify local repo matches the issue

2) 验证本地仓库与Issue匹配

Ensure the current directory is inside a git repo (
```
git rev-parse --show-toplevel
```
).
Identify the target remote (prefer
```
origin
```
); normalize SSH/HTTPS to
```
owner/repo
```
.
If the issue URL repo does not match the local
```
owner/repo
```
, stop and ask the user to switch directories or confirm the target repo.
If no remote or multiple candidates exist, ask the user which repo to use.

确保当前目录在git仓库内（执行
```
git rev-parse --show-toplevel
```
）。
确定目标远程仓库（优先使用
```
origin
```
）；将SSH/HTTPS格式统一为
```
owner/repo
```
格式。
如果Issue URL对应的仓库与本地
```
owner/repo
```
不匹配，请停止操作并要求用户切换目录或确认目标仓库。
如果没有远程仓库或存在多个候选仓库，请询问用户使用哪个仓库。

3) Fetch issue content

3) 获取Issue内容

Prefer

gh

when available:

gh issue view <num> --json title,body,labels,comments,author,createdAt,updatedAt

If URL provided, run against that repo:

gh issue view <num> -R owner/repo --json ...

Fallbacks:

gh api repos/{owner}/{repo}/issues/{num}

and

.../comments

if you need more fields.

If
```
gh
```
is unavailable but
```
GITHUB_TOKEN
```
exists, use
```
curl
```
with the GitHub API.
If neither works, ask the user to paste the issue content.

Capture at least: title, body, labels, environment details, repro steps, expected/actual behavior, and key comment insights.

优先使用

gh

命令（如果可用）：

gh issue view <num> --json title,body,labels,comments,author,createdAt,updatedAt

如果提供了URL，则针对该仓库执行：
```
gh issue view <num> -R owner/repo --json ...
```

备选方案：

若需要更多字段，使用

gh api repos/{owner}/{repo}/issues/{num}

和

.../comments

。

如果
```
gh
```
不可用但存在
```
GITHUB_TOKEN
```
，使用
```
curl
```
调用GitHub API。
如果以上都不可行，请让用户粘贴Issue内容。

至少捕获以下信息：标题、正文、标签、环境细节、复现步骤、预期/实际行为，以及关键评论要点。

4) Analyze the codebase

4) 分析代码库

Translate the issue into concrete signals (keywords, error messages, stack traces, config names).
Use
```
rg
```
to locate relevant code, tests, and configs.
Trace the execution path: entry points -> core logic -> dependencies.
Identify likely failure points: missing checks, edge cases, incorrect assumptions, data shape mismatches, concurrency/timing issues.
If appropriate and safe, run focused tests; otherwise propose targeted tests to validate the hypothesis.

将Issue内容转化为具体的信号（关键词、错误信息、堆栈跟踪、配置名称）。
使用
```
rg
```
定位相关代码、测试和配置文件。
追踪执行路径：入口点 -> 核心逻辑 -> 依赖项。
识别可能的故障点：缺失的检查、边缘情况、错误的假设、数据结构不匹配、并发/时序问题。
如果合适且安全，运行针对性测试；否则建议通过针对性测试来验证假设。

5) Confirm or qualify the issue

5) 确认或限定Issue

Provide evidence with file references and reasoning.
If you can only reason without running tests, state assumptions and confidence.
If evidence is insufficient, list the exact missing info needed.

提供带有文件引用和推理过程的证据。
如果只能通过推理无法运行测试，请说明假设和置信度。
如果证据不足，列出确切需要补充的信息。

6) Respond with a structured analysis

6) 返回结构化分析结果

Include:

Issue summary (expected vs actual)
Repo match verification
Evidence (files/functions)
Root-cause hypothesis or confirmed cause
Suggested fix approach
Questions or missing data
Next steps (tests, logs, repro)

包含：

Issue摘要（预期 vs 实际情况）
仓库匹配验证结果
证据（文件/函数）
根因假设或已确认的原因
建议的修复方案
问题或缺失的数据
后续步骤（测试、日志、复现）

Output conventions

输出规范

Respond in the user's language when clear; default to Chinese for Chinese prompts.
Keep analysis concise but include concrete file pointers and evidence.
Do not claim confirmation without code-based evidence or reproduction.

若明确用户使用的语言，则用该语言回复；对于中文提示，默认使用中文回复。
分析内容保持简洁，但需包含具体的文件指向和证据。
没有基于代码的证据或复现结果时，不要声称已确认问题。