360 Secure SkillHub
360 安全技能中心

Triage Issues

借助已配置的 Gitee MCP Server 拉取仓库开放 Issue,按类型与 P0–P3 优先级分类,识别信息不足、疑似重复与建议关闭项,输出结构化分诊报告;经用户确认后可更新标签、优先级、指派并评论补充信息。

oschina
oschina
热度 65.92
最近更新 2026-05-11 17:29:44
分类:协作依赖:MCP 服务 / 第三方 API应用场景:任务管理 / 报告撰写 / 工作流

安装技能

复制下面的话术发给你的 Agent,它会通过 secure-skills 检索并安装该技能。

执行命令安装此 skill: npx secure-skills-cli add gitee-triage-issues
Metadata
name
triage-issues
description
Use this skill when the user asks to triage issues, organize the issue list, "triage issues", "show me all open issues", "categorize issues", "triage-issues". Requires Gitee MCP Server to be configured.
version
1.0.0
requires
{"mcp-servers":["gitee"]}

Note: If you have mcporter installed locally, you should use mcporter to invoke the MCP tool instead of directly calling the MCP tool. The mcporter CLI provides a more convenient interface and better error handling.

Triage Issues via Gitee MCP

Batch-process a repository's Issues: classify by priority, add labels, identify duplicates, and generate a clear action list.

Prerequisites

  • Gitee MCP Server configured (tools: list_repo_issues, get_repo_issue_detail, update_issue, comment_issue)
  • User must provide: repository owner, repository name
  • Optional: filter criteria (bugs only / unassigned only / specific milestone, etc.)

Steps

Step 1: Fetch the Issue List

Use list_repo_issues to retrieve open Issues:

  • state: open (open Issues only)
  • Paginate if there are many Issues (20 per page)

Step 2: Batch Analysis

For each Issue, quickly evaluate:

Type classification

  • bug: something is broken, throwing errors, or crashing
  • feature: request for new functionality
  • enhancement: improvement to existing functionality
  • question: usage question
  • docs: documentation-related
  • duplicate: same as an existing Issue

Priority assessment

  • P0 - Critical: affects core functionality, production bug, security vulnerability
  • P1 - High: significant functional defect, affects many users
  • P2 - Medium: general improvement, workaround exists
  • P3 - Low: nice-to-have, cosmetic improvement

Status assessment

  • Whether there are sufficient reproduction steps (for bugs)
  • Whether more information is needed
  • Whether a PR is already addressing it

Step 3: Generate Triage Report

Output a structured report:

## Issue Triage Report
Repository: [owner/repo]
Date: [date]
Total open issues: [N]

---

### 🔴 P0 Critical (needs immediate attention)
| # | Title | Type | Note |
|---|-------|------|------|
| #N | [title] | bug | [one sentence explaining urgency] |

### 🟠 P1 High Priority
| # | Title | Type | Note |
|---|-------|------|------|

### 🟡 P2 Medium Priority
| # | Title | Type | Note |
|---|-------|------|------|

### 🟢 P3 Low Priority
| # | Title | Type | Note |
|---|-------|------|------|

---

### Needs More Information
- #N [title]: [what information is missing]

### Possible Duplicates
- #N may be a duplicate of #M: [explanation]

### Recommended for Closure
- #N [title]: [reason, e.g., long inactive / already resolved another way]

Step 4: Update Issue Labels (requires user confirmation)

Ask the user whether to automatically update labels and priorities.

After confirmation, use update_issue for Issues that need updating:

  • Add appropriate labels
  • Update priority
  • Assign to a team member (if specified by the user)

For Issues that need more information, use comment_issue to ask:

Thanks for submitting this issue!

To help us address it more effectively, could you please provide:
- [Missing reproduction steps]
- [Environment and version info]
- [Expected behavior vs. actual behavior]

Thank you!

Notes

  • Triage focuses on classification and organization — no need to deep-dive into technical details of each Issue
  • Always confirm with the user before performing bulk updates to avoid unintended changes
  • If there are more than 50 Issues, consider processing in batches or filtering by a specific label