Skill 详情
tools-repo-review
Explicitly produces engineering-manager repository, team-health, and contribution insights.
使用前先检查
自动化审核只检查相关性,不代表安全审查或推荐。使用前请阅读来源中的说明。
SKILL.md
这段内容是审核时保存的快照。外部来源才是完整且最新的版本。
---
name: tools-repo-review
description: Comprehensive GitHub repository analysis for engineering managers with contribution stats, code quality review, team health metrics, and actionable management outputs
---
# Repo Review Skill (Engineering Manager Edition)
This skill enables AI-powered repository analysis with three modes: contribution statistics, code quality assessment, and team/manager insights. Supports org-wide reviews, single repos, and saved repo groups.
## Prerequisites
- Git CLI installed
- GitHub CLI (`gh`) for org-wide reviews
- Optional: Python 3.10+ with matplotlib for PNG chart generation
## Quick Start
```
Review https://github.com/org/repo
Review the UP organization
Review my side-projects group
```
## Progressive Onboarding
On first use, gather context through conversation:
### Initial Questions
1. **Role**: Engineering Manager | Tech Lead | IC
2. **Focus**: Full review | Quick stats | Code quality only
3. **Time period**: 30 days | 90 days | YTD | Custom
Save preferences to `config.yaml` for future sessions.
### Follow-up Questions (as needed)
- "I found 5 contributors. Focus on anyone specific?"
- "Compare against previous review?"
- "Flag contributors with < X commits?"
---
## Analysis Modes
### Mode 1: Contribution Statistics
Gather quantitative metrics using git commands:
```bash
# Clone repo to temp directory
git clone --quiet <repo_url> /tmp/repo_review_temp
# Contribution stats by author
git log --numstat --format="COMMIT|%H|%an|%ad" --date=short --since="2025-01-01" --all
# Weekly breakdown
git log --format="%ad" --date=short --since="2025-01-01" | sort | uniq -c
# Files touched by author
git log --author="Name" --name-only --pretty=format: | sort | uniq -c | sort -rn
# File type breakdown
git log --numstat --format="" | awk '{print $3}' | grep -oE '\.[^.]+$' | sort | uniq -c | sort -rn
```
**Output metrics:**
- Commits per author
- Lines added/deleted/net per author
- File type breakdown (frontend/backend/config/infra/docs)
- Weekly/monthly activity patterns
### Mode 2: Code Quality Review
Examine the codebase for:
**Architecture & Patterns**
- Project structure organization
- Component/module boundaries
- DRY compliance
- Separation of concerns
**Type Safety**
- TypeScript strictness
- `any` usage
- Proper type exports
**Error Handling**
- Try/catch patterns
- Error boundaries
- Validation at boundaries
**Database/API Patterns**
- Query structure
- N+1 detection
- Input validation
**Security**
- Auth implementation
- Secret handling
- Input sanitization
**Testing Indicators**
- Test file presence
- Coverage indicators
- Test patterns
### Mode 3: Team/Manager Assessment
Combine quantitative data with qualitative code review:
**Per-Contributor Analysis:**
- Volume and consistency
- Scope of work (full-stack vs narrow)
- Quality signals
- Growth indicators
**Team Health:**
- Bus factor
- Knowledge silos
- Cross-functional work
- Velocity trends
---
## Review Targets
### 1. Single Repository
```
Review https://github.com/org/repo
Review https://github.com/org/repo --since 2025-06-01
```
### 2. Organization-Wide
```
Review the UP organization
Review all repos in UP org
```
Uses GitHub CLI:
```bash
gh repo list UP --limit 100 --json name,sshUrl,url
```
### 3. Saved Groups
```
Review my side-projects group
Review client-work group
```
### 4. Managing Groups
```
Add https://github.com/user/repo to side-projects
Create a new group called "experiments"
Remove old-project from experiments
Show all my saved groups
```
---
## Visualization Options
### Mermaid Diagrams (for markdown reports)
**Contribution Timeline:**
```mermaid
gantt
title Contribution Timeline
dateFormat YYYY-MM-DD
section Contributor1
Feature A :2025-12-30, 2d
Feature B :2026-01-02, 1d
section Contributor2
Bug Fix :2026-01-01, 1d
```
**Work Breakdown:**
```mermaid
pie title Work Breakdown
"Frontend" : 45
"Backend" : 30
"Config" : 15
在 GitHub 阅读完整来源 (打开外部页面)