Repo Scout — Repository Discovery & Ranking
Overview
Systematically discover and rank GitHub repositories in a given ecosystem. Produces a structured, actionable ranking document.
Use cases: Open-source contribution targeting, technology landscape surveys, competitive analysis, ecosystem exploration.
Prerequisites
Before starting, the user must have GitHub CLI authenticated:
CODEBLOCK0
If not configured, ask the user to provide:
- 1. GitHub username — for searching and attribution
- GitHub token — run
gh auth login or set INLINECODE1
Without auth, gh API calls will hit rate limits quickly and private repo data won't be accessible.
Workflow
Step 1: Define Scope
Ask the user for (with sensible defaults):
| Parameter | Default | Example |
|---|
| Ecosystem keyword(s) | (required) | "AI agent", "LLM tools", "Kubernetes" |
| Target count |
15 | top 15 by stars |
| Minimum stars | 5,000 | Filter out small repos |
| Language filter |
(any) | Python, TypeScript |
| Additional criteria |
(none) | "must have bug label issues" |
Step 2: Search & Collect
Use multiple search strategies to find candidates:
CODEBLOCK1
For each candidate repository, collect:
| Data Point | How to Get |
|---|
| Star count | GitHub API / web |
| Primary language |
GitHub API |
| Last commit date | GitHub API |
| Open issue count | GitHub API |
| Bug-labeled issues |
gh issue list --label bug --state open --limit 5 |
|
good first issue count | GitHub search |
| CONTRIBUTING.md exists? | Check repo root |
| CI configured? | Check
.github/workflows/ |
| PR template exists? | Check
.github/PULL_REQUEST_TEMPLATE.md |
| License | GitHub API |
Step 3: Score & Rank
Score each repository on a contribution friendliness scale:
| Factor | Weight | Scoring |
|---|
| Actionable bug issues | 30% | 3=many clear bugs, 1=none |
| Activity (recent commits) |
20% | 3=daily, 2=weekly, 1=monthly+ |
| Contribution docs | 15% | 3=CONTRIBUTING+template, 2=partial, 1=none |
| CI/CD health | 15% | 3=green CI, 2=partial, 1=none |
| Community size (stars) | 10% | 3=>50K, 2=>10K, 1=>5K |
| Response time to PRs | 10% | 3=<3d, 2=<7d, 1=>7d |
Step 4: Filter Out
Mark repositories to skip if:
- - Non-code repo (awesome-lists, documentation-only, resource collections)
- Desktop/mobile UI bugs requiring hardware access
- No actionable bug issues (only feature requests or stale issues)
- Archived or unmaintained (no commits in 6+ months)
- Hostile contribution environment (PRs routinely ignored)
Step 5: Produce Ranking Document
Write {workspace}/ecosystem-top{N}.md:
CODEBLOCK2
Output
- -
{workspace}/ecosystem-top{N}.md — Structured ranking document ready for downstream use
Tips
- - When used as part of a contribution campaign, the output feeds directly into the issue-hunter skill for issue analysis.
- For technology evaluation, the ranking + detailed profiles are the final deliverable.
- Re-run periodically to catch ecosystem changes.
Repo Scout — 仓库发现与排名
概述
系统性地发现并排名特定生态系统中的GitHub仓库,生成结构化、可操作的排名文档。
应用场景:开源贡献目标筛选、技术生态调研、竞品分析、生态探索。
前置条件
开始前,用户必须完成GitHub CLI身份认证:
bash
gh auth status # 必须显示Logged in
若未配置,请用户提供:
- 1. GitHub用户名 — 用于搜索和归属
- GitHub令牌 — 运行gh auth login或设置export GH_TOKEN=
未认证时,gh API调用将快速触发频率限制,且无法访问私有仓库数据。
工作流程
步骤1:定义范围
向用户询问以下参数(含合理默认值):
| 参数 | 默认值 | 示例 |
|---|
| 生态关键词 | (必填) | AI agent、LLM tools、Kubernetes |
| 目标数量 |
15 | 按星标数排名前15 |
| 最低星标数 | 5,000 | 过滤小型仓库 |
| 语言筛选 |
(不限) | Python、TypeScript |
| 附加条件 |
(无) | 必须包含bug标签的议题 |
步骤2:搜索与收集
使用多种搜索策略寻找候选仓库:
搜索策略:
- 1. GitHub搜索:{关键词}按星标排序
- awesome-{关键词}精选列表
- 该领域GitHub趋势
- 网页搜索:{关键词} 顶级开源项目 {年份}
对每个候选仓库,收集以下数据:
| 数据点 | 获取方式 |
|---|
| 星标数 | GitHub API / 网页 |
| 主要编程语言 |
GitHub API |
| 最后提交日期 | GitHub API |
| 开放议题数 | GitHub API |
| 标记为bug的议题 | gh issue list --label bug --state open --limit 5 |
| good first issue数量 | GitHub搜索 |
| 是否存在CONTRIBUTING.md? | 检查仓库根目录 |
| 是否配置CI? | 检查.github/workflows/ |
| 是否存在PR模板? | 检查.github/PULL
REQUESTTEMPLATE.md |
| 许可证 | GitHub API |
步骤3:评分与排名
按贡献友好度对每个仓库评分:
| 因素 | 权重 | 评分标准 |
|---|
| 可操作的bug议题 | 30% | 3=大量明确bug,1=无 |
| 活跃度(近期提交) |
20% | 3=每日,2=每周,1=每月以上 |
| 贡献文档 | 15% | 3=CONTRIBUTING+模板,2=部分,1=无 |
| CI/CD健康度 | 15% | 3=绿色CI,2=部分,1=无 |
| 社区规模(星标数) | 10% | 3=>50K,2=>10K,1=>5K |
| PR响应时间 | 10% | 3=<3天,2=<7天,1=>7天 |
步骤4:过滤排除
标记以下情况的仓库为跳过:
- - 非代码仓库(awesome列表、纯文档、资源合集)
- 需要硬件访问的桌面/移动端UI缺陷
- 无可操作的bug议题(仅功能请求或过期议题)
- 已归档或无人维护(6个月以上无提交)
- 不友好的贡献环境(PR经常被忽略)
步骤5:生成排名文档
写入{workspace}/ecosystem-top{N}.md:
markdown
{生态系统} — 前{N}个仓库
生成日期:{date}
关键词:{keywords}
最低星标数:{min_stars}
排名
| 排名 | 仓库 | 星标数 | 语言 | 开放bug数 | 评分 | 备注 |
|---|
| 1 | owner/repo | 45.2K | Python | 12 | 8.5/10 | 活跃,文档完善 |
| 2 |
... | ... | ... | ... | ... | ... |
跳过的仓库
| 仓库 | 原因 |
|---|
| owner/repo | 非代码(awesome列表) |
详细档案
1. owner/repo (45.2K ⭐)
- - 语言:Python
- 最后提交:2天前
- 开放议题:234(12个标记为bug)
- CONTRIBUTING.md:✅
- CI:✅ GitHub Actions
- 评分明细:活跃度 3/3,bug 3/3,文档 2/3,CI 3/3,社区 2/3,响应 2/3
- 备注:非常活跃,社区友好
输出
- - {workspace}/ecosystem-top{N}.md — 结构化排名文档,可直接用于后续流程
提示
- - 作为贡献活动的一部分时,输出结果可直接输入issue-hunter技能进行议题分析。
- 用于技术评估时,排名加详细档案即为最终交付物。
- 定期重新运行以捕捉生态变化。