Token导航 LogoToken导航TokenDH.com
研究检索敏感数据github未标认证来源可访问clear审计提醒

open-source-contributions开源贡献

Agent Skill

open-source-contributions 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

总安装

7,362

周安装

313

GitHub Stars

750

下载量

2,579
CodexClaudeCursorGemini CLI

安装说明

本站只整理中文说明和来源信息,不托管安装包,也不代用户安装。

GitHub

来源数

3

许可证

MIT

最后核验

2026-05-01

来源状态

来源可访问

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:open-source-contributions(开源贡献)
来源仓库:https://github.com/jezweb/claude-skills
仓库路径:skills/open-source-contributions
安装命令:
npx skills add https://github.com/jezweb/claude-skills --skill open-source-contributions
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。不同来源提供的安装方式可能略有差异;本站展示可直接复制的安装命令,安装前请核对来源页面。

skills.shnpx skills
npx skills add https://github.com/jezweb/claude-skills --skill open-source-contributions

简介

创建维护人员友好的拉取请求,同时避免 16 个常见的拒绝错误。

  • 防止使用预 PR 检查脚本提交个人开发工件(SESSION.md、规划文档、调试屏幕截图、临时测试文件)
  • 强制实施三项关键工作流程规则:始终使用功能分支、提交前彻底测试以及将 PR 集中于 400 行以下的单个更改
  • 使用“内容”/“原因”/“如何”结构、传统提交消息以及带有自动关闭关键字的问题链接,提供 PR 描述的模板和指南
  • 包括捆绑的脚本、清单和示例,用于在提交之前扫描工件、清理分支并遵循项目约定

SKILL.md

Open Source Contributions Skill

Version: 1.2.0 | Last Verified: 2026-01-09 | Production Tested: ✅


When to Use This Skill

Auto-triggers: "submit PR to", "contribute to", "pull request for", "open source contribution"

Create maintainer-friendly PRs while avoiding the 16 common mistakes that cause rejection.


What NOT to Include in Pull Requests

Personal Development Artifacts (NEVER Include)

Planning & Notes Documents:

❌ SESSION.md              # Session tracking notes
❌ NOTES.md                # Personal development notes
❌ TODO.md                 # Personal todo lists
❌ planning/*              # Planning documents directory
❌ IMPLEMENTATION_PHASES.md # Project planning
❌ DATABASE_SCHEMA.md      # Unless adding new schema to project
❌ ARCHITECTURE.md         # Unless documenting new architecture
❌ SCRATCH.md              # Temporary notes
❌ DEBUGGING.md            # Debugging notes
❌ research-logs/*         # Research notes

Screenshots & Visual Assets:

❌ screenshots/debug-*.png     # Debugging screenshots
❌ screenshots/test-*.png      # Testing screenshots
❌ screenshot-*.png            # Ad-hoc screenshots
❌ screen-recording-*.mp4      # Screen recordings
❌ before-after-local.png      # Local comparison images

✅ screenshots/feature-demo.png   # IF demonstrating feature in PR description
✅ docs/assets/ui-example.png     # IF part of documentation update

Test Files (Situational):

❌ test-manual.js          # Manual testing scripts
❌ test-debug.ts           # Debugging test files
❌ quick-test.py           # Quick validation scripts
❌ scratch-test.sh         # Temporary test scripts
❌ example-local.json      # Local test data

✅ tests/feature.test.js   # Proper test suite additions
✅ tests/fixtures/data.json # Required test fixtures
✅ __tests__/component.tsx  # Component tests

Build & Dependencies:

❌ node_modules/           # Dependencies (in .gitignore)
❌ dist/                   # Build output (in .gitignore)
❌ build/                  # Build artifacts (in .gitignore)
❌ .cache/                 # Cache files (in .gitignore)
❌ package-lock.json       # Unless explicitly required by project
❌ yarn.lock               # Unless explicitly required by project

IDE & OS Files:

❌ .vscode/                # VS Code settings
❌ .idea/                  # IntelliJ settings
❌ .DS_Store               # macOS file system
❌ Thumbs.db               # Windows thumbnails
❌ *.swp, *.swo            # Vim swap files
❌ *~                      # Editor backup files

Secrets & Sensitive Data:

❌ .env                    # Environment variables (NEVER!)
❌ .env.local              # Local environment config
❌ config/local.json       # Local configuration
❌ credentials.json        # Credentials (NEVER!)
❌ *.key, *.pem            # Private keys (NEVER!)
❌ secrets/*               # Secrets directory (NEVER!)

Temporary & Debug Files:

❌ temp/*                  # Temporary files
❌ tmp/*                   # Temporary directory
❌ debug.log               # Debug logs
❌ *.log                   # Log files
❌ dump.sql                # Database dumps
❌ core                    # Core dumps
❌ *.prof                  # Profiling output

What SHOULD Be Included

✅ Source code changes      # The actual feature/fix
✅ Tests for changes        # Required tests for new code
✅ Documentation updates    # README, API docs, inline comments
✅ Configuration changes    # If part of the feature
✅ Migration scripts        # If needed for the feature
✅ Package.json updates     # If adding/removing dependencies
✅ Schema changes           # If part of feature (with migrations)
✅ CI/CD updates            # If needed for new workflows

Pre-PR Cleanup Process

Step 1: Run Pre-PR Check Script

Use the bundled scripts/pre-pr-check.sh to scan for artifacts:

./scripts/pre-pr-check.sh

What it checks:

  • Personal documents (SESSION.md, planning/*, NOTES.md)
  • Screenshots not referenced in PR description
  • Temporary test files
  • Large files (>1MB)
  • Potential secrets in file content
  • PR size (warns if >400 lines)
  • Uncommitted changes

Step 2: Review Git Status

git status
git diff --stat

Ask yourself:

  • Is every file change necessary for THIS feature/fix?
  • Are there any unrelated changes?
  • Are there files I added during development but don't need?

Step 3: Clean Personal Artifacts

Manual removal:

git rm --cached SESSION.md
git rm --cached -r planning/
git rm --cached screenshots/debug-*.png
git rm --cached test-manual.js

Or use the clean script:

./scripts/clean-branch.sh

Step 4: Update.gitignore

Add personal patterns to .git/info/exclude (affects only YOUR checkout):

# Personal development artifacts
SESSION.md
NOTES.md
TODO.md
planning/
screenshots/debug-*.png
test-manual.*
scratch.*

Writing Effective PR Descriptions

Use the What/Why/How Structure

Template (see references/pr-template.md):

## What?
[Brief description of what this PR does]

## Why?
[Explain the reasoning, business value, or problem being solved]

## How?
[Describe the implementation approach and key decisions]

## Testing
[Step-by-step instructions for reviewers to test]

## Checklist
- [ ] Tests added/updated
- [ ] Documentation updated
- [ ] CI passing
- [ ] Breaking changes documented

## Related Issues
Closes #123
Relates to #456

Commit Message Format

Conventional Commits: <type>(<scope>): <subject>

Types: feat, fix, docs, refactor, test, ci, chore

Example: feat(auth): add OAuth2 support for Google and GitHub

See references/commit-message-guide.md for complete guide.


PR Sizing Best Practices

Research-backed guidelines:

  • Ideal: 50 lines
  • Good: <200 lines
  • Maximum: 400 lines
  • Beyond 400: Defect detection drops significantly

Keep PRs small:

  • One change per PR
  • Use feature flags for incomplete work: if (featureFlags.newAuth) {// New OAuth flow (incomplete but behind flag)} else {// Existing flow}
  • Break by layer: schema → API → frontend → tests

Following Project Conventions

Before contributing:

  1. Read CONTRIBUTING.md (check /, /.github/, /docs/)
  2. Run formatters: npm run lint, npm run format
  3. Match existing patterns (review recent merged PRs)
  4. Test before submitting: npm test && npm run lint && npm run build

Communication Best Practices

Response templates:

  • Implemented: "Good idea! Implemented in [commit hash]"
  • Disagreement: "I see your point. I went with X because Y. Open to alternatives."
  • Clarification: "Could you help me understand what you mean by Z?"
  • Ping (after 1-2 weeks): "Gently pinging this PR. Happy to make changes!"

Common Mistakes That Annoy Maintainers (16 Errors Prevented)

See Critical Workflow Rules section for detailed guidance on Rules 1-3

  1. Not Reading CONTRIBUTING.md - ALWAYS read first, follow exactly
  2. Including Personal Artifacts - SESSION.md, planning/*, screenshots, temp tests (use pre-PR check script)
  3. Massive Pull Requests - Break into <200 lines ideal, <400 max
  4. Not Testing Before Submitting - Run full test suite, test manually, capture evidence (violates RULE 2)
  5. Working on Assigned Issues - Check assignments, comment to claim work
  6. Not Discussing Large Changes First - Open issue or comment before coding
  7. Being Impatient/Unresponsive - Be responsive, ping after 1-2 weeks
  8. Not Updating Documentation - Update README, API docs, inline comments
  9. Ignoring Code Style - Use project's linters/formatters
  10. Ignoring CI Failures - Fix immediately, ask for help if stuck
  11. Including Unrelated Changes - One PR = One Feature (violates RULE 3)
  12. Not Linking Issues - Use "Closes #123" or "Fixes #456"
  13. Committing Secrets - Never commit.env, scan for secrets
  14. Force-Pushing Without Warning - Avoid after review starts
  15. Not Running Build/Tests Locally - Always run before pushing
  16. Working on main/master - ALWAYS use feature branches (violates RULE 1)

GitHub-Specific Best Practices

Critical Workflow Rules (NEVER SKIP)

RULE 1: ALWAYS Work on a Feature Branch

# ✅ CORRECT
git checkout main
git pull upstream main
git checkout -b feature/add-oauth-support
# make changes on feature branch
git commit -m "feat(auth): add OAuth support"

Branch naming: feature/name, fix/issue-123, docs/update-readme, refactor/utils, test/add-tests


RULE 2: Test Thoroughly BEFORE Submitting PR

Never submit without:

  1. Running full test suite: npm test && npm run lint && npm run build
  2. Testing manually (run app, test feature, edge cases)
  3. Capturing evidence (screenshots/videos for visual changes - add to PR description, NOT commits)
  4. Checking CI will pass

Testing checklist template:

## Testing Performed
### Automated Tests
- ✅ All existing tests pass
- ✅ Added 12 new tests for OAuth flow
- ✅ Coverage increased from 85% to 87%

### Manual Testing
- ✅ Tested Google/GitHub OAuth flows end-to-end
- ✅ Verified error handling
- ✅ Tested on Chrome, Firefox, Safari

RULE 3: Keep PRs Focused and Cohesive

One PR = One Feature/Fix

  • Ideal: <200 lines
  • Acceptable: 200-400 lines
  • Large: 400-800 lines (needs justification)
  • Too large: >800 lines (split it)

Keep focused:

  • Plan: What ONE thing does this PR do?
  • During dev: Unrelated bug? Separate branch
  • Before commit: git diff - Is every change necessary for THIS feature?

Break large features into phases:

PR #1: Database schema and models
PR #2: API endpoints
PR #3: Frontend components
PR #4: Integration and tests

Using Draft PRs

Create: gh pr create --draft Mark ready: gh pr ready (when code complete, tests passing, CI passing)

Linking Issues

Auto-closing keywords (in PR description):

Closes #123
Fixes #456
Resolves #789

# Multiple: Fixes #10, closes #20, resolves #30
# Cross-repo: Fixes owner/repo#123

GitHub CLI Essentials

gh pr create --fill                    # Auto-fill from commits
gh pr create --draft                   # Draft PR
gh pr status                           # See your PRs
gh pr checks                           # View CI status
gh pr ready                            # Mark draft as ready

Pre-Submission Checklist

See references/pr-checklist.md for complete version.

Pre-Contribution:

  • Read CONTRIBUTING.md, CODE_OF_CONDUCT.md
  • Commented on issue to claim work
  • Created feature branch (NEVER work on main)

Development:

  • RULE 1: Working on feature branch
  • RULE 2: Tested thoroughly with evidence
  • RULE 3: PR focused on single feature
  • All tests pass: npm test && npm run lint && npm run build
  • Updated documentation

Cleanup:

  • Ran ./scripts/pre-pr-check.sh
  • No personal artifacts (SESSION.md, planning/*, debug screenshots, temp tests)
  • No secrets (.env, credentials)

PR Quality:

  • Focused on one change (<200 lines ideal, <400 max)
  • Title: Conventional Commits format
  • Description: What/Why/How structure
  • Links to issues (Closes #123)
  • Screenshots for visual changes (in PR description)

Post-Submission:

  • Monitor CI, fix failures immediately
  • Respond to feedback promptly

Bundled Resources

See bundled examples and scripts:

  • scripts/pre-pr-check.sh - Scan for artifacts before submission
  • scripts/clean-branch.sh - Remove common personal artifacts
  • references/pr-template.md - PR description template
  • references/pr-checklist.md - Complete checklist
  • references/commit-message-guide.md - Conventional commits guide
  • assets/good-pr-example.md - Well-structured PR example
  • assets/bad-pr-example.md - Common mistakes to avoid

Key Takeaways

  1. RULE 1: ALWAYS use feature branches (never main)
  2. RULE 2: Test thoroughly before submitting (automated + manual + evidence)
  3. RULE 3: Keep PRs focused (<200 lines ideal, one change per PR)
  4. Clean PRs: Remove personal artifacts (SESSION.md, planning/*, debug screenshots)
  5. Read CONTRIBUTING.md: Always read first, follow exactly
  6. Link Issues: Use "Closes #123" to auto-close
  7. Use ./scripts/pre-pr-check.sh: Scan for artifacts before submission

Production Tested: Real-world open source contributions and maintainer feedback

Token Efficiency: ~70% savings vs trial-and-error

Errors Prevented: 16 common mistakes

Last Verified: 2026-01-09

适合场景

01

用户想查找某类 Agent Skill 时

02

需要根据任务场景推荐可安装能力包时

03

需要对比不同来源的安装命令和来源信息时

04

需要参考平台分布和安装热度时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明,方便继续核验

能力 4

补充不同宿主或平台的使用分布数据

能力 5

展示第三方安全扫描或审计结果

安装后应在对应宿主中按原始 README 的触发条件使用;具体调用方式请以来源页面和 README 为准。

平台分布

Claude Code

30.79%
按下载量换算794

Gemini CLI

21.25%
按下载量换算548

Antigravity

18.94%
按下载量换算488

Cursor

12.81%
按下载量换算330

OpenCode

7.58%
按下载量换算195

Codex

3.58%
按下载量换算92

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

可疑

权限和风险

敏感数据

该 Skill 可能接触密钥、Token、环境变量或敏感配置,应进入高风险复核队列,默认不自动发布。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。来源安全扫描存在 warning/failed 结果,不能写成本站确认安全。

来源信息

继续浏览同类 Skills