Token导航 LogoToken导航

Agent Skill · 开发

release-please-protection(释放请保护)

release-please-protection 用于处理 GitHub 仓库、Issue、Pull Request 和代码协作信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要围绕仓库状态、代码变更或协作事项进行整理时使用。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。

release-please-protection 是开发类 Agent Skill,可用宿主为 Codex、Claude、Cursor、Gemini CLI,通过GitHub 仓库获取,权限为接触敏感数据;许可证信息还不明确,第三方扫描已通过,按安装量在开发类的 21,746 个 Skill 里排第 12,657。

总安装 1,176周安装 50GitHub Stars 28下载 412核验 2026-05-01

安装说明

本站只整理中文说明,不托管安装包,也不代用户安装。
GitHub
许可证
unknown

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:release-please-protection(释放请保护)
来源仓库:https://github.com/laurigates/claude-plugins
仓库路径:skills/release-please-protection
安装命令:
npx skills add https://github.com/laurigates/claude-plugins --skill release-please-protection
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。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

skills.shnpx skills
npx skills add https://github.com/laurigates/claude-plugins --skill release-please-protection

这个 Skill 是做什么的

开发类 Skill 面向写代码、改代码和排查问题,一般会读写仓库文件,部分还会执行构建或测试命令。

release-please-protection 用于处理 GitHub 仓库、Issue、Pull Request 和代码协作信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要围绕仓库状态、代码变更或协作事项进行整理时使用。

它可设置保护规则,防止未经授权的合并或强制推送,确保发布流程的安全性。通过 npx skills add 命令从指定仓库安装,具体用法请结合原始 README 进一步确认。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操作。

适用宿主包括 Codex、Claude、Cursor、Gemini CLI,接入前应确认版本、权限和运行环境要求。

来源仓库内路径:skills/release-please-protection

上手步骤

4 步
  1. 1

    确认自己的宿主工具

    公开资料标注支持 Codex、Claude、Cursor、Gemini CLI;先确认自己用的是其中之一,再按对应工具的 Skills 目录放置。

  2. 2

    拿到安装命令

    可以直接复制的命令是 npx skills add https://github.com/laurigates/claude-plugins --skill release-plea…;命令里的占位符换成自己的路径或密钥再执行。

  3. 3

    安装后重启宿主

    命令执行完要完全退出宿主工具再打开,让新装的 Skill 被重新扫描到,否则列表里可能看不到。

  4. 4

    按最小权限试用

    它需要接触敏感数据,先在测试目录或测试账号里跑,确认没问题再放到正式项目。

常见问题

6 条

release-please-protection 怎么安装?

在终端执行 npx skills add https://github.com/laurigates/claude-plugins --skill release-please-protect…。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

release-please-protection 支持哪些 AI 工具?

公开资料标注支持 Codex、Claude、Cursor、Gemini CLI。其他工具只要能读取本地 Skill 目录,一般也能手动放进去,但要自己确认触发方式是否一致。

release-please-protection 需要什么权限?

权限级别是接触敏感数据。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。

release-please-protection 是免费的吗?

公开资料没有标注明确许可证,收费和商用限制都无法确认。正式使用前请到来源仓库核对授权说明。

release-please-protection 用的人多吗?

公开数据是累计安装 1,176 次,近一周安装 50 次,下载 412 次,GitHub 28 星。这些数字来自第三方目录统计,只能当作热度参考,不代表质量或安全性。

release-please-protection 安全吗?

第三方扫描结果为通过,但这只覆盖已知的静态检查项,不能替代自己读一遍源码和安装命令。

值得选的地方和要先想清楚的地方

值得选的地方

  • 同时适配 4 种工具,换客户端不用重新找替代品。
  • 第三方扫描已通过,常见的高风险行为没有命中。

先想清楚的地方

  • 需要接触敏感数据,装到常用环境前要先评估影响面。
  • 许可证没有明确标注,能不能商用、能不能再分发都不确定。
  • 第三方目录整理,版本更新和行为细节以来源仓库为准。

适不适合你

更适合

  • 已经在用 Codex,想直接扩展它能力的人。
  • 需要处理开发类任务、又不想每次都手写提示词的人。

不太适合

  • 环境里有大量密钥、凭证,不方便给第三方 Skill 读取的人。
  • 需要明确授权才能用到公司项目里的场景。
  • 需要厂商级技术支持和 SLA 的生产系统。

开发类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
skill-vetter

技能审查员 · OpenClaw

8,779,018986操作浏览器MIT-0通过
ontology

本体论 · OpenClaw

6,636,960576只读MIT-0通过
entra-app-registration

内部应用程序注册 · Codex、Claude

6,631,872764接触敏感数据未明确通过
entra-app-registration

内部应用程序注册 · Codex、Claude

3,757,680201接触敏感数据未明确通过
nano-banana-pro

纳米香蕉专业版 · OpenClaw

3,526,898364接触敏感数据MIT-0有提醒
release-please-protection

释放请保护 · 本页

1,17628接触敏感数据未明确通过

按安装量取同分类前几名;数字来自第三方目录统计,只作热度参考,不代表质量结论。

SKILL.md 原文

Release-Please Protection

Automatically detects and prevents manual edits to release-please managed files across all projects.

Overview

This skill provides proactive detection and warnings for files managed by Google's release-please automation tool. It helps prevent merge conflicts and workflow disruptions by identifying problematic edit attempts before they occur.

When This Skill Activates

The skill automatically activates in these scenarios:

  1. Direct edit requests to protected files
  2. User mentions of version bumps, releases, or changelog updates
  3. Broad refactoring that might touch version-controlled files
  4. Documentation updates that could include CHANGELOG.md
  5. "Fix all issues" or similar sweeping requests

Protected Files

Hard Protection (Permission System)

These files are completely blocked from editing by Claude Code's permission system:

  • **/CHANGELOG.md - All changelog files in any location

Operations blocked: Edit, Write, MultiEdit Operations allowed: Read (for analysis and context)

Soft Protection (Skill Detection)

These files trigger warnings and suggestions before edits:

Package Manager Manifests (Version Fields)
  • package.json → "version": "x.y.z" (npm/Node.js)
  • pyproject.toml → version = "x.y.z" (Python/uv)
  • Cargo.toml → version = "x.y.z" (Rust/cargo)
  • .claude-plugin/plugin.json → "version": "x.y.z" (Claude Code plugins)
  • pom.xml → <version>x.y.z</version> (Maven/Java)
  • build.gradle → version = 'x.y.z' (Gradle)
  • pubspec.yaml → version: x.y.z (Dart/Flutter)

Why soft protection? Claude Code's permission system operates at the file level, not field level. Blocking entire manifest files would prevent legitimate dependency updates via automated tools (npm, cargo, uv, etc.).

Detection Logic

Before attempting any edit, the skill checks:

1. File Path Analysis

if file_path ends with "CHANGELOG.md":
    → Inform user of hard permission block
    → Explain release-please workflow
    → Suggest conventional commit approach

2. Content Pattern Matching

if file is package manifest AND edit touches version field:
    → Warn about release-please management
    → Explain why manual edits cause conflicts
    → Offer to edit OTHER fields (but not version)
    → Provide conventional commit template

3. Intent Recognition

if user request contains keywords: "version", "release", "bump", "changelog":
    → Proactively explain release-please workflow
    → Check if files in scope are protected
    → Suggest proper approach before attempting edits

Response Templates

When CHANGELOG.md Edit Attempted

⚠️ CHANGELOG.md Protection Active

I cannot edit CHANGELOG.md files - they're protected by Claude Code's permission
system because they're managed by release-please automation.

**Why this matters:**
Manual edits to CHANGELOG.md cause merge conflicts with automated release PRs,
breaking your release workflow.

**Proper workflow:**
1. Make your changes with conventional commit messages:
   - feat: adds new features (minor version bump)
   - fix: bug fixes (patch version bump)
   - chore: maintenance tasks (no version bump)
   - BREAKING CHANGE: breaking changes (major version bump)

2. Release-please automatically:
   - Detects conventional commits
   - Generates CHANGELOG entries
   - Creates a release PR with version bumps
   - Tags releases when PR is merged

**Example commit:**

feat(auth): add OAuth2 support

Implements OAuth2 authentication flow with PKCE. Includes refresh token rotation and session management.

**Need to override?**
For emergency fixes, temporarily edit ~/.claude/settings.json:
1. Comment out the CHANGELOG.md deny rules
2. Make your edits
3. Re-enable protection

When Version Field Edit Detected

⚠️ Version Field Protection Warning

I notice you're asking me to modify the version field in [file_name]. This field
is typically managed by release-please automation.

**Current version:** [current_version]
**Requested change:** [requested_change]

**Why manual edits are problematic:**
- Creates merge conflicts with release-please PRs
- Can cause version numbering inconsistencies
- Bypasses conventional commit history tracking

**I can help in two ways:**

1. **Automated approach (recommended):**
   I'll help you craft a conventional commit message that will trigger the
   appropriate version bump when release-please creates its next release PR.

   For [requested_change], you should use:
   - `feat:` for minor version bump
   - `fix:` for patch version bump
   - `feat!:` or `BREAKING CHANGE:` for major version bump

2. **Manual override (emergency only):**
   If you absolutely need to manually set the version (e.g., aligning with
   external requirements), I can edit other fields in [file_name] but will
   skip the version field. You'll need to edit that manually or temporarily
   disable the skill protection.

Which approach would you prefer?

When Broad Refactoring Requested

📋 Release-Please Files Detected

I've identified several release-please managed files in the scope of this
refactoring:

**Protected files found:**
- plugins/dotfiles-core/CHANGELOG.md (hard block)
- plugins/dotfiles-toolkit/.claude-plugin/plugin.json (version field)
- [other files...]

**My approach:**
✅ I'll refactor all other files as requested
⚠️ I'll skip protected files and explain why
📝 I'll provide a summary of skipped changes

**If you need version/changelog updates:**
I'll generate appropriate conventional commit messages that will trigger
release-please to make those changes automatically.

Should I proceed with this approach?

Conventional Commit Guide

The skill provides instant conventional commit templates based on the type of change:

Feature Addition

feat(scope): brief description

Detailed explanation of what was added and why.
Can be multiple paragraphs.

Refs: #issue-number

Bug Fix

fix(scope): brief description

Explanation of the bug and how it was fixed.

Fixes: #issue-number

Breaking Change

feat(scope)!: brief description

BREAKING CHANGE: Explanation of what breaks and migration path.

Details about the new behavior.

Refs: #issue-number

Chore (No Version Bump)

chore(scope): brief description

Maintenance work that doesn't affect functionality.
Examples: dependency updates, refactoring, docs.

Integration with Other Skills

This skill works alongside:

  • Chezmoi Expert - Ensures dotfiles templates don't manually edit versions
  • Git Workflow - Enforces conventional commits before creating PRs
  • GitHub Actions - Aware of release-please workflow configurations

Skill Configuration

Located in dot_claude/skills/release-please-protection/ (source) which becomes ~/.claude/skills/release-please-protection/ after chezmoi apply:

  • SKILL.md - This file (skill definition)
  • patterns.md - Protected file pattern reference
  • workflow.md - Detailed release-please workflow guide

Limitations

What This Skill Cannot Prevent

  1. Explicit overrides - If you explicitly instruct me to edit despite warnings
  2. Out-of-context files - Files not in the current context window
  3. External tools - Commands like sed, awk, or direct bash edits
  4. Git operations - Manual git commit with modified protected files

What This Skill DOES Prevent

  1. Accidental edits - Catching mistakes before they happen
  2. Workflow violations - Explaining proper release-please patterns
  3. Merge conflicts - Preventing automated PR conflicts
  4. Version inconsistencies - Maintaining semantic versioning discipline

Emergency Overrides

If you absolutely must manually edit protected files:

Temporary Permission Override

# 1. Edit global settings
vim ~/.claude/settings.json

# 2. Comment out deny rules
"deny": [
  "Bash(git add .)",
  "Bash(git add -A)",
  "Bash(git add --all)",
  // "Edit(**/CHANGELOG.md)",
  // "Write(**/CHANGELOG.md)",
  // "MultiEdit(**/CHANGELOG.md)"
]

# 3. Make your edits

# 4. Re-enable protection (uncomment the lines)

# 5. Verify with chezmoi
chezmoi diff ~/.claude/settings.json
chezmoi apply  # If template is out of sync

Skill Bypass (Not Recommended)

# Temporarily disable skill
mv .claude/skills/release-please-protection .claude/skills/release-please-protection.disabled

# Make edits

# Re-enable
mv .claude/skills/release-please-protection.disabled .claude/skills/release-please-protection

Success Metrics

This skill is working properly when:

✅ All CHANGELOG.md edit attempts are blocked with helpful explanations ✅ Version field modifications trigger warnings and alternatives ✅ Conventional commit suggestions match the requested changes ✅ Users understand the release-please workflow after first warning ✅ No merge conflicts occur with automated release PRs ✅ Version numbers follow semantic versioning consistently

Further Reading

平台分布

Codex

32.42%
按下载量换算134

Claude

31.99%
按下载量换算132

Cursor

19.44%
按下载量换算80

Gemini CLI

9.47%
按下载量换算39

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

接触敏感数据

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。目前只有一个公开来源,建议同时核对源仓库。

继续浏览同类 Skills