Token导航 LogoToken导航

Agent Skill · 开发

checkpoint(checkpoint 搜索)

checkpoint 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。

checkpoint 是开发类 Agent Skill,可用宿主为 Codex、Claude、Cursor、Gemini CLI,通过GitHub 仓库获取,权限为权限未标注;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 15,714。

总安装 569周安装 23GitHub Stars 4下载 178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:checkpoint(checkpoint 搜索)
来源仓库:https://github.com/petekp/agent-skills
仓库路径:skills/checkpoint
安装命令:
npx skills add https://github.com/petekp/agent-skills --skill checkpoint
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/petekp/agent-skills --skill checkpoint

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。

用于处理 GitHub 仓库、Issue、Pull Request 和代码协作信息。适合围绕仓库状态、代码变更或协作事项进行整理。可结合来源仓库和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态及是否会触发联网或命令执行。当前归类为开发,功能描述一致,使用方式明确。

checkpoint 属于开发类 Skill,可作为该场景下的辅助能力补充。

来源仓库内路径:skills/checkpoint

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/petekp/agent-skills --skill checkpoint;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

    权限没有明确标注,第一次使用时只给一个无关紧要的任务,观察它到底读了哪些文件和网络。

常见问题

6 条

checkpoint 怎么安装?

在终端执行 npx skills add https://github.com/petekp/agent-skills --skill checkpoint。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

checkpoint 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。

checkpoint 需要什么权限?

权限级别没有明确标注。第一次使用时建议只给无关紧要的任务,观察它实际读写了哪些内容。

checkpoint 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。

checkpoint 用的人多吗?

公开数据是累计安装 569 次,近一周安装 23 次,下载 178 次,GitHub 4 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。

checkpoint 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

  • 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

  • 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
checkpoint

checkpoint 搜索 · 本页

5694权限未标注未明确通过

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

Checkpoint

Pause, assess, and surface intelligent next-step options to the user.

When to Trigger Proactively

Suggest this skill (without being asked) when detecting:

  • Completion signal: A feature, fix, or milestone just finished
  • Uncertainty signal: Requirements unclear, multiple valid paths, or low confidence in current direction
  • Complexity signal: Scope expanded, unexpected dependencies emerged, or task is taking longer than expected
  • Drift signal: Work may have diverged from user's original intent
  • Quality signal: Code works but may benefit from review, testing, or refactoring

Workflow

1. Context Assessment

Silently evaluate the current state across these dimensions (do not output this analysis):

  • Progress: What has been accomplished? What remains?
  • Quality: Is the work solid, or are there rough edges?
  • Alignment: Does recent work match what the user actually wants?
  • Uncertainty: What assumptions were made? What's unclear?
  • Risk: What could go wrong? What hasn't been tested?
  • Efficiency: Is there a better path forward?

2. Generate Options

Based on assessment, generate 2-5 contextually-appropriate options. Draw from (but don't limit to) these archetypes:

ArchetypeWhen Relevant
Commit progressMeaningful progress made, good stopping point
Systems auditComplex changes, potential for bugs or regressions
Prioritize/planMultiple pending tasks, unclear what matters most
Re-evaluate decisionsLow confidence in recent choices, new information available
Clarify with userAssumptions made, requirements ambiguous
Test/verifyCode works but edge cases untested
Refactor/clean upCode functional but messy
DocumentComplex logic that needs explanation
Step backMay be overcomplicating or missing simpler solution
Continue current pathClear next step, no reason to pause

Option generation principles:

  • Options should be meaningfully different, not variations of the same thing
  • Include at least one "continue forward" option when momentum is valuable
  • Include at least one "pause and verify" option when risk is present
  • Avoid analysis paralysis - fewer sharp options beat many vague ones

3. Select Recommendation

Choose one option as recommended. The recommendation should reflect:

  • What would a thoughtful senior engineer do here?
  • What reduces risk of wasted effort or rework?
  • What serves the user's underlying goals (not just stated requests)?

4. Present via AskUserQuestion

Use AskUserQuestion with this structure:

Question: "What would you like to do next?"
Header: "Next step" (or contextually appropriate 1-2 words)
Options: [generated options with descriptions]

Option format:

  • label: Action verb phrase (e.g., "Commit current progress", "Run systems audit")
  • description: 1 sentence explaining what this involves and why it might be valuable

Recommendation:

  • Place recommended option FIRST in the list
  • Append "(Recommended)" to its label
  • Include rationale in the description

Example Output

For a scenario where a feature was just implemented but with some shortcuts:

AskUserQuestion:
  question: "Feature implementation complete. What would you like to do next?"
  header: "Next step"
  options:
    - label: "Review and refactor (Recommended)"
      description: "Clean up the shortcuts taken during implementation before they become technical debt. The core logic works but could be more maintainable."
    - label: "Add test coverage"
      description: "Write tests for the new feature to catch edge cases and prevent regressions."
    - label: "Commit and move on"
      description: "The feature works - commit it and tackle the next task. Can refactor later if needed."
    - label: "Walk me through what was built"
      description: "Explain the implementation so you can verify it matches your expectations before proceeding."

Anti-Patterns

  • Don't overthink: This skill should take seconds, not minutes
  • Don't list every possible option: Curate the most valuable 2-5
  • Don't recommend "ask user" when the situation is clear: Have a point of view
  • Don't trigger too frequently: Reserve for genuine decision points, not every minor step
  • Don't explain the assessment process: Just present the options naturally

平台分布

Codex

36.62%
按下载量换算65

Claude

30%
按下载量换算53

Cursor

17.83%
按下载量换算32

Gemini CLI

7.83%
按下载量换算14

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

权限未标注

当前来源未能明确判断权限范围,默认进入异常复核队列。

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