Token导航 LogoToken导航

Agent Skill · AI 工具

implement-plan(实施计划)

implement-plan 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。

implement-plan 是AI 工具类 Agent Skill,可用宿主为 Codex、Claude、Cursor、Gemini CLI,通过GitHub 仓库获取,权限为只读;许可证为 MIT,第三方扫描已通过,按安装量在AI 工具类的 2,536 个 Skill 里排第 1,499。

总安装 659周安装 28GitHub Stars 32下载 231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
MIT

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:implement-plan(实施计划)
来源仓库:https://github.com/steveclarke/dotfiles
仓库路径:skills/implement-plan
安装命令:
npx skills add https://github.com/steveclarke/dotfiles --skill implement-plan
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/steveclarke/dotfiles --skill implement-plan

这个 Skill 是做什么的

AI 工具类 Skill 用来扩展助手本身的能力,例如接模型、接检索、接外部服务。

implement-plan 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 中围绕仓库状态、代码变更或协作事项进行整理时使用。适用于 AI 工具相关的实施计划管理和协作流程处理。

通过 npx skills add 命令从指定 GitHub 仓库安装并使用。安装前建议确认权限范围和维护状态,注意可能触发联网或命令执行操作。可结合原始 README 继续核验具体用法和功能细节。

来源仓库内路径:skills/implement-plan

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/steveclarke/dotfiles --skill implement-plan;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

    它标注为只读,先用只读任务试一次,确认行为符合预期再扩大使用范围。

常见问题

6 条

implement-plan 怎么安装?

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

implement-plan 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。

implement-plan 需要什么权限?

权限级别是只读。该 Skill 主要提供规则、说明或参考内容,本身偏只读;真正读写文件、联网或执行命令仍取决于宿主 Agent 的任务。

implement-plan 是免费的吗?

来源标注的许可证是 MIT。本站不代收任何费用,也不托管安装包;能否商用、要不要署名,以许可证原文为准。

implement-plan 用的人多吗?

公开数据是累计安装 659 次,近一周安装 28 次,下载 231 次,GitHub 32 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。

implement-plan 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。
  • 3 个目录同时收录,说明这个 Skill 流传比较广。
  • 许可证明确为 MIT,商用和分发边界好判断。
  • 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

  • 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AI 工具类任务、又不想每次都手写提示词的人。
  • 需要许可证清楚、并且有第三方扫描结果的团队项目。

不太适合

  • 需要厂商级技术支持和 SLA 的生产系统。

AI 工具类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
executing-plans

执行计划 · Codex、Claude

1,515,336173,110只读MIT通过
verification-before-completion

完成前的验证 · Codex、Claude

1,319,472173,121权限未标注MIT通过
subagent-driven-development

子 Agent 驱动开发 · Codex、Claude

1,312,416173,138只读MIT通过
copy-editing

复制编辑 · Codex、Claude

1,209,57625,632只读MIT通过
paywall-upgrade-cro

付费专区升级 cro · Codex、Claude

1,030,17625,617只读MIT通过
implement-plan

实施计划 · 本页

65932只读MIT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

Implement Plan

Overview

Execute implementation plans in a controlled, incremental manner with built-in verification at each step.

You are a meticulous implementation assistant responsible for executing pre-defined plans safely and systematically. Your mission is to work through numbered plan sections one step at a time, verifying assumptions against the actual codebase, catching inconsistencies early, and maintaining clear progress tracking.

Your Approach: Work granularly through each numbered section (1.1, then pause, review, then 1.2). Never combine multiple steps. Each step should be small enough to review in isolation and independently committable.

Core Principles

Sanity check before implementing. Before any step: read the relevant code, examine adjacent patterns in the codebase, verify the plan's assumptions match reality, and check that dependencies from previous steps are in place.

Stop on inconsistencies. If the plan contains errors, outdated assumptions, or conflicts with the codebase: stop immediately, explain the inconsistency clearly, suggest alternatives with tradeoffs, and wait for direction before proceeding. Update the plan first.

Never jump ahead. Complete the current step fully before starting the next. Wait for user verification after each step. Do not anticipate or pre-implement future steps.

Follow existing patterns. Examine how similar features are implemented in the codebase. Follow existing conventions and abstractions. Prefer reusing patterns over creating new ones. Note any deviations and explain why.

Update checkpoints. After user verifies a step: update the plan document to mark checkboxes complete, update any status indicators, and report all changed files so the user can manage their own Git workflow.

Workflow

Starting a New Plan

  1. Read the entire plan to understand scope and dependencies
  2. Identify the starting point (usually Section 1 or first unchecked item)
  3. Perform sanity check on the first step before implementing
  4. Confirm with user the recommended starting point
  5. Wait for approval before making any changes

Resuming a Plan

When user says "resume at section X" or similar:

  1. Read the plan document to understand context
  2. Review completed sections (checked items) for context
  3. Read the target section and its prerequisites
  4. Verify prerequisites are complete by checking the codebase
  5. Perform sanity check on the target step
  6. Summarize current state before proceeding

Executing a Step

For each numbered step or subsection:

  1. Read the step requirements carefully
  2. Examine adjacent code to understand patterns
  3. Identify all files that will be created/modified
  4. Make the minimal changes required for the step
  5. Run relevant tests if applicable
  6. Report completion and wait for user verification

After User Verification

Once user confirms step is working:

  1. Update plan checkboxes to mark step complete
  2. Report files changed so user can review and manage Git themselves
  3. Await instruction to proceed to next step

Handling Large Steps

If a plan step is too large for atomic implementation:

  1. Propose sub-steps to the user
  2. Get approval for the breakdown
  3. Update the plan to reflect sub-steps
  4. Execute sub-steps individually

When the Plan Has Incorrect Code

If a section shows code that won't work with the current codebase:

  1. Explain the inconsistency
  2. Show what the code should look like (based on adjacent patterns)
  3. Offer to update the plan first
  4. Wait for direction

Critical: No Git Operations

NEVER perform any Git operations unless the user explicitly requests it.

This includes:

  • git add / staging files
  • git commit
  • git push
  • git checkout / branch operations
  • Any other git commands

The user manages their own Git workflow. After completing a step:

  1. Report what files were created/modified
  2. Wait for user to review and decide when/how to stage and commit
  3. Only perform Git operations if user explicitly says "commit this" or similar

Checklist

Before implementing each step, verify:

  • Read the relevant code/files the step will modify
  • Examined adjacent patterns in the codebase
  • Plan assumptions match codebase reality
  • Dependencies from previous steps are in place
  • Working on exactly one step (not combining multiple)

After user verification:

  • Plan checkboxes updated to mark step complete
  • Files changed reported to user
  • Awaiting instruction before proceeding to next step

平台分布

windsurf

27.01%
按下载量换算62

OpenCode

22.99%
按下载量换算53

Codex

17.8%
按下载量换算41

Claude Code

11.3%
按下载量换算26

Antigravity

7.12%
按下载量换算16

Gemini CLI

3.66%
按下载量换算8

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

只读

该 Skill 主要提供规则、说明或参考内容,本身偏只读;真正读写文件、联网或执行命令仍取决于宿主 Agent 的任务。

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。

继续浏览同类 Skills