task-planning
Skill2 starsChanged 2 months agoMITView on GitHub
任务规划工作流:把多步骤需求拆成可执行、可逐项验证的任务清单。当接到多步骤需求、跨多文件的新功能开发、跨模块改动或需要评估影响范围时使用;覆盖阅读现有实现、识别相似模式、拆分任务与为每个任务指定验证方式。
8 other projects use a version of this.
Adopted by
Most recently taken up by
- NoobyGains/godmodetheir own version, first seen Nov 1, 2026
- microsoft/amplifier-app-papercliptheir own version, first seen Oct 25, 2026
- mustafahasanain/frappe-workflowtheir own version, first seen Oct 8, 2026
- lasp/adamant-skillstheir own version, first seen Aug 13, 2026
- TTflysky/sirenhuisuotheir own version, first seen Aug 4, 2026
- 5kuper/ai-docs-kittheir own version, first seen May 25, 2026
Other versions8 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 8
- NoobyGains/godmode★ 118same name
- lasp/adamant-skills★ 4same name
- TTflysky/sirenhuisuo★ 4same name
- zebbern/termstack★ 4same name
- microsoft/amplifier-app-paperclip★ 1same name
- youjun011/skills★ 1same name
- 5kuper/ai-docs-kit★ 0same name
- mustafahasanain/frappe-workflow★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository5 other files it gives its agents.
- code-reviewSkill
- project-initializationSkill
- verificationSkill
- feature-developmentSkill
- {{skill-name}}Skill
The file
Files
Task Planning(任务规划)
Purpose
把一个多步骤需求变成一份可执行的任务计划:任务清单 + 影响范围 + 每步验证方式, 并落实文件变更 / 数据流 / 测试方案三要素。产出的是计划,不是代码实现。
When to Use
- 当接到多步骤需求或新功能开发,预计改动跨多个文件 / 模块时
- 当需求与现有实现的关系不明,需要先读代码确认改动点时
- 当用户要求「先给方案 / 拆任务 / 评估工作量」再动手时
- 当同一需求要拆给多个会话或多个人并行执行,需要一份共同依据时
When NOT to Use
- 单文件小改动(改一个函数、一处样式):直接实现,结束前用
verification收尾 - 已有 issue 或设计文档列出了明确任务清单:按文档执行,不重复规划
- 定位报错根因:改用
debugging-and-error-recovery,定位后再回来规划修复 - 测试先行的开发循环:改用
test-driven-development - 贯穿数据库到 UI 的全栈功能开发:改用
feature-development(其七步顺序更具体)
Inputs
- 需求描述(issue、对话或设计文档):缺失或含关键歧义时先问清目标与验收标准,不自行假设
- 仓库读权限(现有实现与 Git 状态):缺失时报告并停止,不凭记忆规划
- 项目入口资料(CLAUDE.md、package.json、目录结构):缺失时按常规布局探索,并把结论记为假设
Workflow
1. Inspect
- 读
CLAUDE.md/README拿项目约定;读package.json的scripts与依赖,确认可用的检查命令 git status+git log --oneline -10:确认当前分支、未提交改动与近期改动风格- 找相似实现:按关键词检索同类组件、API、类型与测试,记录 1-3 个可参照的
文件:行号 - 列出可能受影响的模块及其现有约束(命名、分层、状态管理、测试组织方式)
2. Plan
- 把需求拆成有序任务:每条以动词开头,附一条可判定的完成标准
- 每个任务标注影响范围:改哪些文件、哪些调用方受影响
- 每个任务指定验证方式,具体到命令形态(如
npx vitest run src/foo、npx tsc --noEmit) - 发现歧义停下先问;不在计划里替用户决定实现路径
3. Implement
- 本步产出计划本身,不写业务代码:文件变更清单(新增 / 修改 / 删除)、 数据流(输入 → 处理 → 输出 → 存储)、测试方案(新增哪些用例、覆盖哪些路径)
- 计划默认落在回复正文;仅当用户要求时创建计划文件,用编号 + 复选框表达,保持可勾选、可被后续任务引用
4. Verify
- 逐条对照需求:每个验收点都有对应任务,无遗漏
- 每个任务都有影响范围与验证方式,且粒度可独立验证
- 用检索复核影响范围:被改符号的全部引用处都已列入计划
Rules
- 必须先读现有实现再拆任务,禁止规划不存在的接口或不存在的文件路径
- 禁止把「实现功能」写成一个黑盒任务;最小任务粒度 = 一次可独立验证的改动
- 禁止无关重构:计划中的每一项都必须能追溯到需求原文
- 关键歧义必须先问;用户未答复前不产出最终计划
- 计划不得引入需求之外的架构或依赖变更
Common Failure Modes
| 症状 | 纠正 |
| 计划里出现「优化」「完善」这类不可判定任务 | 改写成「改哪个文件、达到什么可验证状态」 |
| 某任务依赖未确认的外部接口 | 标记为阻塞项,先问清再继续拆分 |
| 影响范围遗漏调用方 | 检索被改符号的全部引用处补齐 |
| 任务列表没有配套验证方式 | 给每个任务补一条具体命令或检查动作 |
Common Rationalizations
| Agent says | Correct response |
| "需求很清楚,不用先读代码" | 先列出 1-3 个相似实现文件,再开始拆任务 |
| "任务粒度粗一点没关系,写的时候再说" | 每个任务给出可判定的完成标准与验证命令 |
| "影响范围大概就这些文件" | 检索被改符号的全部引用处核对 |
| "歧义先按我的理解做,做完再问" | 关键歧义停下先问,不带假设出计划 |
Verification
- 计划中的每个文件路径 → 用
ls/ Read 确认存在,或明确标注为「新增」 - 每个被改符号 → 检索其引用处,证明影响范围覆盖完整
- 计划三要素 → 文件变更 / 数据流 / 测试方案各至少一节,缺一不算完成
Exit Criteria
- 任务有序、逐条可独立验证,且每条带完成标准
- 文件变更 / 数据流 / 测试方案三要素齐备
- 每个任务标注影响范围与验证方式
- 关键歧义已问清,或已作为阻塞项显式列出
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

