Token导航 LogoToken导航

Agent Skill · 研究检索

replit-plan(重新制定计划)

replit-plan 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。

replit-plan 是研究检索类 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,按安装量在研究检索类的 73,675 个 Skill 里排第 44,681。

总安装 899周安装 36GitHub Stars 12下载 291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:replit-plan(重新制定计划)
来源仓库:https://github.com/sergio-bershadsky/ai
仓库路径:skills/replit-plan
安装命令:
npx skills add https://github.com/sergio-bershadsky/ai --skill replit-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。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/sergio-bershadsky/ai --skill replit-plan

这个 Skill 是做什么的

研究检索类 Skill 主要解决「找资料、核对事实、把来源整理清楚」的问题,通常会让助手先检索再总结。

用于查找、检索和筛选相关信息。适合在 Codex、Claude、Cursor、Gemini CLI 中根据关键词快速定位结果。可结合来源仓库和原始 README 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文件读写。安装方式:通过 npx skills add 从 GitHub 仓库安装。

来源仓库内路径:skills/replit-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/sergio-bershadsky/ai --skill replit-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

replit-plan 怎么安装?

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

replit-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。

replit-plan 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。

replit-plan 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。

replit-plan 用的人多吗?

公开数据是累计安装 899 次,近一周安装 36 次,下载 291 次,GitHub 12 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。

replit-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。
  • 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。
  • 需要处理研究检索类任务、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。

研究检索类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
find-skills

寻找技能 · Codex、Claude

31,199,81316,486操作浏览器MIT有提醒
self-improving-agent

自我改进 Agent · OpenClaw

16,464,0023,396接触敏感数据MIT-0通过
soultrace

灵魂追踪 · Codex、Claude

7,588,94412需要联网未明确有提醒
self-improving

自我完善 · OpenClaw

6,829,2841,046需要联网MIT-0通过
microsoft-foundry

微软代工厂 · Codex、Claude

6,614,144764调用外部服务未明确有提醒
replit-plan

重新制定计划 · 本页

89912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

Replit Task Planner

Break down projects into iterative development phases with checkpoints optimized for Replit Agent's workflow.

When to Use

  • Converting a large project into manageable phases
  • Planning checkpoint strategy for complex builds
  • Creating iterative prompts from a PRD or idea
  • Structuring work for Replit Agent's extended autonomy mode

Replit Agent Modes

Build Mode (Default)

  • Agent writes code directly
  • Best for: Implementation, coding, fixing bugs
  • Creates checkpoints automatically

Plan Mode

  • Agent discusses without modifying code
  • Best for: Architecture decisions, exploring approaches, reviewing PRDs
  • Free (no checkpoint charges)

Edit Mode

  • Agent makes targeted changes to specific files
  • Best for: Precise modifications, refactoring specific sections

Checkpoint Strategy

Replit Agent creates checkpoints after each prompt. Effective planning means:

  1. One checkpoint = one logical unit of work
  2. Each phase should be testable independently
  3. Phases build on each other sequentially
  4. Rollback-friendly boundaries

Task Breakdown Template

# [Project Name] - Development Plan

## Overview
**Project:** [Brief description]
**Total Phases:** [Number]
**Estimated Checkpoints:** [Number]

---

## Phase 1: [Phase Name]
**Goal:** [One sentence describing what's achieved]
**Dependencies:** None (starting point)

### Tasks
1. [Specific task 1]
2. [Specific task 2]
3. [Specific task 3]

### Deliverables
- [ ] [Testable outcome 1]
- [ ] [Testable outcome 2]

### Prompt for Replit Agent

[Phase goal description]. Specifically:

  1. [Task 1 with details]
  2. [Task 2 with details]
  3. [Task 3 with details]

Stop after completing these tasks so I can review before continuing.

### Verification Steps
- [ ] [How to verify task 1]
- [ ] [How to verify task 2]

---

## Phase 2: [Phase Name]
**Goal:** [One sentence]
**Dependencies:** Phase 1 complete

### Tasks
[...]

### Prompt for Replit Agent
[...]

---

## Rollback Points
- After Phase 1: [What's safe to rollback to]
- After Phase 2: [What's safe to rollback to]

## Risk Mitigation
| Risk | Phase | Mitigation |
|------|-------|------------|
| [Risk] | [Phase #] | [How to handle] |

Procedure

Step 1: Understand the Project

Gather from user:

  1. What is being built?
  2. What are all the features/requirements?
  3. Any dependencies between features?
  4. Are there external integrations?

Step 2: Identify Natural Boundaries

Look for:

  • Setup vs Features — Config/auth first, then features
  • Independent Features — Features that don't depend on each other
  • Dependent Features — Features that require others to exist
  • Polish vs Core — Error handling, edge cases last

Step 3: Create Phases

Phase 1: Foundation Always start with:

  • Project setup
  • Database schema
  • Authentication (if needed)
  • Basic navigation/layout

Phase 2-N: Core Features Group related functionality:

  • One major feature per phase
  • Include its UI, API, and data needs
  • Keep phases 30-60 minutes of Agent work

Final Phase: Polish Always end with:

  • Error handling
  • Loading states
  • Edge cases
  • Responsive fixes
  • Final testing

Step 4: Write Prompts for Each Phase

Each prompt should:

  • State the goal clearly
  • List specific tasks (numbered)
  • Include technical details
  • Request a stop point for review

Template:

Implement [Feature Name].

Requirements:
1. [Specific requirement with details]
2. [Specific requirement with details]
3. [Specific requirement with details]

Technical notes:
- [Relevant constraint or approach]
- [Relevant constraint or approach]

Create a checkpoint when complete so I can review before the next phase.

Step 5: Define Verification Steps

For each phase, include how to verify:

  • What to click/test in the UI
  • What API calls to make
  • What database records to check

Step 6: Present the Plan

## Development Plan: [Project Name]

**Phases:** [Number]
**Approach:** [Brief strategy explanation]

---

[Full plan with all phases]

---

## How to Use This Plan

1. **Start with Plan Mode**: Copy Phase 1 prompt, select Plan Mode
2. **Review Agent's approach**: Ensure it aligns with expectations
3. **Switch to Build Mode**: Approve and let Agent implement
4. **Verify at checkpoint**: Test deliverables before continuing
5. **Proceed to next phase**: Copy next prompt and repeat

## Iteration Strategy
If issues arise:
- Minor issues: Describe and ask Agent to fix in current phase
- Major issues: Rollback to previous checkpoint, refine prompt, retry

Ready to start?

Phase Sizing Guidelines

Phase SizeWork AmountExample
Small15-30 minAdd a single form with validation
Medium30-60 minFull CRUD for one entity
Large60-90 minFeature with UI + API + data

Ideal phase: Medium (30-60 minutes)

Signs phase is too large:

  • More than 5 major tasks
  • Touches more than 4-5 files
  • Multiple unrelated features
  • "Build the entire X" language

Signs phase is too small:

  • Single task that takes < 10 minutes
  • No testable deliverable
  • Could easily be combined with adjacent phase

Common Phase Patterns

Pattern 1: CRUD Feature

Phase N: [Entity] Management
1. Database model and migrations
2. API endpoints (list, create, read, update, delete)
3. List view with table/cards
4. Create/edit form (modal or page)
5. Delete with confirmation

Deliverables:
- [ ] Can create new [entity]
- [ ] Can view list of [entities]
- [ ] Can edit existing [entity]
- [ ] Can delete [entity]

Pattern 2: Authentication

Phase 1: Authentication Setup
1. Configure auth provider (Supabase/Clerk/etc.)
2. Create signup page with form
3. Create login page with form
4. Add protected route wrapper
5. Add user context/hook
6. Add logout functionality

Deliverables:
- [ ] New user can sign up
- [ ] Existing user can log in
- [ ] Protected pages redirect to login
- [ ] User can log out

Pattern 3: Dashboard

Phase N: Dashboard
1. Create dashboard layout
2. Add metric cards with data fetching
3. Add recent activity list
4. Add quick action buttons
5. Implement responsive grid

Deliverables:
- [ ] Dashboard loads with correct data
- [ ] Metrics update when data changes
- [ ] Recent activity shows latest items
- [ ] Layout works on mobile

Pattern 4: External Integration

Phase N: [Service] Integration
1. Set up environment variables for API keys
2. Create service wrapper/client
3. Implement core integration function
4. Add error handling for API failures
5. Connect to UI trigger point

Deliverables:
- [ ] API key configuration works
- [ ] Integration function returns expected data
- [ ] Errors handled gracefully
- [ ] UI reflects integration state

Example: Full Task Breakdown

Project: Todo App with Categories

# Todo App - Development Plan

## Overview
**Project:** Todo app with categories, due dates, and search
**Total Phases:** 4
**Estimated Checkpoints:** 5

---

## Phase 1: Foundation
**Goal:** Set up project with database and basic structure
**Dependencies:** None

### Tasks
1. Initialize React + Vite project with TailwindCSS
2. Set up SQLite database with Prisma
3. Create Todo and Category models
4. Set up Express API with basic structure
5. Create app shell with header and main area

### Deliverables
- [ ] Project runs without errors
- [ ] Database creates successfully
- [ ] Empty app shell renders

### Prompt

Set up a Todo app with React, Vite, TailwindCSS, Express, and SQLite with Prisma.

  1. Initialize React + Vite project
  2. Add TailwindCSS configuration
  3. Create Express backend with /api prefix
  4. Set up Prisma with SQLite
  5. Create models:

- Category: id (uuid), name (string), color (string) - Todo: id (uuid), title (string), completed (boolean), due_date (date nullable), category_id (uuid nullable, FK)

  1. Create basic app layout with header showing "Todo App"

Stop when the shell runs and database migrates successfully.

---

## Phase 2: Category Management
**Goal:** Full CRUD for categories
**Dependencies:** Phase 1

### Tasks
1. API endpoints for categories
2. Category list sidebar
3. Add/edit category modal
4. Delete category with confirmation
5. Color picker for categories

### Deliverables
- [ ] Can create category with name and color
- [ ] Categories display in sidebar
- [ ] Can edit category
- [ ] Can delete category

### Prompt

Implement category management for the Todo app.

  1. Create API endpoints:

- GET /api/categories - list all - POST /api/categories - create new - PUT /api/categories/:id - update - DELETE /api/categories/:id - delete

  1. Create sidebar showing category list with colored dots
  2. Add "New Category" button that opens modal
  3. Modal has: name input, color picker (preset colors), save/cancel
  4. Category item has edit and delete buttons
  5. Delete shows confirmation before removing

Use a simple color picker with 8 preset colors.

Stop when categories can be created, edited, and deleted.

---

## Phase 3: Todo Management
**Goal:** Full todo functionality with categories and due dates
**Dependencies:** Phase 2

### Tasks
1. API endpoints for todos
2. Todo list component with checkboxes
3. Add todo form
4. Edit todo modal
5. Category assignment dropdown
6. Due date picker
7. Filter todos by category (click sidebar)

### Deliverables
- [ ] Can create todo with optional category and due date
- [ ] Can mark todo complete/incomplete
- [ ] Can edit todo details
- [ ] Can delete todo
- [ ] Clicking category filters todos

### Prompt

Implement todo management with category assignment and due dates.

  1. Create API endpoints:

- GET /api/todos?category_id=x - list todos, optional filter - POST /api/todos - create - PUT /api/todos/:id - update - DELETE /api/todos/:id - delete

  1. Main area shows todo list:

- Checkbox to toggle complete (strikethrough when done) - Title text - Category badge (colored) - Due date badge (red if overdue) - Edit/delete buttons on hover

  1. Add todo form at top:

- Title input (required) - Category dropdown (optional) - Due date picker (optional)

  1. Click category in sidebar to filter todos (highlight selected)
  2. Edit todo opens modal with all fields editable

Sort todos: incomplete first, then by due date (soonest first).

Stop when todos can be fully managed with categories and dates.

---

## Phase 4: Search and Polish
**Goal:** Add search and polish UX
**Dependencies:** Phase 3

### Tasks
1. Search input in header
2. Search filters todos by title (client-side)
3. Empty states for no todos / no results
4. Loading states
5. Error handling with toast messages
6. Responsive design adjustments

### Deliverables
- [ ] Search filters todos in real-time
- [ ] Empty states display appropriately
- [ ] Errors show user-friendly messages
- [ ] Works on mobile screens

### Prompt

Add search functionality and polish the UX.

  1. Add search input in header (right side)
  2. Search filters todos by title (case-insensitive, client-side)
  3. Add empty states:

- No todos: "No todos yet. Add one above!" - No results: "No todos match your search" - No category selected + no todos: "Select a category or add a todo"

  1. Add loading spinner for initial data fetch
  2. Add error toast for API failures
  3. Responsive adjustments:

- Mobile: hide sidebar, add hamburger menu - Sidebar overlays on mobile when menu open

Test all flows and fix any visual issues.

This is the final phase - ensure everything works smoothly.

---

## Rollback Points
- After Phase 1: Clean slate with just structure
- After Phase 2: Categories work, can redo todo implementation
- After Phase 3: Full functionality, can redo polish

## Risk Mitigation
| Risk | Phase | Mitigation |
|------|-------|------------|
| Prisma schema issues | 1 | Test migration before proceeding |
| Filter state complexity | 3 | Use URL params for filter state |
| Mobile responsive issues | 4 | Test each breakpoint individually |

Output Format

# Development Plan: [Project Name]

## Summary
| Metric | Value |
|--------|-------|
| Total Phases | [Number] |
| Estimated Build Time | [Range] |
| Checkpoint Strategy | [Brief] |

---

[Phase-by-phase breakdown with prompts]

---

## Quick Reference

### Phase Prompts (Copy-Paste Ready)

**Phase 1:**

[Prompt]

**Phase 2:**

[Prompt]

[Continue for all phases]

---

## Verification Checklist

After each phase, verify:
- [ ] Phase 1: [Key items]
- [ ] Phase 2: [Key items]
[...]

Ready to begin?

Rules

  1. ALWAYS start with foundation phase — Setup, schema, auth first
  2. ALWAYS end with polish phase — Error handling, edge cases last
  3. ALWAYS include verification steps — Testable deliverables per phase
  4. ALWAYS size phases appropriately — 30-60 minutes ideal
  5. NEVER combine unrelated features — One major feature per phase
  6. NEVER skip rollback planning — Document safe rollback points
  7. PREFER explicit prompts — Include technical details in each phase prompt

平台分布

Codex

35.86%
按下载量换算104

Claude

30.21%
按下载量换算88

Cursor

19.39%
按下载量换算56

Gemini CLI

9.51%
按下载量换算28

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