AI workflow toolkit

Give your AI a workflow.

Pick a work scenario, then pick the AI tool you already use. RecoWork gives you ready-to-use prompts, project rules, knowledge templates, and continuation memory.

Prompt Init CLI usage Explore packs

3 starter workflow packs

6 supported tools

1 local scaffold

rw init project --platform codex .

Prompt

Pick a scenario Daily AI work Project workflow Structured learning

Pick a tool ChatGPT Mobile Codex / Cursor Notion / 飞书

01 No prompt writing from scratch

Choose a ready workflow pack and export it for your tool.

02 More than one prompt

Get steps, checklists, memory cards, and project templates together.

03 Keep using it when tools change

The same scenario can generate ChatGPT, Codex, Cursor, or Feishu versions.

Starter workflow packs

Start with the three most common AI work scenarios.

general

general-ai-workflow

Helps everyday users run tasks reliably in mobile or web AI apps with init prompts, task prompts, memory cards, and review checklists.

For ChatGPT Mobile / Claude Mobile / Kimi
Reduces context loss and inconsistent outputs

project

project-engineering

Sets up AGENTS.md, project skills, knowledge folders, quality gates, and durable capture for AI-assisted projects.

For Codex, Cursor, and Claude Code
Gives a project an AI working protocol from day one

learning

learning-engineering

Turns learning into roadmaps, chapters, exercises, homework feedback, reviews, and progress tracking.

For LangChain, coding, product, design, and more
Teaches chapter by chapter instead of dumping answers

Supported tools

Use your current tool. RecoWork generates the right version.

ChatGPT Mobile Segmented prompts, task cards, and continuation memory.

Claude Mobile Project instructions, context boundaries, and deep task guidance.

Kimi / 豆包 Chinese mobile prompts, summaries, steps, and copyable templates.

Codex AGENTS.md, skills, knowledge, and project rules.

Cursor rules.md, project context, and coding collaboration rules.

Notion / 飞书 Knowledge pages, task templates, reviews, and team SOPs.

Prompt Init

Ask AI to initialize the package for you.

01 Copy the init prompt

Use the prompt template with the RecoWork repo already filled in, then choose the package, platform, and target location.

02 Paste it into an AI assistant

Use Codex, ChatGPT, Claude, Cursor, or any AI that can create files or structured content.

03 Review the generated package

The AI should show the file tree, generated files, and the first step to use the package.

You are a RecoWork package initializer.

Source repository:
https://github.com/recoluan/recowork

Package:
project-engineering

Target platform:
codex

Target location:
current project

Read the repository, locate packages/project-engineering, generate the files this package needs for Codex, include a README, show the file tree, and tell me the first step.

CLI

Initialize workflow packs with RecoWork.

rw list
rw platforms
rw init project --platform codex .
rw init general --platform chatgpt-mobile ./my-ai-workflow
rw init learning -p notion ./langchain-study

How it works

Choose a workflow pack, then choose your target tool. The CLI creates files you can copy into chat or drop into a project.

For builders

If you want to extend packs, the internal structure is simple.

pack.yaml

Describes who the pack is for, what problem it solves, and which tools it supports.

core/

Stores the reusable method for the scenario, such as steps, principles, and checklists.

adapters/

Stores tool-specific versions for ChatGPT, Codex, Cursor, Feishu, and more.

examples/

Stores real scenarios so users can understand how to apply the pack.