@guru-ai-product/ai-product-kit

v0.2.251226142524

Published

4 months ago

Sync the AI Product Kit Skill bundle through npx without cloning the repository.

0High
0Medium
0Low

charge.woo

mcp model-context-protocol ai-product-kit

AI Product Kit

AI Product Kit is a doc-first toolkit designed for product managers and cross-functional teams, offering end-to-end guidance from discovery to launch.

Looking for the Chinese version? See README_CN.md.

📁 Repository Structure

PACKAGE_ROOT/
├── README.md / README_CN.md    # English / Chinese entry points
├── package.json                # NPM package manifest
├── bin/                        # Distribution-ready CLI entry points
├── mcp/                        # MCP server implementation and templates
├── skills/                     # Canonical lifecycle Skills and reusable guides
└── tools/build_package/        # Bundled build scripts for MCP + template refresh

Documentation Standards: All documentation follows the standards defined in docs/guide_writing_standard/README.md, covering document structure, content organization, formatting, and template writing guidelines.

🎯 Who This Helps

Product managers – start with the aipk_requirements-intake, aipk_requirements-documentation, and aipk_requirements-changes skills for research, PRDs, and change management.
Designers – invoke the aipk_update-requirements-from-design and aipk_format-requirement-pages-with-design skills when you need to align composite mockups, split panels, or embed visuals into requirement pages.
Engineers – use the aipk_development skill for architecture notes and delivery checklists.
QA teams – use the aipk_requirements-documentation skill for acceptance criteria and pair with aipk_development for validation tasks.
Operators & growth – rely on the aipk_aso-new-release skill for launch updates, store copy, and asset guidance.

📚 Core Capabilities

Discovery & analysis – market studies, user interviews, competitor reviews.
Product planning – PRD templates, backlog framing, prioritization.
Design systems – UI/UX principles, prompt and content guidelines.
Delivery enablement – architecture references, API design, dev workflows.
Quality practice – test plans, automation strategy, regression tooling.
Release & operations – go-to-market, post-launch metrics, growth loops.

⚙️ Environment Requirements

Node.js 18+ – required to run npx @guru-ai-product/ai-product-kit@latest from ROOT.
npm 9+ – provides access to the scoped package and carries your authenticated session, so the sync command can download the latest Skill bundle.

🛠️ Usage Guide

AI Product Kit provides two ways to access its capabilities: Traditional Skills and MCP (Model Context Protocol) Server.

🔍 Choosing Between Skills and MCP

| Method | Best For | How It Works | |--------|----------|--------------| | Skills | Full project lifecycle, document templating, collaborative workflows, and IDE/CLI agents that operate inside your repository (Cursor, Codex, Claude Code, Gemini CLI) | Downloads skill bundles to .claude/skills/ directory for agent access | | MCP | Real-time skill queries, programmatic access, tool integration, and agents that run outside the project workspace (e.g., Cherry Studio) | Runs a server that provides skills as tools via Model Context Protocol |

Choose one mode per project. Do not copy both templates or mix the automation blocks.

📋 Using Skills (Traditional Method)

Skills provide templates, checklists, and guidance for each phase of the product lifecycle. They are downloaded to your local project and accessed by agents through the file system.

1. 📦 Installation & Update

Choose one of the following methods depending on whether you want a direct install or an OpenSkill install:

Option A – Direct install

Simple, one-command setup with no extra dependencies beyond npm.

# Install or update Skills from your project root directory
npx -y @guru-ai-product/ai-product-kit@latest ai-product-kit

This downloads the Skill bundle (templates, scripts, and GURU_AI.md files) to .claude/skills/ in your working tree. The bundle is ready to use immediately—agents learn where every Skill lives from AGENTS.md, so no explicit initialization command is required. See Quick Start below for lifecycle-specific prompts.

Option B – OpenSkill install

Flexible approach that lets you add other community Skill bundles in addition to the official set.

Ensure the CLI is installed globally:
```
npm i -g openskills
```
Run the install command from your project root, pointing to the desired Skill:
```
openskills install https://github.com/castbox/ai-product-workflow
```
Sync the Skill metadata so your agents see the newly added Skill:
```
openskills sync
```

These steps can be repeated for any other Skill repositories you want to bake into this workspace.

⚡ Using MCP Server

The MCP Server makes AI Product Kit skills available as tools through the Model Context Protocol. Configure your preferred MCP client to connect to the server.

1. 🔧 Configure MCP Client

Add this configuration to your MCP client so you can access AI Product Kit skills:

Option A: Using npx (Recommended)

Codex CLI – add the entry below to ~/.codex/config.toml:

[mcp_servers.ai_product_kit]
command = "npx"
args = ["-y", "-p", "@guru-ai-product/ai-product-kit@latest", "ai-product-kit-mcp"]
timeout = 60

Cursor IDE – add to your settings.json or via GUI:

{
  "mcpServers": {
    "ai-product-kit": {
      "command": "npx",
      "args": ["-y", "-p", "@guru-ai-product/ai-product-kit@latest", "ai-product-kit-mcp"]
    }
  }
}

Other MCP Clients – use the same pattern:

Command: npx
Arguments: ["-y", "-p", "@guru-ai-product/ai-product-kit@latest", "ai-product-kit-mcp"]

Option B: Using Local Installation

If you install the package locally:

npm install @guru-ai-product/ai-product-kit

Then configure with:

{
  "mcpServers": {
    "ai-product-kit": {
      "command": "./node_modules/.bin/ai-product-kit-mcp",
      "args": []
    }
  }
}

3. 🛠️ Available Tools

Once configured, the MCP Server provides the product_kit_skill tool:

Tool Name: product_kit_skill
Purpose: Retrieve any AI Product Kit skill guide or template
Parameters:
skill_name (string, required) - Name of the skill to retrieve (e.g., aipk_requirements-documentation)

The tool returns:

skill_name: The requested skill name
description: Skill description from the SKILL.md file
cache_path: Absolute path to cached content (for subsequent reading)

4. 🚀 Initialize Your Project

Use the MCP tool to initialize your project:

Initialize my project using the init-project skill with the product_kit_skill tool.

5. 📚 Supported Skills

All AI Product Kit skills are available via MCP:

aipk_init_project - Project bootstrapping
aipk_requirements-intake - Intake and research automation
aipk_requirements-documentation - PRD drafting + iteration templates
aipk_requirements-changes - Change documentation and prompts
aipk_initiative-planning - Initiative evaluation and scoping
aipk_update-requirements-from-design - Sync requirements with design mockups
aipk_format-requirement-pages-with-design - Table formatting for design pages
aipk_aso-new-release - ASO copy and asset guidance
aipk_development - Engineering playbooks
aipk_skill_generate - Skill creation standards
aipk_tool_prompts - Reusable AI prompt catalog

🚀 Quick Start

Agents already know where each Skill lives because the mappings are declared in AGENTS.md. Grab a prompt by lifecycle stage:

Project setup audit
```
Init project.
```
Expected skill: aipk_init_project

Discovery & intake

Use pre-collected SensorTower、七麦、点点 data to produce the intake brief.

Expected skill: aipk_requirements-intake

PRD drafting

Generate a multimodal search PRD based on (path of the draft) using the standard template.

Expected skill: aipk_requirements-documentation

Change control

Summarize v1.2→v1.3 sharing flow changes in the standard change log.

Expected skill: aipk_requirements-changes

Design sync
```
Use (path of design image) to update the linked requirements and refresh mapping.csv.
```
Expected skills: aipk_update-requirements-from-design, aipk_format-requirement-pages-with-design

Requirements review

Review v1.3 of all requirements and flag gaps or inconsistencies.