AI Guards - Feature Planning Tool

AI-powered feature planning tool with MCP (Model Context Protocol) integration for AI assistants like Cursor, Claude Desktop, and Windsurf.

Features

✨ Simple & Focused: Just plan features - no complex configuration
🤖 MCP Integration: Works seamlessly with AI-powered editors
📋 Structured Plans: Generate consistent, comprehensive feature plans
🚀 Quick Setup: Initialize in seconds

Installation

Global Installation

npm install -g ai-guards

Quick Start

# Initialize in your project
npx ai-guards init

# Create your first plan
npx ai-guards plan --title "Add user authentication"

Usage

Commands

Examples

# Initialize project with default .plans folder
ai-guards init

# Initialize with custom folder name
ai-guards init --folder "my-plans"

# Create a plan with custom title
ai-guards plan --title "Implement dark mode"

# Create a plan with custom title and author
ai-guards plan --title "Add search feature" --author "John Doe"

MCP Integration

AI Guards can be integrated with MCP-compatible tools like Cursor, Claude Desktop, and Windsurf.

Setup

Add to your MCP configuration:

{
  "mcpServers": {
    "ai-guards": {
      "command": "npx",
      "args": ["-y", "ai-guards-mcp"]
    }
  }
}

Available MCP Tools

• init: Initialize AI Guards in your project with optional custom folder name
• create-plan: Create a new feature plan with title and optional author
• plan-feature: Prompt template for comprehensive feature planning

For detailed MCP setup instructions, see MCP Integration Guide.

Directory Structure

AI Guards supports multiple directory structures:

Default (new projects):

.plans/                 # Feature development plans

Custom folder:

my-plans/               # Your custom folder name
.ai-guards-config       # Stores the custom folder name

Legacy (backward compatibility):

.ai-guards/
└── plans/              # Feature development plans

Plan Structure

Each plan includes:

• Scope: Clear definition of what will be built
• Functional Requirements: What the feature should do
• Non-Functional Requirements: Performance, security, scalability
• Guidelines & Packages: Development standards and dependencies
• Threat Model: Security considerations
• Execution Plan: Step-by-step implementation steps

Example Plan

---
id: plan-001
title: Add user authentication
createdAt: 2024-01-15
author: Developer
status: draft
---

## 🧩 Scope
Implement login, logout, and session validation endpoints using JWT.

## ✅ Functional Requirements
- POST /login: authenticate user, return JWT
- POST /logout: invalidate client-side token
- Middleware: validate token on protected routes

## ⚙️ Non-Functional Requirements
- API response time < 200ms
- Log all auth attempts

## 📚 Guidelines & Packages
- Use jsonwebtoken (MIT license)
- Follow REST API conventions

## 🔐 Threat Model (Stub)
- Credential stuffing
- Token replay attacks

## 🔢 Execution Plan
1. Create auth module scaffold
2. Implement POST /login with JWT
3. Implement logout functionality
4. Add token validation middleware
5. Write comprehensive tests

License

MIT

Command Reference

Initialize Project

# Initialize with default .plans folder
ai-guards init

# Initialize with custom folder name
ai-guards init --folder "my-plans"

Creates the specified directory structure for storing plans.

Create Plans

# Basic plan
ai-guards plan

# With custom title
ai-guards plan -t "Feature name"

# With custom title and author
ai-guards plan -t "Feature name" -a "Author Name"

Plans are saved in your configured plans directory and can be edited with any text editor.