context-forge
Advanced tools
AI orchestration platform with autonomous teams, enhancement planning, migration tools, 25+ slash commands, checkpoints & hooks. Multi-IDE: Claude, Cursor, Windsurf, Cline, Copilot
CLI tool for instant Claude Code project scaffolding with context engineering best practices
Transform your project ideas into AI-ready applications with intelligent context engineering
🎯 Optimized for Claude Code with deep integration for slash commands, hooks, and PRPs
Also supports Cursor, Windsurf, Cline, Copilot, Gemini, and more AI coding assistants
🦄 Nuxt 4 SupportNEW: Full Nuxt 4 Integration Complete support for Nuxt 4 with Vue 3 and modern development!
context-forge initSelect "Nuxt 4" as frontend framework |
🤖 AI-Powered PRP GenerationContinued: Intelligent, feature-specific PRPs with AI enhancement!
context-forge ai-keys --provider openai
|
⚡ 25+ Slash CommandsComprehensive command library:
|
🪝 7 Claude Code Hooks
|
📄 Core Files
|
⚡ .claude/commands/
|
🪝 .claude/hooks/
|
Product Requirement Prompts (PRP) now available for Claude, Cursor, Windsurf, Cline, Copilot & Gemini!
Structured feature implementation with validation gates across all major AI coding assistants!
Context Forge is a powerful CLI tool that bridges the gap between project requirements and AI-assisted development. By implementing Andre Karpathy's context engineering principles, it generates comprehensive documentation structures that enable AI IDEs to understand and build your project efficiently—without hallucinations or context confusion.
🌐 Universal AI IDE Support - Works with ALL major AI coding assistants
⚡ Zero Configuration - Intelligent defaults for each IDE
🎯 One Command - Generate configs for multiple IDEs simultaneously
🔄 Format Conversion - Switch between IDEs without starting over
🔗 Claude Hooks Integration - Seamless context preservation during compaction
📦 No Lock-in - Your project, your choice of AI assistant
 Claude Code Full PRP support |  Cursor IDE PRP + MDC format |  Windsurf PRP + Cascade AI | |
 Cline PRP + Context mgmt |  Roo Code Hierarchical rules |  Gemini PRP + CLI & Code Assist |  GitHub Copilot Custom instructions |
Coming Soon: Amazon CodeWhisperer, Tabnine, and more!
Need help understanding how each IDE uses its configuration? Check out our detailed guides:
📚 Complete Claude Features Guide - Master the new advanced features:
Transform your project development with intelligent, feature-specific Product Requirement Prompts (PRPs) powered by OpenAI and Anthropic Claude.
Unlike template-based PRPs, AI-powered PRPs are dynamically generated for each specific feature in your project, containing:
# 1. Set up your AI provider (one-time setup)
context-forge ai-keys --provider openai
# OR
context-forge ai-keys --provider anthropic
# 2. Initialize project with AI-powered PRPs
context-forge init --ai-prp
# Your API keys are securely encrypted and stored in ~/.context-forge-keys
Watch as Context Forge generates intelligent PRPs for each feature:
🤖 Generating AI PRP for: User Authentication & Authorization...
✅ AI content generated for: User Authentication & Authorization
📝 Generated: PRPs/feature-auth-prp.md
🤖 Generating AI PRP for: Google Drive Cloud Streaming...
✅ AI content generated for: Google Drive Cloud Streaming
📝 Generated: PRPs/feature-google-drive-cloud-streaming-prp.md
Traditional Template PRP:
# TODO: Add implementation details
# TODO: Add data models based on requirements
AI-Powered PRP for Google Drive Integration:
### AI-Generated Implementation Strategy
Leverage Google Drive API for file organization, streaming, and redundancy.
Use Google's Picker API for intelligent folder structure management.
Implement JWT for access control.
### ⚠️ AI-Identified Gotchas
- Make sure to have proper error handling for API failures
- Ensure you're not exceeding Google Drive API's rate limits
- Be careful with user's OAuth tokens
### ✅ AI-Recommended Best Practices
- Encrypt OAuth tokens
- Use Google's recommended best practices for Drive API
- Always validate files before streaming
Context Forge provides deep integration with Claude Hooks Manager through 4 powerful hooks that enhance your development workflow and maintain context throughout long sessions.
With Claude Code v1.0.48+, the PreCompact hook ensures your PRPs and project context survive conversation compaction:
Intelligently manages which files stay in context as you work:
Runs validation before code submission:
Tracks PRP implementation progress:
When Dart task management is enabled, Context Forge generates additional hooks for comprehensive task tracking:
Automatically updates Dart task progress based on code changes:
Adds intelligent comments to Dart tasks with code analysis:
Maintains smart mapping between Dart tasks and code files:
Enable Dart Integration:
context-forge init
# Select "Enable Dart task integration? Yes"
Generated Dart Integration Files:
.claude/dart_progress.json - Task progress tracking.claude/task_mapping.json - File-to-task relationships.claude/task_comments.json - Automated task comments.claude/task_suggestions.json - AI-generated task suggestions# Generate project with hooks
context-forge init # Select "Enable Claude Code hooks? Yes"
# Or copy hooks from another project
context-forge copy-hooks --source ../claude-hooks-repo/hooks
# Hooks are created in:
# .claude/hooks/
This integration is especially powerful for:
Context Forge introduces a powerful checkpoint system that pauses AI development at critical milestones for human verification, ensuring you maintain control over the implementation process.
/checkpoint command anytime/checkpoint [description] - Create manual checkpoint/should-checkpoint - AI asks if checkpoint is needed/milestone-gate [milestone] - Major milestone verification🛑 CHECKPOINT: Human Verification Required
Analytics Dashboard - I've completed database schema setup:
✅ What I've Accomplished:
- Created PostgreSQL schema with 5 tables
- Set up foreign key relationships
- Added indexes for query optimization
🧪 Please Test:
1. Run: npm run db:migrate
2. Check tables exist: psql -c "\dt"
3. Verify foreign keys: psql -c "\d+ users"
📋 Critical Verification Points:
- [ ] All tables created successfully
- [ ] Foreign keys properly linked
- [ ] No data loss from existing tables
Please respond:
- ✅ "Approved" - Continue to next task
- ❌ "Issues: [description]" - Fix before proceeding
Checkpoints are configured during project initialization or can be added to existing projects:
# .context-forge/checkpoints.yaml
checkpoints:
- name: "Database Migration"
phase: "infrastructure"
requires_approval: true
validation_steps:
- "Run migration scripts"
- "Verify data integrity"
- "Check rollback procedures"
npm install -g context-forge
# Initialize a new project
context-forge init
# Initialize with AI-powered PRPs (optional)
context-forge init --ai-prp
# Set up AI provider keys (optional)
context-forge ai-keys
# Or use npx without installation
npx context-forge init
Note: Context Forge is a CLI tool and should be installed globally. Do not install it as a project dependency with npm install context-forge as this will add unnecessary files to your project.
# Initialize in current directory
context-forge init
# Specify output directory
context-forge init --output ./my-project
# Use with existing PRD file
context-forge init --prd ./requirements.md
# Skip interactive prompts with config
context-forge init --config ./context-forge.json
# Run validation on existing project
context-forge validate
# Retrofit existing projects with AI-optimized documentation
context-forge analyze
# Execute PRPs with Claude Code for one-pass implementation
context-forge run-prp feature-name
# NEW: Plan technology migration
context-forge migrate --target "Next.js"
# NEW: Copy Claude Code hooks from another project
context-forge copy-hooks --source ../claude-hooks-repo/hooks
# NEW: Set up AI provider keys for enhanced PRP generation
context-forge ai-keys --provider openai
context-forge ai-keys --provider anthropic
$ context-forge init
? Project name: Analytics Dashboard
? Project type: fullstack
? Project description: Real-time analytics dashboard with data visualization
? How would you like to provide the PRD? Create new PRD
? Frontend framework: nextjs
? Backend framework: fastapi
? Database: postgresql
? Select features: authentication, dashboard, realtime-updates, data-export
? Project timeline: standard
? Team size: small
? Enable PRP generation? Yes
? Enable AI-powered PRP generation? Yes
? Enable validation system? Yes
? Enable Human-in-the-Loop checkpoints? Yes
? Enable Claude Code hooks? Yes
✅ Project initialized successfully!
Generated files:
- CLAUDE.md (Next.js 15 specific configuration)
- Docs/Implementation.md (4 development stages)
- Docs/project_structure.md
- PRPs/base.md (Implementation prompt with validation gates)
- .claude/commands/ (20+ slash commands)
- .claude/hooks/ (4 Claude Code hooks)
- .context-forge/config.json
Next steps:
1. cd analytics-dashboard
2. Set up AI keys (optional): context-forge ai-keys
3. Review CLAUDE.md for project rules
4. Use /prime-context to load project knowledge and activate architect mode
5. Start with Stage 1 in Docs/Implementation.md
$ context-forge init --output user-service
? Project name: User Service
? Project type: api
? Project description: User management microservice with JWT auth
? Frontend framework: none
? Backend framework: express
? Database: mongodb
? Authentication method: jwt
? Enable Docker support? Yes
? Enable GitHub Actions? Yes
Generated structure:
user-service/
├── CLAUDE.md # Express.js specific rules
├── Docs/
├── PRPs/
├── Dockerfile
├── docker-compose.yml
└── .github/workflows/
$ context-forge init --output nuxt-dashboard
? Project name: Analytics Dashboard
? Project type: fullstack
? Project description: Modern analytics dashboard with real-time data
? Frontend framework: nuxt4 # NEW: Nuxt 4 option
? Styling framework: tailwind
? State management: pinia
? Backend framework: fastapi
? Database: postgresql
? Authentication method: jwt
? Enable AI-powered PRP generation? Yes
Generated structure:
nuxt-dashboard/
├── CLAUDE.md # Nuxt 4 + Vue 3 specific rules
├── Docs/
├── PRPs/
│ ├── nuxt-dashboard-prp.md
│ └── feature-auth-prp.md # AI-enhanced for Nuxt 4
└── .claude/commands/ # 25+ slash commands
✨ Nuxt 4 Features Included:
- New app/ directory structure documented
- Vue 3 Composition API guidelines
- TypeScript integration with auto-imports
- SSR and data fetching best practices
- Pinia state management setup
# context-forge.json
{
"projectName": "E-Commerce Platform",
"projectType": "fullstack",
"description": "Modern e-commerce platform with AI recommendations",
"prd": {
"content": "# E-Commerce Platform\n\n## Overview\n...",
"features": [
{
"id": "auth",
"name": "User Authentication",
"priority": "must-have",
"complexity": "medium"
},
{
"id": "catalog",
"name": "Product Catalog",
"priority": "must-have",
"complexity": "complex"
},
{
"id": "cart",
"name": "Shopping Cart",
"priority": "must-have",
"complexity": "medium"
},
{
"id": "ai-recommendations",
"name": "AI Product Recommendations",
"priority": "should-have",
"complexity": "complex"
}
]
},
"techStack": {
"frontend": "react",
"backend": "django",
"database": "postgresql",
"cache": "redis",
"search": "elasticsearch"
},
"extras": {
"prp": true,
"validation": true,
"docker": true,
"cicd": true,
"testing": true
}
}
# Run with config
$ context-forge init --config context-forge.json --output ecommerce-platform
You're building an MVP for a startup. You need to move fast but maintain quality:
# 1. Initialize with MVP timeline
context-forge init --preset startup-mvp
# 2. Review generated Implementation.md
cat Docs/Implementation.md
# Shows 2-week sprint plan with daily tasks
# 3. Start development with Claude Code
# Open CLAUDE.md in Claude Code
# Follow Stage 1 tasks systematically
# 4. Validate progress daily
context-forge validate --levels syntax,tests
Migrating a legacy system to modern stack:
# 1. Create detailed PRD first
context-forge init --prd legacy-migration-prd.md
# 2. Select enterprise options
# - Choose 'enterprise' timeline
# - Enable all validation options
# - Select comprehensive testing
# 3. Generated structure includes:
# - Migration strategy in Implementation.md
# - Rollback procedures
# - Comprehensive test suites
# - Security audit checklists
24-hour hackathon, need to prototype fast:
# 1. Quick setup
npx context-forge init --preset hackathon
# 2. Skip optional features
# - Disable PRP (too detailed for hackathon)
# - Basic validation only
# - Focus on core features
# 3. Get building in < 5 minutes
# CLAUDE.md has simplified rules for rapid development
NEW in v3.1.3+: The analyze command allows you to retrofit existing codebases with AI-optimized documentation. This is perfect for:
# Run in your existing project directory
cd /path/to/your/project
context-forge analyze
# Specify output directory
context-forge analyze --output ./ai-docs
# Target specific IDEs
context-forge analyze --ide claude,cursor
# Skip AI analysis for faster setup
context-forge analyze --no-ai
📁 Generated Files:
├── CLAUDE.md (UPDATED - appended retrofit section)
├── Docs/
│ ├── Implementation.md
│ ├── project_structure.md
│ ├── UI_UX_doc.md
│ └── Bug_tracking.md
└── PRPs/
├── user-authentication-prp.md
├── payment-integration-prp.md
└── api-v2-prp.md
The analyze command is intelligent enough to:
NEW in v3.1.4: The migrate command helps you safely transition between technology stacks with comprehensive planning, risk analysis, and rollback strategies.
Perfect for:
# Interactive migration planning
context-forge migrate
# Specify target technology
context-forge migrate --target "Next.js"
# Quick mode (skip detailed analysis)
context-forge migrate --quick --target "FastAPI"
# Analyze only (no file generation)
context-forge migrate --analyze-only
🔄 Migration Analysis Complete!
Source: Flask + Jinja2 + SQLAlchemy
Target: Next.js + FastAPI + Prisma
Complexity: HIGH (Score: 78/100)
Recommended Strategy: parallel-run
📁 Generated Migration Files:
├── CLAUDE.md (Migration-aware configuration)
├── PRPs/
│ ├── migration-overview.md
│ ├── phase-1-infrastructure.md
│ ├── phase-2-api-migration.md
│ ├── phase-3-frontend-migration.md
│ └── rollback-procedures.md
└── .claude/
├── commands/migration/
│ ├── migration-status.md
│ ├── migration-checkpoint.md
│ └── migration-rollback.md
└── hooks/
├── MigrationCheckpoint.py
└── DataIntegrityCheck.py
Context Forge creates IDE-specific documentation structures:
project-folder/
├── CLAUDE.md # Main context file with tech-stack specific rules
├── Docs/
│ ├── Implementation.md # Staged development plan
│ ├── project_structure.md # Folder organization
│ ├── UI_UX_doc.md # Design specifications
│ └── Bug_tracking.md # Bug tracking template
├── PRPs/ # Product Requirement Prompts (if enabled)
│ ├── {project}-prp.md # Base implementation PRP
│ ├── {project}-planning.md # Architecture planning (if enterprise/team)
│ └── {feature}-prp.md # Feature-specific PRPs (if retrofit)
├── .claude/ # Claude Code specific configuration
│ ├── commands/ # Slash commands (20+ commands)
│ │ ├── PRPs/ # PRP management commands
│ │ ├── development/ # Development workflow commands
│ │ ├── checkpoints/ # Human-in-the-loop commands
│ │ └── migration/ # Migration commands (if applicable)
│ └── hooks/ # Claude Code hooks (if enabled)
│ ├── PreCompact.py # Context preservation hook
│ ├── ContextRotation.py # Smart file switching
│ ├── PreSubmit.py # Quality gates
│ ├── PRPTracking.py # Progress monitoring
│ ├── DartProgressUpdater.py # Dart task progress tracking
│ ├── AutoTaskCommenter.py # Automated task documentation
│ └── TaskCodeMapper.py # File-to-task mapping
└── ai_docs/ # AI documentation curation (if enabled)
└── README.md # Documentation guide
project-folder/
├── .cursorrules # Main Cursor rules file
└── .cursor/
└── rules/
├── global.md # Global development rules
├── project.md # Project-specific rules
├── prp-overview.mdc # PRP implementation overview
├── prp-stage-1.mdc # Foundation setup tasks
├── prp-stage-2.mdc # Core feature implementation
├── prp-stage-3.mdc # Advanced features & polish
└── prp-validation.mdc # Validation gates & checks
project-folder/
└── .clinerules/ # Configuration directory
├── README.md # Main configuration
├── context.md # Project context
├── rules.md # Development rules
├── patterns.md # Code patterns
├── prp-overview.md # PRP implementation overview
├── prp-stage-1.md # Foundation setup tasks
├── prp-stage-2.md # Core feature implementation
├── prp-stage-3.md # Advanced features & polish
└── prp-validation.md # Validation gates & checks
project-folder/
├── GEMINI.md # Main Gemini configuration
└── .gemini/
├── context/
│ ├── project.md # Project context
│ ├── architecture.md # Architecture docs
│ └── guidelines.md # Development guidelines
├── prp/ # PRP files (if features defined)
│ ├── overview.md # PRP implementation overview
│ ├── stage-1-foundation.md
│ ├── stage-2-core.md
│ ├── stage-3-advanced.md
│ └── validation.md
└── config.yaml # Gemini config with PRP settings
The main context file that Claude Code reads first. It contains:
Your roadmap with staged development:
Each stage contains:
Advanced prompts for complex implementations:
When you run context-forge init, you'll be guided through:
Context Forge includes a powerful validation system that ensures code quality:
# Run all critical validations
context-forge validate
# Run specific validation levels
context-forge validate --levels syntax,tests
# Run all validations including optional ones
context-forge validate --all
# Generate detailed report
context-forge validate --report
# Validate in specific directory
context-forge validate --path ./my-project
| Level | Description | Critical | Commands |
|---|---|---|---|
| syntax | Type checking and linting | ✅ Yes | tsc, eslint, mypy, etc. |
| tests | Unit and integration tests | ✅ Yes | jest, pytest, rspec, etc. |
| coverage | Code coverage analysis | ❌ No | With coverage reporters |
| build | Production build | ✅ Yes | Framework-specific build |
| security | Vulnerability scanning | ❌ No | npm audit, safety, etc. |
📊 Validation Report - Analytics Dashboard
==================================================
✅ Status: PASSED
Total: 5 | Passed: 5 | Failed: 0
✅ syntax:tsc --noEmit (342ms)
✅ syntax:eslint src --ext .ts,.tsx (567ms)
✅ tests:jest --coverage (4521ms)
✅ build:next build (8234ms)
✅ security:npm audit (1023ms)
Full report saved to: .validation-reports/latest-report.md
Skip interactive prompts with a configuration file:
{
"projectName": "My App",
"projectType": "fullstack",
"description": "A collaborative platform",
"techStack": {
"frontend": "nextjs",
"backend": "fastapi",
"database": "postgresql",
"auth": "jwt"
},
"features": [
{
"id": "auth",
"name": "Authentication",
"priority": "must-have"
}
],
"extras": {
"prp": true,
"validation": true,
"docker": true
}
}
Use with: context-forge init --config context-forge.json
# Run the init command
context-forge init
# Answer the interactive prompts
# Context Forge will analyze your requirements and generate appropriate documentation
# 1. Start with CLAUDE.md
cat CLAUDE.md
# This is your project's "constitution" - rules Claude Code will follow
# 2. Check your implementation plan
cat Docs/Implementation.md
# This breaks down development into manageable stages
# 3. Review project structure
cat Docs/project_structure.md
# This defines how your code should be organized
# After completing each stage
context-forge validate
# Before committing code
context-forge validate --levels syntax,tests
# For comprehensive check
context-forge validate --all --report
Create your own templates in ~/.context-forge/templates/:
# Custom tech stack template
~/.context-forge/templates/claude/my-stack.md
# Custom validation commands
~/.context-forge/templates/validation/my-validation.json
Add pre/post hooks in .context-forge/hooks/:
# pre-init.sh - Run before initialization
#!/bin/bash
echo "Setting up environment..."
# post-init.sh - Run after initialization
#!/bin/bash
echo "Installing dependencies..."
npm install
# .github/workflows/validate.yml
name: Context Forge Validation
on: [push, pull_request]
jobs:
validate:
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v3
- uses: actions/setup-node@v3
- run: npm install -g context-forge
- run: context-forge validate --all
# Clone the repository
git clone https://github.com/webdevtodayjason/context-forge.git
cd context-forge
# Install dependencies
npm install
# Build the project
npm run build
# Run in development mode
npm run dev
# Run tests
npm test
# Run linting
npm run lint
context-forge/
├── src/
│ ├── cli/ # CLI entry point and commands
│ ├── commands/ # Command implementations
│ ├── generators/ # Document generators
│ ├── templates/ # Handlebars templates
│ ├── data/ # Tech stack configurations
│ └── types/ # TypeScript definitions
├── templates/ # Document templates
└── tests/ # Test files
This project was inspired by and built upon the work of amazing developers:
Contributions are welcome! Please read our Contributing Guide for details on our code of conduct and the process for submitting pull requests.
This project is licensed under the MIT License - see the LICENSE file for details.
Made with ❤️ by the Context Forge community
Empowering developers to build smarter, not harder
FAQs
AI orchestration platform with autonomous teams, enhancement planning, migration tools, 25+ slash commands, checkpoints & hooks. Multi-IDE: Claude, Cursor, Windsurf, Cline, Copilot
The npm package context-forge receives a total of 173 weekly downloads. As such, context-forge popularity was classified as not popular.
We found that context-forge demonstrated a healthy version release cadence and project activity because the last version was released less than a year ago. It has 1 open source maintainer collaborating on the project.
Did you know?
Socket for GitHub automatically highlights issues in each pull request and monitors the health of all your open source dependencies. Discover the contents of your packages and block harmful activity before you install or update your dependencies.
Security News
Multiple high-impact npm maintainers confirm they have been targeted in the same social engineering campaign that compromised Axios.
Security News
Axios compromise traced to social engineering, showing how attacks on maintainers can bypass controls and expose the broader software supply chain.
Security News
Node.js has paused its bug bounty program after funding ended, removing payouts for vulnerability reports but keeping its security process unchanged.