A collaborative platform for managing and executing LLM prompts with intelligent agents, conversation management, and powerful tool integrations.
⚠️ Alpha Version - This project is currently in active development and not finalized yet. Features and APIs may change.
- 🤖 Intelligent Agents: Create chatbot and task agents with customizable behavior
- 💬 Smart Conversations: Automatic summarization reduces token usage by up to 70%
- 🛠️ Powerful Tools: Built-in tools for web search, calculations, API calls, and more
- 🎯 Default Providers: Pre-configured access to OpenAI, Anthropic, Google, and DeepSeek with latest models
- 📊 Organization Management: Multi-tenant structure with role-based access control
- 🔧 API Integration: Easy external API configuration with authentication
- ⚡ Performance Optimized: Token-efficient processing with intelligent model selection
- Node.js 18+
- MongoDB 4.4+
- An OpenAI API key (or other supported LLM provider)
# Clone the repository
git clone https://github.com/LLM-Crafter/llm-crafter.git
cd llm-crafter
# Install dependencies
npm install
# Configure environment
cp .env.example .env
# Edit .env with your configuration
# Start the server
npm run dev🎉 That's it! The application automatically initializes with default providers for OpenAI, Anthropic, Google, and DeepSeek, so you can start creating agents immediately.
We provide a JavaScript SDK for easy integration with the LLM Crafter API:
npm install @llm-crafter/sdkimport { LLMCrafterClient } from '@llm-crafter/sdk';
const client = new LLMCrafterClient(
'your-api-key',
'https://your-domain.com/api/v1'
);
// Execute a prompt
const result = await client.executePrompt(
'org-id',
'project-id',
'prompt-name',
{
variable: 'value',
}
);
// Chat with an agent
const chat = await client.startAgentChat('agent-id', 'Hello!');
console.log(chat.response.data.response);# Register a user
curl -X POST http://localhost:3000/api/v1/auth/register \
-H "Content-Type: application/json" \
-d '{"name": "John Doe", "email": "john@example.com", "password": "secure-password"}'
# Login and get token
curl -X POST http://localhost:3000/api/v1/auth/login \
-H "Content-Type: application/json" \
-d '{"email": "john@example.com", "password": "secure-password"}'
# Create organization, project, API key, and agent
# See the Getting Started guide for complete instructionsStart the documentation development server:
npm run docs:devThe documentation will be available at http://localhost:5173.
Build the documentation for production:
npm run docs:buildPreview the built documentation:
npm run docs:preview┌─────────────────┐ ┌─────────────────┐ ┌─────────────────┐
│ Frontend │ │ API Layer │ │ Database │
│ Applications │◄──►│ Express.js │◄──►│ MongoDB │
└─────────────────┘ └─────────────────┘ └─────────────────┘
│
┌─────────────────┐
│ Services │
│ - Agents │
│ - Tools │
│ - OpenAI │
│ - Summarize │
└─────────────────┘
- Node.js 18+
- MongoDB 4.4+
- Git
# Clone the repository
git clone hhttps://github.com/LLM-Crafter/llm-crafter.git
cd llm-crafter
# Install dependencies
npm install
# Start MongoDB
brew services start mongodb-community # macOS
sudo systemctl start mongod # Linux
# Configure environment
cp .env.example .env
# Edit .env with your settings
# Start development server
npm run dev# Required
MONGODB_URI=mongodb://localhost:27017/llm-crafter
JWT_SECRET=your-super-secret-key-min-32-chars
PORT=3000npm testnpm run lintLLM Crafter uses a hierarchical structure:
- Organizations: Top-level entities with team management
- Projects: Containers for related AI resources
- Agents: AI assistants and task workers
- Tools: Extend agent capabilities
- Stateful conversations with memory
- Automatic conversation summarization
- Multi-turn interactions
- Stateless task execution
- Single-purpose operations
- High-volume processing
Built-in tools include:
- Web Search: Search the internet for information
- Calculator: Mathematical calculations
- API Caller: HTTP requests to external APIs
- JSON Processor: Parse and manipulate JSON data
- Current Time: Get current date and time
# Local MongoDB
MONGODB_URI=mongodb://localhost:27017/llm-crafter
# MongoDB Atlas
MONGODB_URI=mongodb+srv://user:pass@cluster.mongodb.net/llm-crafter
# Replica Set
MONGODB_URI=mongodb://host1:27017,host2:27017/llm-crafter?replicaSet=rs0Configure default provider credentials:
OPENAI_API_KEY=sk-your-openai-key
ANTHROPIC_API_KEY=your-anthropic-keyOr manage API keys through the API for better security and organization.
# Build image
docker build -t llm-crafter .
# Run with Docker Compose
docker-compose up -d# Install PM2
npm install -g pm2
# Start application
pm2 start ecosystem.config.js --env production
# Monitor
pm2 monit# Production environment variables
NODE_ENV=production
PORT=3000
MONGODB_URI=mongodb://localhost:27017/llm-crafter-prod
JWT_SECRET=your-production-secret-key- Token Reduction: Up to 70% reduction in token usage
- Cost Savings: Significant reduction in LLM API costs
- Response Speed: 30-50% faster responses for long conversations
- Context Quality: Preserves important information while removing redundancy
LLM Crafter automatically selects cost-effective models:
- Production tasks:
gpt-4o - Development:
gpt-4o-mini - Summarization:
gpt-4o-mini - Reasoning tasks:
o1-mini
We welcome contributions! Please see our Contributing Guide for details.
- Fork the repository
- Create a feature branch
- Make your changes
- Add tests if applicable
- Run tests and linting
- Submit a pull request
We use ESLint and Prettier for code formatting:
npm run lint
npm run formatThis project is licensed under the MIT License - see the LICENSE file for details.
- Documentation: Coming soon
- GitHub: https://github.com/LLM-Crafter/llm-crafter
- Issues: https://github.com/LLM-Crafter/llm-crafter/issues
- Documentation: Complete guides and API reference
- GitHub Issues: Bug reports and feature requests
Built with ❤️ by the LLM Crafter team