Genbase

A modular platform for specialized AI agents that collaborate to solve complex problems.

What is Genbase?

Genbase is an open platform that enables AI-powered collaboration through specialized modules. Instead of building one agent that knows a little about everything, Genbase allows you to combine domain-specific agents that each bring deep expertise to their area.

Key Features

• Specialized Modules: Package domain expertise into modules with purpose-built agents
• Collaborative Workflows: Agents work together through well-defined relationships
• Git-Based Workspaces: Agents operate on real repositories with full filesystem access
• Extensible Architecture: Add new capabilities by creating or combining modules

Why Genbase?

Traditional AI assistants try to know everything but often lack depth in specialized areas. Genbase takes a different approach:

┌─────────────────┐   ┌─────────────────┐   ┌─────────────────┐
│  Next.js Expert │   │PostgreSQL Expert│   │ Docker Expert   │
│                 │◄──►                 │◄──►                 │
└─────────────────┘   └─────────────────┘   └─────────────────┘

• Web Developer Module: Deep knowledge of frameworks, components and best practices
• Database Module: Expertise in schema design, query optimization and administration
• Deployment Module: Specialized in containerization, networking and cloud services

Modules collaborate through actions, sharing context, and workspace access - creating solutions no single agent could provide alone.

Getting Started

Running with Docker Compose (Recommended)

1. Clone the repository:

   git clone https://github.com/genbase-ai/genbase.git
   cd genbase

2. Copy the environment templates:

   cp docker/.env.template docker/.env
   cp engine/.env.template engine/.env
   cp studio/.env.template studio/.env

3. Edit the .env files with your credentials (at minimum, add an LLM API key):

   nano engine/.env

4. Make scripts executable:

   chmod +x scripts/*.sh

5. Start Genbase:

   ./scripts/docker-run.sh up

6. Access the applications:

   • Studio: http://localhost:5173
   • Engine API: http://localhost:8000

Running Locally

1. Make sure you have Python 3.11+ and Node.js 18+ installed.
2. Follow steps 1-3 above.
3. Run the local script:

   ./scripts/run-local.sh

Quick Demo

1. Create a new project
2. Add the Next.js Web Developer module from the registry
3. Start the development workflow
4. Ask the specialized agent to create a feature
5. Watch as it generates code, runs tests, and explains its approach

Building and Deployment

Building for Production

To build both applications for production:

./scripts/build.sh

Docker Commands

• Start all services: ./scripts/docker-run.sh up
• Stop all services: ./scripts/docker-run.sh down
• View logs: ./scripts/docker-run.sh logs
• Rebuild images: ./scripts/docker-run.sh build
• Restart services: ./scripts/docker-run.sh restart

Key Environment Variables

Engine (FastAPI)

The Engine supports many environment variables. Key ones include:

# LLM API Keys (At least one is required)
ANTHROPIC_API_KEY=your_anthropic_api_key_here
OPENAI_API_KEY=your_openai_api_key_here
GROQ_API_KEY=your_groq_api_key
MISTRAL_API_KEY=your_mistral_api_key
DEEPSEEK_API_KEY=your_deepseek_api_key

# Core settings
REGISTRY_URL=http://localhost:5173
DATA_DIR=".data"
ADMIN_USERNAME="admin"
ADMIN_PASSWORD="your_secure_password_here"
DATABASE_URL="postgresql://user:password@host/database?sslmode=require"

# Additional settings
LOG_LEVEL="INFO"
API_HOST="0.0.0.0"
API_PORT="8000"
# ... and many more (see engine/.env.template)

Studio (React/Vite)

# Core settings
VITE_ENGINE_URL="http://localhost:8000"
VITE_API_USERNAME="admin"
VITE_API_PASSWORD="your_secure_password_here"

Note: When running with Docker Compose, the Engine container will use its own .env file mounted as a volume, allowing for runtime configuration changes without rebuilding the container.

Testing and Troubleshooting

Quick Test Guide

1. Check environment files

   • Verify .env, engine/.env, and studio/.env exist and contain the right variables

2. Test the applications

   • Engine API: Make API calls to http://localhost:8000
   • Studio UI: Open http://localhost:5173 in your browser
   • Try logging in with the configured admin credentials

3. Common troubleshooting

   • View logs: ./scripts/docker-run.sh logs
   • Restart services: ./scripts/docker-run.sh restart
   • Check container status: docker ps -a

Documentation

• Getting Started Guide
• Core Concepts
• Architecture
• Creating Modules

Contributing

We welcome contributions! See CONTRIBUTING.md for details on how to get started.

License

Genbase is licensed under the Apache 2.0 License - see the LICENSE file for details.