Prism

Prism is a free and open-source, self-hosted web agent. It supports multiple LLM providers, GitHub integration, and secure sandboxed tool execution.

Features

• Multi-Provider LLM Support: OpenAI, Anthropic, Google AI, OpenRouter, Groq, DeepSeek, and Ollama

• Secure by Design: AES-256 encrypted API key storage, JWT authentication, sandboxed Docker execution

• GitHub Integration: OAuth-based repository access for code operations

• File Uploads: Support for images, PDFs, and code files

• Configurable Tool Approval: Auto-execute by default, with per-tool approval settings

• Mobile Responsive: Access from any device through your hosted instance

• API Access: Generate API keys for external client access

• Modern UI: ChatGPT/Claude-like chat interface with dark/light themes

Quick Start

Prerequisites

• Docker and Docker Compose
• Go 1.20+ (for local development)
• Node.js 20+ (for local development)

Installation

1. Clone the repository:

   git clone https://github.com/jackulau/Prism.git
   cd prism

2. Copy the environment file:

   cp .env.example .env

3. Generate an encryption key:

   openssl rand -hex 32

   Add this to your .env file as ENCRYPTION_KEY.

4. Start the application:

   make prod

5. Open http://localhost:3000 in your browser.

Development

For local development with hot reload:

# Install dependencies
make setup

# Start development servers
make dev

Or run components separately:

# Terminal 1: Backend
make run-backend

# Terminal 2: Frontend
make run-frontend

Configuration

Environment Variables

Variable	Description	Default
PORT	Backend server port	8080
DATABASE_URL	SQLite database path	./data/prism.db
ENCRYPTION_KEY	32-byte hex key for encrypting API keys	(required)
JWT_SECRET	Secret for JWT tokens	change-me-in-production
GITHUB_CLIENT_ID	GitHub OAuth App client ID	(optional)
GITHUB_CLIENT_SECRET	GitHub OAuth App client secret	(optional)
OLLAMA_HOST	Ollama API endpoint	http://localhost:11434

LLM Providers

Users configure their own API keys through the Settings UI. Prism supports a wide range of providers:

Provider	Models	Tools	Vision	Best For
OpenAI	GPT-4.1, o3, o4-mini	Yes	Yes	General purpose, reasoning
Anthropic	Claude Opus/Sonnet/Haiku 4.5	Yes	Yes	Coding, analysis
Google AI	Gemini 2.5 Flash/Pro	Yes	Yes	Large context (1M tokens)
OpenRouter	200+ models	Varies	Varies	Access to all models
Groq	Llama 3.3, Mixtral	Yes	Some	Ultra-fast inference
DeepSeek	DeepSeek V3, R1, Coder	Yes	No	Cost-effective reasoning
Ollama	Any local model	Some	Some	Privacy, no API costs

Quick Setup

1. Click the Settings icon in the sidebar
2. Select a provider and enter your API key
3. Choose a model from the dropdown

For detailed setup instructions and model recommendations, see Provider Documentation.

GitHub OAuth Setup

1. Go to GitHub Settings > Developer Settings > OAuth Apps
2. Create a new OAuth App
3. Set the callback URL to http://your-domain/api/v1/github/callback
4. Add the Client ID and Secret to your .env file

Architecture

prism/
├── backend/           # Go backend (Fiber)
│   ├── cmd/server/    # Entry point
│   └── internal/
│       ├── api/       # HTTP handlers & WebSocket
│       ├── llm/       # LLM provider implementations
│       ├── tools/     # Tool execution & sandbox
│       ├── github/    # GitHub integration
│       └── security/  # Crypto, JWT, auth
├── frontend/          # React + TypeScript
│   └── src/
│       ├── components/  # UI components
│       ├── hooks/       # React hooks
│       ├── services/    # API & WebSocket clients
│       └── store/       # Zustand state
└── sandbox/           # Docker sandbox images

Security

• API Keys: Encrypted at rest with AES-256-GCM
• Passwords: Hashed with Argon2id
• Sessions: JWT with 15-minute access tokens
• Tool Execution: Isolated Docker containers with:
  • Memory limits (512MB default)
  • CPU limits (0.5 cores default)
  • Timeout (60 seconds default)
  • Network isolation
  • Read-only root filesystem

API

WebSocket

Connect to /api/v1/ws?token=<access_token> for real-time chat streaming.

REST Endpoints

• POST /api/v1/auth/register - Register
• POST /api/v1/auth/login - Login
• GET /api/v1/conversations - List conversations
• POST /api/v1/chat/completions - Send chat message (non-streaming)

Contributing

Contributions are welcome! Please read our Contributing Guide first.

License

Apache License 2.0 - see LICENSE for details.

Support

• GitHub Issues
• Discussions