Your agent. Your infra. Your data.
LFG is fully open source. Run the entire AI product-development agent on your own server with Docker, bring your own model and API key, and keep every byte of your data in-house. No vendor lock-in.
Quick start
Up and running in one command
The default setup uses SQLite and local file storage, so there's nothing external to provision for local dev.
# Clone and install (needs Bun) git clone https://github.com/lfg-hq/lfg.git && cd lfg bun install # Configure: set BETTER_AUTH_SECRET + one AI provider key cp example.env .env # Start (runs migrations, serves on :3000) bun run dev
Sandboxes
Run builds on your own machines
Every app the agent builds runs in an isolated sandbox. Choose where that compute lives.
Docker (self-hosted)
Each sandbox is an isolated Docker container on your own server, from a prebuilt image with Node, Python, and the coding agents baked in. No third-party compute, your code never leaves your box.
Managed VMs (hosted)
Prefer not to manage compute? Point LFG at hosted Firecracker micro-VMs and get public preview URLs out of the box. Flip a single env var to switch backends.
Bring your own model
Works with every major model
Use your own API keys. Switch models per task, cost, or capability, no provider lock-in.
Claude
GPT
Gemini
Grok
DeepSeek
Kimi
GLM
Anthropic · OpenAI · Google · xAI · DeepSeek · Moonshot · Z.ai
Your database
SQLite out of the box, or point it at your own Postgres. Your data lives where you put it.
Your files
Store uploads on the local disk, or plug in S3. One env var flips between them.
Your keys
Secrets are encrypted at rest and stay on your infrastructure. Nothing phones home.
Prefer we run it for you?
Don't want to manage the infrastructure? We offer fully managed, dedicated LFG instances, hosted, scaled, and maintained by us, with your models and isolated data. Tell us what you need and we'll set you up.
Custom deployments, SSO, SLAs, and volume pricing available on request.