One API for everything.
Create sessions, send input and stream events. Use the TypeScript SDK, the React components or plain HTTP.
For developers
Your product. Agents inside.
Build the experience that makes you different. Opengeni runs sessions, tools and sandboxes underneath.
How it works
Your app calls one API. Opengeni does the rest.
Work that keeps going.
Long tasks survive restarts and deploys, and pick up exactly where they stopped.
Your team sees it all.
Sessions your product starts show up in the Opengeni app, so teammates can follow and review them.
Run it
From your laptop. To your infrastructure.
The whole stack. One launcher.
Install Bun and Docker, add a model key, and run one command. It starts everything, including the web app your team uses.
Local development
git clone https://github.com/Cloudgeni-ai/opengeni.git
cd opengeni
cp .env.example .env
# Configure model credentials in .env
bun run devOpen the app at the URL printed by the launcher. Ports are selected per checkout.
Compute
Run where the work belongs.
Integration
Your interface. Our execution layer.
@opengeni/sdk/chatAdd agent chat
A server-side chat handler with product, tenant, user, and conversation mapping.
@opengeni/sdkControl the full session
A TypeScript client for sessions, workspaces, tools, files, and event streams.
@opengeni/reactEmbed a session experience
Session hooks, a composer, and timeline surfaces you can style for your product.
Open source
Contributing guide Read it. Run it. Make it better.
Apache 2.0, down to the deployment tooling. The same code runs the managed service.
Ship your firstagent feature.
Start free, add one handler to your backend, and let your team follow what it does.