OpenVoleThe self-hosted agent OS

Run a fleet of AI agents on your own hardware, against any model, from one dashboard. Peer-to-peer networked. Nothing phones home.

Get Started

Browse Paws

View on GitHub

🏠 Self-hosted & private

Your agents, your hardware, your data. Point a space at a local Ollama model and nothing ever touches the cloud.

🔌 Model-agnostic

One unified brain paw speaks Anthropic, OpenAI, Gemini, xAI, and Ollama. Switch providers per agent with a single env var.

🖥️ A server, not a script

vole serve is a control plane — create, start, stop, and chat with a whole fleet of agents from one browser dashboard. No processes to babysit.

🕸️ VoleNet

A peer-to-peer agent mesh. Remote tools become local, cheap workers share one brain, memory syncs — Ed25519-signed, no central server.

🧩 Microkernel, zero lock-in

A tiny, LLM-ignorant core. Every capability is a Paw you can swap, sandbox, or write yourself. Bring your own system prompt.

🪟 An app platform

A Paw can ship its own UI, rendered under the dashboard's Apps tab — the way apps live on a desktop. Extend the agent console into a full self-hosted app, no extra server, no LLM in the loop.

The OpenVole control plane managing a running space

One vole serve control plane managing a fleet of agents — each its own isolated space.

A dashboard you extend into a full app

A Paw can ship its own UI, rendered as a panel under the Apps tab — the way apps live on a desktop. The panel talks to its paw's own tools directly — no LLM, no extra port — so the dashboard grows from an agent console into a full self-hosted app.

The Apps tab — an embedded paw panel