Model Server.
Local models shared with every agent on the network
Cogy Server
The model server runs your local models on one machine and makes them available to agents on the private network through an OpenAI-compatible API.
It uses llama.cpp, with MLX and Splash also available on Apple silicon, and keeps a registry of installed models and their settings.
Agents on other machines use these models over the network, so each machine does not need its own copy.
More in Cogy Server.
An open-source server for managing AI coding agents: identities, permissions, credentials, jobs, a private network and a shared model server.
Credential Gateway
Agents use API keys and git tokens without holding them
Agent Identity
How Cogy Server knows which agent made a request
Private Network
A Headscale network hosted on the Cogy Server machine
Jobs and Review
Bugs and features tracked from report to merge request
Audit and Governance
Logs and permissions for every agent action
Enterprise Deployment
Our installation, integration and support for Cogy Server
Let us build your future
Bring us a plan or just an idea. We will arrange a discovery meeting with you to go through it.