Introduction
Sevorix is a runtime containment layer for autonomous AI agents. Rather than inspecting only the prompt going into a model, it sits on the wire between an agent and the systems it acts on — the shell, the network, and (on Linux) the kernel — and decides, per action, whether that action is allowed.
Two things are documented on this site:
- Runtime — the enforcement product itself: the daemon,
sevsh, the policy engine, and the CLI surfaces around them. - Hub API — the REST API behind
hub.sevorix.com, covering accounts, signing keys, and artifact publishing.
Start here
To get Sevorix running on a machine, work through these in order:
- Install — requirements, then the quick (scripted) or manual install.
- Installer Prompts — what each of the installer's questions does, and what declining it costs.
- Start and Verify — log in, start the daemon, and confirm enforcement is actually on.
Where to go next
| If you want to… | Go to |
|---|---|
| Install and configure the enforcement layer | Getting Started → Install |
| Find where a file was installed | Runtime → File Locations |
| Fix a problem with an existing install | Runtime → Troubleshooting |
| Remove Sevorix | Runtime → Uninstall |
| Understand how policies are written and evaluated | Runtime → Policy Format |
| Look up a command | Runtime → CLI Reference |
| Call the hub from your own code | Hub API → Authentication |