Installing Community Edition.
Community Edition installs like real self-hosted software, not a hosted-only product with a token local mode. Prerequisites, a containerized quick-start path, and a verification step to confirm the control library and AI inference connection are both working.
The install path
- Prerequisites
A supported Linux host with a container runtime, and an AI inference endpoint you control — that endpoint is the one genuinely external dependency, because Community Edition ships with no bundled provider.
- Containerized quick-start
One compose-style definition stands up the application, its datastore, and the initial admin access. No compiled artifacts to hunt down, no hidden phone-home step between you and a running instance.
- Verification
The quick-start ends with an explicit check: the control library loads its framework packs, and the configured inference endpoint answers a test request. An install is not finished until both are confirmed.
- First-run configuration
From a verified running instance to your organization: entities, users, roles, and your first integrations — the next documentation page.
Two design decisions shape the path. The inference endpoint is required up front, at install time, rather than patched in later — because the governance workflows are the product, and they call that endpoint from the first run; an install that skipped it would be an install of nothing in particular. And the datastore stands up empty of assumptions: no seed content, no demo entities, nothing to clean out before the instance is yours.
The install path is deliberately boring. If installation is the interesting part of your governance platform, something has gone wrong — the interesting part is what runs afterwards. Upgrades follow the same philosophy: the same definition that stood the instance up moves it forward, with a changelog entry per release and no undocumented migration steps.
Continue to configuration