Deploy
Ship a service to real Kubernetes via GitOps — environments, per-PR preview environments, and variables/secrets. Forge-native.
The Deploy section is the runtime tier of the platform. It takes a service’s repo and turns it into a running app on real Kubernetes, with sane defaults for the 90% case and full K8s manifests underneath when you need them.
In the app, Deploy is a scope-aware section: pick a service from the top-bar scope picker and Deploy’s tabs — Services, Deployments, Environments, Variables, Previews, Activity — all narrow to it.
What you get out of the box
- Push-to-deploy from the forge. Start a deploy on any repo, review the auto-generated
k8s/Kustomize tree, merge the PR — your app is live in ~30 seconds athttps://<app>.<owner>.proxifai.appwith TLS. - Environments. Each service has named targets —
dev,qa,preprod,prod— each wired to a deployments (GitOps) repo, optional approval gates, and promotion rules. See Environments. - Preview environments per PR. Every open PR can get its own ephemeral environment with an optional clone of the production database, torn down on PR close. See Previews.
- Variables & secrets. Config managed per service as a key × environment matrix, with org-wide defaults inherited unless overridden. See Variables.
- GitOps is the source of truth. Every change — a UI edit, a CLI command, or a raw
git pushto the deployments repo — converges on the same reconciler and applies in seconds.
How it differs from a typical PaaS
| proxifai Deploy | Generic PaaS | |
|---|---|---|
| Where deploy config lives | Standard K8s manifests in your repo | Proprietary YAML |
| Edits | UI/CLI generate Kustomize patches via PR or commit | UI mutates server-side state |
| Power-user escape | kubectl + Helm + CRDs (via Infrastructure) | None |
| Forge integration | First-class (issues / PRs / agents) | None or weak |
| Preview environments | Per-PR with optional DB cloning | Per-PR (some) |
Get started
pfai deploy from a fresh clone — live on Kubernetes in 30 seconds.
dev → prod targets, approval gates, and promotion.
Ephemeral per-PR environments with optional DB cloning.
Per-service config across every environment.
The tiny .proxifai/config.yaml metadata file.
Databases, Kafka, cache, and isolated clusters.