feat(deploy): render blueprint for the autorouter sandbox gateway

Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>

deploy/autorouter-sandbox/README.md

@@ -22,4 +22,6 @@ Without Docker: `uv run --no-sync litellm --config deploy/autorouter-sandbox/con
 
 ## Deploy
 
+Render: New > Blueprint, pick this repo and branch. The root `render.yaml` builds this Dockerfile, prompts for `UPSTREAM_LITELLM_API_KEY`, and generates `LITELLM_MASTER_KEY` (read it from the service's Environment tab). To use the sandbox DB natively instead of proxying, add `DATABASE_URL`, `LITELLM_SALT_KEY` and `LITELLM_LICENSE` in the Environment tab
+
 `.github/workflows/deploy-autorouter-sandbox.yml` builds and pushes `ghcr.io/<owner>/litellm-autorouter-sandbox:{sha,latest}` on every push to `litellm_autorouter_sandbox_deploy`. If the repo secret `AUTOROUTER_SANDBOX_KUBECONFIG_B64` is set (base64 kubeconfig; optional repo variable `AUTOROUTER_SANDBOX_NAMESPACE`), it also applies `k8s.yaml` and rolls the deployment to the new image. Create the env secret once with the command at the top of `k8s.yaml`

render.yaml

@@ -1,12 +1,19 @@
 services:
   - type: web
-    name: openai-proxy
-    runtime: image
-    image:
-      url: ghcr.io/berriai/litellm:main-stable
+    name: litellm-autorouter-sandbox
+    runtime: docker
+    plan: standard
+    region: oregon
+    autoDeploy: true
+    dockerfilePath: ./deploy/autorouter-sandbox/Dockerfile
+    dockerContext: .
+    healthCheckPath: /health/liveliness
     envVars:
       - key: PORT
-        value: 4000
-    numInstances: 1
-    healthCheckPath: /health/liveliness
-    autoDeploy: true
+        value: "4000"
+      - key: UPSTREAM_LITELLM_BASE_URL
+        value: https://gateway.litellm-sandbox.ai
+      - key: UPSTREAM_LITELLM_API_KEY
+        sync: false
+      - key: LITELLM_MASTER_KEY
+        generateValue: true