3.3 KiB
Clone and run https://git.komnenos.me/springfw/random-proxy-simple in ./random-proxy-simple.
Goal: get the app running once on port 3001 with OpenAI and Anthropic Replit AI Integrations wired, and make it publishable.
Bearer token for this deployment: DEPLOYMENT_GENERATED_PROXY_TOKEN
Important: wire OpenAI and Anthropic via the ai-integrations-openai and ai-integrations-anthropic skills. Call setupReplitAIIntegrations directly. Do not call searchIntegrations. OpenAI and Anthropic are Replit AI Integrations that provide AI_INTEGRATIONS_* variables; they are not OAuth connectors.
Do not:
- add it to a parent workspace
- move the proxy repo into artifacts/
- modify source files
- edit package.json
- create config.json or /tmp config files
- inspect/free unrelated ports
- use nohup/setsid/background shell tricks
- call searchIntegrations
- search the Replit integrations catalog
- create provider API secrets manually
- use public OpenAI or Anthropic endpoints
- invent, rotate, print, store, or replace the bearer token
- run a foreground server command directly in the shell
- run verification commands
- run chat-completion smoke tests unless asked
- configure production deployment through the root .replit [deployment] section
- create or publish a kind = "api" artifact for this proxy
Steps:
-
Clone into ./random-proxy-simple.
-
Wire OpenAI and Anthropic via the ai-integrations-openai and ai-integrations-anthropic skills:
- call setupReplitAIIntegrations directly
- do not call searchIntegrations
-
From ./random-proxy-simple, run: pnpm install --ignore-workspace
-
Create the smallest persistent workflow named "Random Proxy Simple" with this exact run command:
cd /home/runner/workspace/random-proxy-simple && node build.mjs && PORT=3001 PROXY_API_KEY=DEPLOYMENT_GENERATED_PROXY_TOKEN node --enable-source-maps dist/index.mjs
-
Start/restart that workflow once.
-
Create a new react-vite artifact at previewPath = "/". Keep its artifact kind as "web"; this is the publishable signal Replit needs. Do not convert it to kind = "api".
-
Use verifyAndReplaceArtifactToml on that web artifact. Preserve the artifact's existing kind/title/id/integratedSkills metadata. Replace only the service production config so it builds and runs the proxy:
[services.production]
[services.production.build] args = ["sh", "-c", "cd random-proxy-simple && pnpm install --ignore-workspace && pnpm build:fast"]
[services.production.run] args = ["sh", "-c", "cd random-proxy-simple && PORT=3001 PROXY_API_KEY=DEPLOYMENT_GENERATED_PROXY_TOKEN NODE_ENV=production node --enable-source-maps dist/index.mjs"]
[services.env] PORT = "3001" PROXY_API_KEY = "DEPLOYMENT_GENERATED_PROXY_TOKEN" NODE_ENV = "production"
[services.production.health.startup] path = "/api/healthz"
-
If artifacts/mockup-sandbox exists, keep it as its own artifact and configure its production mode to serve a static Vite build from artifacts/mockup-sandbox/dist. Do not move or configure this proxy repo as mockup-sandbox.
-
Do not run health, dashboard, model-list, or chat-completion verification commands.
Report only:
- local URL
- whether setupReplitAIIntegrations wired OpenAI and Anthropic
- whether a kind = "web" artifact was configured with proxy production build/run commands
- that no verification commands were run