Clone and run https://git.komnenos.me/springfw/random-proxy-simple in ./random-proxy-simple. Goal: get the app running once, with Replit AI Integrations wired, on port 3000. Bearer token for this deployment: DEPLOYMENT_GENERATED_PROXY_TOKEN Use that exact token as the proxy bearer token in the run command. Do not invent, rotate, print, or replace it. Do not ask for another token. Do not: - add it to a parent workspace - move it 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 - search for or install integration skills/plugins/packages - create provider API secrets manually - use public OpenAI or Anthropic endpoints - run a foreground server command directly in the shell - set secrets or environment variables for the proxy token - run verification commands - run chat-completion smoke tests unless asked Steps: 1. Clone into ./random-proxy-simple. 2. Use Replit's built-in AI Integrations feature only. Do not look for skills, plugins, packages, npm modules, or marketplace integrations. Enable exactly these Replit AI Integration providers: - OpenAI - Anthropic After provisioning, rely on Replit to inject these variables automatically: - AI_INTEGRATIONS_OPENAI_BASE_URL - AI_INTEGRATIONS_OPENAI_API_KEY - AI_INTEGRATIONS_ANTHROPIC_BASE_URL - AI_INTEGRATIONS_ANTHROPIC_API_KEY 3. Create or update one persistent Replit workflow named "Random Proxy Simple" with this exact run command: cd random-proxy-simple && pnpm dev -- --proxy-api-key DEPLOYMENT_GENERATED_PROXY_TOKEN 4. Start/restart that workflow once. 5. Do not run health, dashboard, model-list, or chat-completion verification commands. Report only: - local URL - whether OpenAI and Anthropic AI Integrations were wired - that no verification commands were run