e-dream.ai worker

Running

There are two ways to use this project:

A) Local development (run worker locally)

1. nvm install

2. nvm use

3. npm install

4. Ensure Redis is available locally (for example on macOS: brew services start redis)

5. Create .env.

6. Build: npm run build

7. Start the worker and admin UI: npm start

8. In another terminal, run jobs using JSON files:

   • node dist/prompt.js prompt/deforum-fish.json (creates prompt/deforum-fish.mp4)
   • node dist/prompt.js prompt/animatediff-dog.json (creates prompt/animatediff-dog.mp4)
   • node dist/prompt.js prompt/uprez-example.json (creates prompt/uprez-example.mp4)
   • By default, output files are saved alongside the input JSON with a .mp4 extension
   • Use -o to specify a custom output path: node dist/prompt.js prompt/deforum-fish.json -o my-custom-name.mp4

Notes:

• The worker now always returns a presigned URL, and the CLI downloads to your machine.

Required env vars (local development)

• RUNPOD_API_KEY
• RUNPOD_ANIMATEDIFF_ENDPOINT_ID
• RUNPOD_HUNYUAN_ENDPOINT_ID
• RUNPOD_DEFORUM_ENDPOINT_ID
• RUNPOD_UPREZ_ENDPOINT_ID
• ADMIN_PASS (password for username admin in the admin UI)
• Redis: use local redis info (REDIS_HOST=localhost, REDIS_PORT=6379, REDIS_PASSWORD='').

B) Use deployed worker only (no local server)

1. Build once: npm run build
2. Point the CLI at the same Redis used by the worker. The CLI process needs a few env vars due to validation:
   • Required: REDISCLOUD_URL

node dist/prompt.js prompt/animatediff-dog.json -o ./out/dog.mp4
node dist/prompt.js prompt/uprez-example.json -o ./out/uprez.mp4
node dist/prompt.js prompt/deforum-fish.json -o ./out/deforum.mp4

How it works

• The worker returns { r2_url: <presigned-url> }.
• The local CLI downloads from the provided URL to your specified -o path (or the default path when -o is omitted).

Ops

1. Visit https://www.runpod.io/console/serverless/user/endpoint/89vs9h0qx0g966?tab=requests to see recent activity and cancel jobs you don't wish to keep running.
2. https://apps.apple.com/us/app/red-2-ui-for-redis/id1491764008?mt=12 is helpful for seeing what's in Redis
3. https://gpu-worker-0ac312b41451.herokuapp.com/admin/queues/ for the stage and https://gpu-worker-prod-8775591876d9.herokuapp.com/admin/queues for alpha Bull Dashboard (login the user "admin" and the password from the .env)

Marketing Email CLI (Worker)

Marketing sends are triggered from the worker/ project and processed by the marketing worker.

Prerequisites

• worker/.env must include:
  • BACKEND_URL (backend API base URL, e.g. https://api-stage.infinidream.ai/api/v1)
  • BACKEND_API_KEY (admin API key)
  • MARKETING_EMAIL_SECRET (shared internal secret)
  • MARKETING_CONCURRENCY (optional, default 50)
• Resend template must include variable UNSUBSCRIBE_URL in the template body
  • Example: <a href="{{UNSUBSCRIBE_URL}}">Unsubscribe</a>

Build worker

cd worker
pnpm run build

One-command send (starts worker + triggers send)

pnpm run marketing:send -- --template-id tpl_123 --dry-run

pnpm run marketing:send -- --template-id tpl_123 --limit 100 --offset 0

Send to one specific user:

pnpm run marketing:send -- --template-id tpl_123 --email user@example.com

or by user id:

pnpm run marketing:send -- --template-id tpl_123 --user-id 42

The command starts the marketing worker in-process and will auto-exit when the queue is empty. You can also stop it early with Ctrl+C.