blacktroll-worker

Lend your CPU to the BlackTroll network and help run AI for everyone — no clone, no config, no GPU required.

Run it

npx blacktroll-worker

That's it. On first run it automatically installs Ollama, downloads a small model, and connects to the network. Leave the terminal open — your CPU is now a worker.

Requires: Node.js 20+ (a normal, signed installer).

Options

# weak/old PC — smaller model, cap CPU cores so your machine stays responsive
npx blacktroll-worker --model qwen2.5:0.5b --threads 2

# stronger PC — bigger, smarter model
npx blacktroll-worker --model qwen2.5:7b

# point at a different network
npx blacktroll-worker --url https://api.blacktroll.meme

Run npx blacktroll-worker --help for all flags.

How it works

Your machine opens an outbound websocket to the network's orchestrator (so no firewall or port setup is needed), registers the model it can serve, and waits. When someone sends a prompt, the orchestrator routes it to your worker, your CPU generates the answer via Ollama, and the tokens stream back. The orchestrator stores nothing and never learns who the user is.