Provider Client Introduction¶
The Flockadoo Provider Client is a lightweight service that runs next to a local LLM engine (vLLM, vLLM Metal, or llama.cpp). It reports health to Flockadoo, registers models and pricing for your flock, and forwards inference requests to that engine.
The setup wizard installs the engine if it is missing, picks an approved model that fits this machine, and writes config. You do not install the engine yourself.
Start here if you have not installed yet:
What it does¶
- Report provider health (GPU, models) to the Flockadoo orchestrator
- Register models and pricing with the platform
- Accept inference requests from the orchestrator and proxy them to your local LLM server
- Open an outbound connection to Flockadoo so the platform can send inference requests back to your machine without inbound firewall ports
How Flockadoo reaches your machine¶
The client opens an outbound connection to Flockadoo when it starts, and reconnects if that connection drops. Flockadoo sends inference requests back over that connection.
- You do not install another program.
- You do not open inbound firewall ports.
- You do not set a public URL. The address is the platform URL already in your config.
After first run¶
The client publishes your available models and sets initial costs from platform averages. Log in to the dashboard and adjust model pricing per flock if needed — see Model pricing.