Expand description
usepod-agent setup — device-flow pairing on first run.
Replaces the v0.1.x ceremony of “enroll on the dashboard, copy a 40-char host_token, paste it into agent.toml”. The flow now:
- Generate (or load) the agent’s Ed25519 identity.
- Probe well-known local backend ports (vLLM :8000, llama.cpp :8080, LM Studio :1234, Ollama :11434).
- POST /v1/host/pair/issue to the coordinator with the agent pubkey; receive a short pair_code + poll_token.
- Print the pair_code prominently and start long-polling /v1/host/pair/poll, sending detected backends as capabilities so the dashboard can render the model picker live.
- When the operator hits Activate in the dashboard, the next poll response delivers host_token + provider_id + activated_models.
- Write a complete agent.toml from the discovered + operator- configured state. Operator never edits a file.