# Host.Rodeo

> Agent-first LLM gateway — one OpenAI-compatible token, auto-routed to the fit-for-purpose source (free -> local/customer-owned -> paid), always fulfilled, with a signed receipt on every call showing what it would have cost elsewhere.

Point any OpenAI-compatible client at `https://api.host.rodeo/v1` with `model: "auto"`. Do not name a model.

## Wire it up

- [AGENTS.md](https://api.host.rodeo/AGENTS.md): drop this in your repo — everything a coding agent needs, and nothing it could work out for itself.
- [Live contract](https://api.host.rodeo/v1/contract): what the ranch can serve right now, machine-readable.
- [OpenAPI](https://api.host.rodeo/v1/openapi.json): generated from that same contract — every modality, tools, and both wire protocols.
- [One command](https://api.host.rodeo/v1/init.py): `curl -fsSL https://api.host.rodeo/v1/init.py | python3 - --key <your key>` wires an app end to end — config, a client, and one real call. `https://api.host.rodeo/v1/init` says what it does first.
- [Get a key](https://host.rodeo): sign in and connect an app.

## What keeps working

Every request word this gateway has ever accepted keeps parsing. We remove words from the documentation, never from the parser — so if something you already send is no longer mentioned here, it still works and you need change nothing. `https://api.host.rodeo/v1/contract` says it in full under `compat_covenant`.