---
name: coagent8-doctor
description: Show which models, reasoning efforts and output prices each CoAgent8 backend accepts, or check backend health (installation, sign-in, quota) through the CoAgent8 doctor tool. Use when the user asks which model CoAgent8 can use, gives a price limit, asks why a backend fails, or wants CoAgent8 status.
---

# CoAgent8 doctor

1. For models call `doctor` with `scope: models` (the default). It reads cached
   catalogs and lists, per backend, models, reasoning efforts, output prices,
   price freshness and each model's approval tag: `ask: <rule>`, `deny` or
   `allow`. Pass
   `backend` (`codex`, `claude` or `agy`) to narrow it and `detail: full` for the
   same report as JSON.
2. For problems call `doctor` with `scope: health`. It checks installation,
   sign-in and quota and takes longer than the models view.
3. To apply a user's price limit, pass `output_price_below` to consult or
   review and leave `model` unset; the server picks a model whose verified price
   is below the limit. Do not pick a model from the list to work around it.
4. A model tagged `ask` runs only after the user approves it in the form the
   server shows; tell the user its price and that the form will follow before
   you call it. You cannot approve it. A model tagged `deny` never runs: the user
   denied it, so do not call it. `allow` is the user's standing approval.

Do not edit CoAgent8 or client configuration files to change defaults or
approval: the running server may read a different configuration file than the
default path. For installation or registration problems point the user to
`coagent8 status` and `coagent8 repair`, or the VS Code command CoAgent8: Set Up
Coding CLIs.