# Record Ollama

Canonical: https://hypaware.ai/docs/cli-reference/record-ollama

<!-- @ref LLP 0474#setup [explains]: setup checks readiness and prints routes without loading a model; verify needs a fresh saved request and response -->
Plugin: `@hypaware/ollama`, added when you choose **Ollama** in `hyp setup`.
See the [Ollama integration](https://hypaware.ai/docs/integrations/ollama/index.md) for the workflow.

## `hyp ollama setup`

```text
hyp ollama setup [--upstream URL] [--json]
```

Checks the direct Ollama service and its model inventory and prints the routes
to use for your next CLI or SDK launch. It never starts Ollama, downloads or
loads a model, or sends inference. `--upstream` saves a custom service address
in place of the default `http://127.0.0.1:11434`. A failure, or a daemon restart
that failed or could not be confirmed, returns `1`; invalid arguments return `2`.

## `hyp ollama verify`

```text
hyp ollama verify --model NAME [--json]
```

Sends one fixed check prompt to the named model. It succeeds only when a fresh
request and its completed response are saved and visible to your queries; any
other result returns `1`. A missing or invalid model name returns `2`.