Missing or unclear documentation for Ollama model usage (`generic.` prefix)

[SipkeW]

Description

The documentation currently does not explain how to configure and use local Ollama models with fast-agent.

Through experimentation, I discovered that the correct way to reference an Ollama model is by using the generic. prefix in the default_model field of fastagent.config.yaml.

For example:

default_model: generic.llama3

This works out of the box, even without a models: section. However, this behavior is undocumented in:

The README

The output of fast-agent setup

The comments in the generated config file

Steps to reproduce

Run fast-agent setup

Try to use llama3 or qwen3:8b via Ollama

Use --model=qwen3:8b or default_model: qwen3:8b — this fails

Use --model=generic.qwen3:8b or default_model: generic.qwen3:8b — this works, but isn’t documented

(i'm using qwen3:8b, but the same goes for any running ollama model)

Suggested fix

Please update the documentation to:

Mention the generic. prefix explicitly as the provider for Ollama

Provide working examples for default_model, commented out sections for different providers

Clarify that no models: block is required in this case

Thanks for the great project!