LLM providers - MUTAGENT

An LLM provider holds the model key a workspace uses (bring your own key). The Python SDK lists, adds and tests LLM providers. The calls are on client.llm_providers and take your API key. Request bodies and responses are typed models from mutagent.models. This page requires mutagent-sdk 0.4.1. The key you send is stored encrypted, and no call returns it.

import os

from mutagent import Mutagent
from mutagent.models import CreateProviderConfigRequest, LlmProviderType

client = Mutagent(api_key=os.environ["MUTAGENT_API_KEY"], server_url="https://api.mutagent.io")

created = client.llm_providers.create_provider_config(
    CreateProviderConfigRequest(
        name="OpenAI", provider=LlmProviderType.OPENAI, api_key=os.environ["OPENAI_API_KEY"]
    )
)

providers = client.llm_providers.list_provider_configs()
print([config.name for config in providers.configs])

result = client.llm_providers.test_provider_config(created.id_)
print(result.success, result.message)

Call	What it does
`create_provider_config`	Adds an LLM provider: `name`, `provider` (an `LlmProviderType`), and the key as `api_key`. The key is tested with the LLM provider first. If the test fails, the call raises the provider’s error and nothing is saved.
`list_provider_configs`	Lists the workspace’s LLM providers as `configs`, with `total`.
`test_provider_config`	Calls the LLM provider with the stored key and returns `success` and `message`.

The same work from the command line: mutagent providers.