Skip to main content
The providers section lets you define reusable provider configurations for any OpenAI-compatible API — without modifying docker-agent’s source code. This is useful for self-hosted models, API proxies, enterprise gateways, and any service that implements the OpenAI chat completions API.
If a service supports the /v1/chat/completions endpoint, you can use it with docker-agent. No source code changes are needed.

​
Provider configuration

Define a provider under the top-level providers key: 
providers:
  my_provider:
    api_type: openai_chatcompletions  # openai_chatcompletions | openai_responses
    base_url: https://api.example.com/v1
    token_key: MY_API_KEY             # name of the env var holding the token

​
Provider properties

PropertyDescriptionDefault
api_typeAPI schema: openai_chatcompletions or openai_responsesopenai_chatcompletions
base_urlBase URL for the API endpoint—
token_keyName of the environment variable containing the API token—

​
API types

  • openai_chatcompletions — Standard OpenAI Chat Completions API (/v1/chat/completions). Works with most OpenAI-compatible endpoints including vLLM, Ollama, LiteLLM, and Mistral.
  • openai_responses — OpenAI Responses API. Use this for newer OpenAI models that require the Responses API format.

​
Using a custom provider

Reference the provider by name from a model definition, then reference the model from an agent: 
providers:
  my_provider:
    api_type: openai_chatcompletions
    base_url: https://api.example.com/v1
    token_key: MY_API_KEY

models:
  my_model:
    provider: my_provider
    model: gpt-4o
    max_tokens: 32768

agents:
  root:
    model: my_model
    instruction: You are a helpful assistant.

​
Shorthand syntax

Once a provider is defined, you can reference it with the inline provider/model shorthand without defining a named model: 
agents:
  root:
    model: my_provider/gpt-4o-mini
The provider’s base_url, token_key, and api_type are applied automatically.

​
How it works

When you reference a custom provider, docker-agent:
  1. Applies the provider’s base_url to the model (if not already set on the model).
  2. Applies the provider’s token_key to the model (if not already set on the model).
  3. Stores the provider’s api_type in provider_opts.api_type.
  4. Routes the request to the appropriate API client.

​
Examples

providers:
  local_llm:
    api_type: openai_chatcompletions
    base_url: http://localhost:8000/v1
    # no token_key needed for unauthenticated local servers

models:
  llama:
    provider: local_llm
    model: meta-llama/Llama-3.2-3B-Instruct

agents:
  root:
    model: llama

Build docs developers (and LLMs) love

Get started for freeTalk to us