Configuration File Reference

Complete reference for the config.json file.

Full Schema

{
  "defaultProvider": "azure-foundry",
  "providers": {
    "azureFoundry": {
      "endpoint": "https://YOUR-RESOURCE.services.ai.azure.com",
      "apiKey": "your-api-key",
      "authMode": "apiKey",
      "models": [{ "id": "gpt-4o", "description": "GPT-4o" }],
      "chatCompletionPath": "/chat/completions",
      "apiVersion": "2024-10-21"
    },
    "azureAgents": {
      "endpoint": "https://RESOURCE.services.ai.azure.com/api/projects/PROJECT",
      "agentId": "my-agent-id",
      "apiKey": "optional-api-key",
      "apiVersion": "2025-05-15-preview"
    },
    "azureAnthropic": {
      "endpoint": "https://RESOURCE.openai.azure.com/anthropic",
      "apiKey": "your-api-key",
      "models": [{ "id": "claude-opus-4-6", "description": "Claude Opus 4-6" }]
    },
    "azureResponses": {
      "endpoint": "https://RESOURCE.openai.azure.com",
      "apiKey": "your-api-key",
      "models": [{ "id": "gpt-5.2-codex", "description": "GPT-5.2 Codex" }]
    }
  },
  "telemetry": false
}

Root Fields

defaultProvider

• Type: string
• Default: "azure-foundry"
• Values: "azure-foundry", "azure-agents", "azure-anthropic", "azure-responses"

The provider used when --provider is not specified on the command line.

telemetry

• Type: boolean
• Default: false

Telemetry opt-in switch. No telemetry is currently implemented — this is a placeholder for future use.

Azure Foundry Provider

Path: providers.azureFoundry

For standard Azure OpenAI chat completions API.

Field	Type	Required	Default	Description
endpoint	string (URL)	Yes	—	Azure resource URL
apiKey	string	No	—	API key for authentication
authMode	enum	No	"apiKey"	"apiKey", "azureCli", "managedIdentity"
models	array	No	[]	Available models
chatCompletionPath	string	No	"/chat/completions"	API path suffix
apiVersion	string	No	"2024-08-01-preview"	API version parameter

URL Construction

{endpoint}/openai/deployments/{model}{chatCompletionPath}?api-version={apiVersion}

Azure Agents Provider

Path: providers.azureAgents

For Azure AI Foundry Agent Service (threads/runs API).

Field	Type	Required	Default	Description
endpoint	string (URL)	Yes	—	AI Foundry project endpoint
agentId	string	Yes	—	Agent/assistant ID
apiKey	string	No	—	API key (or uses Azure CLI auth)
apiVersion	string	No	"2025-05-15-preview"	API version parameter

Azure Anthropic Provider

Path: providers.azureAnthropic

For Anthropic models (Claude) deployed on Azure AI Foundry. Uses the Anthropic Messages API through Azure's /anthropic gateway.

Field	Type	Required	Default	Description
endpoint	string (URL)	Yes	—	https://RESOURCE.openai.azure.com/anthropic
apiKey	string	Yes	—	Azure API key
apiVersion	string	No	"2023-06-01"	Anthropic API version header
models	array	No	[]	Available models

URL Construction

{endpoint}/v1/messages

Uses the x-api-key header for authentication and anthropic-version header for API versioning.

Azure Responses Provider

Path: providers.azureResponses

For models that use the OpenAI Responses API instead of Chat Completions (e.g. gpt-5.2-codex, codex-mini).

Field	Type	Required	Default	Description
endpoint	string (URL)	Yes	—	https://RESOURCE.openai.azure.com
apiKey	string	Yes	—	Azure API key
models	array	No	[]	Available models

URL Construction

{endpoint}/openai/v1/responses

Uses the api-key header for authentication. No API version parameter is needed.

Validation

The config file is validated against a Zod schema at load time. Invalid values produce clear error messages:

caretforge doctor
# ✗  Config valid: Invalid configuration: ...