custom_provider_instancesTier 1 · 70% confidence

ai-agents-custom-provider-inst-users-need-to-connect-to-multiple-openai-anthropic-bbfde87e

agent: ai_agents

When does this happen?

IF Users need to connect to multiple OpenAI/Anthropic-compatible providers (e.g., OpenRouter, Together AI) or run multiple instances of the same provider for different environments/teams, but the current architecture enforces a one-to-one relationship between provider type and instance.

How others solved it

THEN Introduce a CustomProvider implementation that assigns a unique provider ID to each instance, includes provider type specification (OpenAI/Anthropic), and allows instance-specific configuration (custom base URLs, API keys). This maintains backward compatibility by using the unique ID as the provider key, thereby breaking the one-to-one constraint while preserving database uniqueness constraints.

// Example conceptual config: custom provider instances with unique IDs
{
  "providers": [
    {
      "id": "custom-openai-1",
      "type": "openai",
      "baseUrl": "https://openrouter.ai/api/v1",
      "apiKey": "env:OPENROUTER_KEY"
    },
    {
      "id": "custom-anthropic-2",
      "type": "anthropic",
      "baseUrl": "https://my-private-proxy.example.com",
      "apiKey": "env:PRIVATE_ANTHROPIC_KEY"
    }
  ]
}

Related patterns

Have you seen this in your site?

Connect AgentMinds to match against your tech stack automatically.

Run diagnostics