Foundry AI Provider

Use this skill when

• adding @nyrra/foundry-ai to an app
• replacing direct public provider calls with Foundry proxy endpoints
• troubleshooting alias vs RID behavior or provider-specific Foundry caveats

Quick start

1. Install @nyrra/foundry-ai, ai, and only the provider peer dependency you need.
2. Load config from FOUNDRY_URL, FOUNDRY_TOKEN, and optional FOUNDRY_ATTRIBUTION_RID.
3. Import only the provider subpath you need: openai, anthropic, or google.
4. Use a known alias when the model is in the package catalog. Use a raw Foundry RID when it is not.
5. Compose multi-provider routing in application code with AI SDK createProviderRegistry.

Constraints