API reference

The OpenAPI document is the source of truth for the public and admin HTTP API:

• apps/gateway/openapi.yaml — OpenAPI 3.1 specification. Importable into Bruno, Postman, Insomnia, and similar tools.

Authentication

• Inference (/v1/*): master key or virtual key, via Authorization: Bearer <key> (alternatives: the x-api-key header, or ?api_key=<key> for EventSource).
• Admin (/admin/*): requires the master key (MASTER_KEY from .env).

Importing into Bruno

Use Import Collection → OpenAPI V3 and select apps/gateway/openapi.yaml.

After import, set the server variable:

• baseUrl — e.g. http://localhost:4000

Set the bearer token in Bruno's auth UI:

• Inference requests may use the master key or a virtual key.
• Admin requests require the master key.

Typical flow to serve a model

1. POST /admin/deployments — create a deployment: publicModel + provider + upstreamModel + the inline API key.
2. POST /admin/keys — issue a virtual key with allowedModels.
3. Call an inference endpoint with the public model name in the model field.

Optional: Admin → Deployments - Resolve validates the profile/operations/transports without saving.