Module mockai_api 

Standalone MockAI HTTP API Standalone MockAI HTTP API.

MockAI exists in the codebase but until now was only invoked from the OpenAPI route generator — i.e., a hosted mock with a spec got AI-augmented responses on its existing routes. There was no way to reach MockAI directly without a spec, which made it impossible to prototype an “AI persona” mock or hand-craft a one-off intelligent response without a full OpenAPI document.

This module exposes the engine over HTTP. Same functional contract the internal callers use: a Request goes in (method/path/body/etc) and a Response comes out (status_code/body/headers).

Endpoints

• POST /__mockforge/api/mockai/generate — generate one response
• GET /__mockforge/api/mockai/status — config / availability probe

Auth / cost

MockAI typically calls an LLM provider, so an API key needs to be configured at server start. The endpoint returns 503 with a clear reason if the engine isn’t available — same surface the OpenAPI handler treats absence with.

Structs

GenerateRequest

JSON body accepted by POST /__mockforge/api/mockai/generate.

GenerateResponseBody

JSON body returned from POST /__mockforge/api/mockai/generate.

MockAiApiState

Cheap-to-clone shared state. Holds the existing MockAI instance built at server startup; the API handler reads through the RwLock so a future hot-reload of the model config (e.g., re-uploading a spec) doesn’t require a router rebuild.

Functions

mockai_api_router

Build the MockAI standalone API router. Mount under /__mockforge/api/mockai.