Skip to content

API_REFERENCE.md

Latest commit

 

History

History
60 lines (41 loc) · 2.48 KB

API_REFERENCE.md

File metadata and controls

60 lines (41 loc) · 2.48 KB
Raw

📖 vromvrom-engine API Reference

The vromvrom-engine exposes two types of APIs: a native REST API powered by FastAPI, and an OpenAI-compatible Proxy for universal integration.

🇫🇷 Version française disponible → API_REFERENCE.fr.md

1. Interactive Documentation (Swagger)

All native engine endpoints are interactively documented thanks to FastAPI. Once the engine is running (python gui_server.py), open your browser to:

👉 http://localhost:8000/docs

You will find:

  • The interface to test each endpoint.
  • Expected JSON schemas for inputs and outputs.
  • Error codes.

2. The OpenAI-Compatible Proxy (/v1)

This is one of the most powerful features of the engine. It exposes a 100% OpenAI API compatible endpoint, allowing you to use vromvrom-engine as an LLM backend for any IDE (Cursor, Cline, Aider) or third-party interface.

Endpoint: POST /v1/chat/completions

IDE Configuration (e.g., Cursor)

  • Base URL: http://localhost:8000/v1
  • API Key: The value of your MOTEUR_API_KEY defined in your .env.
  • Model: Use any model configured in your config.json (e.g., gemini-2.5-pro, github/gpt-4o, or auto to let the engine choose via the Elo system).

How it works under the hood

Even though your IDE thinks it's talking to OpenAI, the engine intercepts the request, applies Elo Routing, uses its own keys (via the KeyPool), and triggers the Circuit Breaker in case of a provider error, then returns the response formatted as OpenAI.

3. The Native Engine API (/api/*)

Here is an overview of the main engine-specific routes:

General Execution

  • POST /api/execute: Sends a complex task to the engine (Routing -> Planner -> Executor -> Reviewer).
  • POST /api/execute/stream: Same but returns the response as Server-Sent Events (SSE) for real-time streaming.

Home Assistant Integration (Home Automation)

  • GET /api/ha/state/{entity_id}: Retrieves the current state of a sensor (e.g., sensor.living_room_temperature).
  • POST /api/ha/control: Executes a service (e.g., turning on a light). 
    {
  "entity_id": "light.living_room",
  "service": "turn_on",
  "service_data": {"brightness": 255}
}

Workflows & Models Management

  • GET /api/workflows: Lists available DAG execution graphs.
  • GET /api/models: Lists available LLMs and their current Elo scores.

For the full specification, check the Swagger UI at http://localhost:8000/docs.