VDCS – Vehicle Diagnostic Confidence System

Status: Production Ready (Fly.io Compatible)
Type: Deterministic Rule Engine + Optional AI Enrichment

Philosophy

VDCS is a "Glass Box" diagnostic intelligence engine. It is designed to be:

1. Deterministic: The same input always produces the exact same output.
2. Honest: It explicitly states when it is guessing versus when it knows.
3. Secure: No external calls (unless configured), strictly validated inputs.
4. Stateless: No user data is persisted.

API Contract (Android)

• Endpoint: POST /diagnose
• Format: JSON Input -> JSON Output
• Schema: See /docs for full OpenAPI spec.

Example Request

{
  "vehicle": { "make": "Honda", "model": "Civic", "year": 2010, "mileage_km": 260000, "fuel_type": "petrol" },
  "environment": { "region": "US_METRO", "city_size": "medium" },
  "dtcs": [{ "code": "P0420", "status": "active" }]
}

Running Locally

1. Install: pip install -r requirements.txt
2. Env: Copy .env.example to .env (Optional for Gemini).
3. Run: uvicorn app.main:app --reload

Running on Fly.io (Docker)

The project includes a production-ready Dockerfile.

1. Build: docker build -t vdcs .
2. Run: docker run -p 8080:8080 --env-file .env vdcs

Gemini Integration (Optional)

Gemini is used strictly to refine the tone of the summary. It CANNOT change urgency or costs. To enable:

1. Set ENABLE_GEMINI=true in .env.
2. Set GEMINI_API_KEY=your_key.
3. Set GEMINI_MODEL=gemini-1.5-pro.

Security

• No Auth: Currently public API (add API Gateway for auth).
• Input Validation: Strict Pydantic v2 rules.
• Safe AI: Gemini fails silently if it errors/timeouts, ensuring 100% uptime for the rule engine.