Quick Start

Quick Start

Get Amadeus running in under 10 minutes.

---

Prerequisites

Requirement	Version	Notes
Python	3.11+	Required
uv package manager	Latest	Recommended; or use pip
Docker Desktop	Latest	For containerised setup or sandboxed code execution
PostgreSQL	15+	Production; SQLite is the default for development
Redis	5+	Caching and rate limiting
LLM API key or local GGUF model	—	At least one required

---

Local Installation

# 1. Clone the repository
git clone https://github.com/adityatawde9699/Amadeus-AI.git
cd Amadeus-AI

# 2. Install dependencies
uv sync --all-extras --dev
# or: pip install -e ".[all]"

# 3. Configure environment
cp .env.example .env
# Edit .env — at minimum set GROQ_API_KEY and SECRET_KEY

# 4. Run database migrations
uv run alembic upgrade head

# 5. Start the server
uv run uvicorn src.api.server:app --reload --host 0.0.0.0 --port 8000

Interactive API docs available at http://localhost:8000/docs (requires DEBUG=true).

---

Offline / Local-First Mode

Download a GGUF model (e.g., Phi-3 mini) and place it in the Model/ directory:

pip install llama-cpp-python

Then set in .env:

SLM_MODEL_PATH=./Model/phi-3-mini-4k-instruct-q4.gguf
LOCAL_ONLY_MODE=true

---

Build the Workspace Search Index (optional)

Enables the search_workspace tool over your local files:

python scripts/index_workspace.py --root "C:\Users\ASUS\Downloads" --max-chunks 15000

# Subsequent runs are incremental — only changed files are re-embedded

---

Docker (Development)

docker-compose up --build
# API available at http://localhost:8000
# Interactive docs at http://localhost:8000/docs

---

Docker (Production)

# Set POSTGRES_PASSWORD and SECRET_KEY in .env first
docker-compose --profile prod up --build -d

# View logs
docker-compose logs -f api-prod

The production profile runs Gunicorn with 4 Uvicorn workers (2 CPU / 1 GB RAM limit). Redis and PostgreSQL ports are not exposed to the host — they remain internal to the Docker bridge network.

Windows shortcut: Run Setup_Amadeus.bat for guided first-time setup with auto-generated SECRET_KEY.

---

Generate a Test JWT

import jwt, datetime

payload = {
    "sub": "admin",
    "iat": datetime.datetime.utcnow(),
    "exp": datetime.datetime.utcnow() + datetime.timedelta(hours=24),
}
token = jwt.encode(payload, "your-secret-key", algorithm="HS256")
print(token)

Use the token in all protected API calls:

Authorization: Bearer <token>

---

← Home | Configuration-Reference →