Skip to main content

Quickstart Fundamentals

The deployment of Zyrabit SLM requires technical precision and environmental isolation. This guide provides a step-by-step procedure to initialize your environment and launch the core services.

Zyrabit adopts a Zero-Config Bias: The engine automatically detects and utilizes the optimal compute kernels (Metal, CUDA, or AVX2) based on your host hardware.


1. Environment Preparation​

Before executing the orchestration, ensure your Standard Toolchain is configured. We utilize uv for hermetic dependency management and Docker for container isolation.

Install the uv Toolchain​

uv ensures that Python dependencies remain isolated and reproducible.

curl -LsSf https://astral.sh/uv/install.sh | sh

Verify Docker Status​

Ensure Docker is running on your host system.

docker info

2. Project Initialization​

Follow these steps to clone the repository and prepare the local environment.

Step 1: Clone the Repository​

Download the Zyrabit SLM source code to your local machine.

git clone https://github.com/Zyrabit-tech/zyrabit-SLM.git

Step 2: Navigate to Project Root​

Change your working directory to the repository folder.

cd zyrabit-SLM

Step 3: Initialize the Virtual Environment​

Create a hermetic environment using uv to manage orchestration dependencies.

uv venv

Step 4: Activate the Environment​

Load the isolated environment into your current shell session.

source .venv/bin/activate

Step 5: Sync Dependencies​

Install the required Python packages defined in the project.

uv pip install -r requirements.txt

3. Launching the Stack​

Zyrabit uses a unified CLI script (zyra.sh) that automatically chooses the best configuration for your hardware. Local/Dev mode is the default — no extra flags required.

First-Time Installation (Local / Dev Mode)​

./zyra.sh install

This single command:

  1. Detects your hardware (RAM, GPU, Apple Silicon Metal, Tenstorrent)
  2. Runs interactive configuration (model selection, inference engine, ReAct agent)
  3. Generates/updates .env configuration safely
  4. Builds Docker images for the API, Web UI, MCP, and ChromaDB
  5. Starts all containers on direct local ports
  6. Verifies stack health and readiness

Access Your Services​

ServiceURL
Web UIhttp://localhost:3000
APIhttp://localhost:8082/v1
Healthhttp://localhost:8082/v1/health
Grafanahttp://localhost:3001
ChromaDBhttp://localhost:8000

Verify Health​

./zyra.sh verify

[!TIP] On the first run, the system will download approximately 3GB of container images and the foundational SLM model. Once running, test with: curl http://localhost:8082/v1/health


4. Going to Production​

When ready to deploy with a domain, HTTPS, and PostgreSQL:

./zyra.sh install --production

This activates:

  • Traefik reverse proxy with automatic TLS certificates
  • Custom domain configuration
  • PostgreSQL as a persistent relational database (via --profile db)
  • Strict rate limiting and security headers