Abraham Gómez

This guide turns the official repository quick start into a copy-and-paste sequence for a Mac with Apple Silicon and 16 GB of RAM. The repository is the source of truth; hardware-specific performance should be verified on the target machine.

1. Check the host

Install Docker Desktop for Apple Silicon and make sure it is running. Then clone the repository:

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

Run the project's diagnostic command before starting containers:

./zyra-up.sh doctor

On a 16 GB machine, close memory-heavy applications before pulling models. The doctor command is the authoritative check provided by the repository.

2. Configure the environment

The README instructs you to copy the example environment file:

cp zyrabit-slm/example.env zyrabit-slm/.env

Open it locally and set the values required by the project, including SLM_URL and DB_URL when requested by the current checkout:

open -e zyrabit-slm/.env

Never commit this file. If the project asks for a token, use a local secret rather than writing it into this article:

ZYRABIT_TOKEN=replace-with-a-local-token

The exact variable name and authentication flow must match the checked-out repository version. This guide does not invent an API endpoint or token format.

3. Install the stack

Use the unified lifecycle command:

./zyra-up.sh install

According to the README, this builds images, starts the stack, and pulls the required AI models. On the first run it can take time and consume significant disk and memory.

4. Verify services

After installation, run the visual and API health checks exposed by the repository:

./zyra-up.sh verify

If you need to inspect the running stack:

./zyra-up.sh start
./zyra-up.sh stop

5. Useful lifecycle commands

doctor  Diagnose RAM, hardware, and environment
install Build images, start the stack, and pull models
start   Bring up the infrastructure
stop    Tear down the infrastructure
build   Build Docker images without starting services
verify  Run visual health checks and API probes
dev     Run native local development with uv

6. If the Mac is memory constrained

Start with doctor, then avoid launching optional services until the base stack is healthy. The repository documents n8n as optional and describes an Ollama-backed model network. Choose smaller local models where the current configuration permits it, and record the result rather than assuming a benchmark.

Source and screenshots

The commands and architecture in this guide are derived from the official Zyrabit SLM README. The repository source used for this post did not provide a verified Mac M1 Pro screenshot sequence, so this version intentionally uses terminal commands instead of fabricated screenshots.