Getting started with Docker Compose
Install Akmatori with Docker Compose, create the admin password and connect an LLM provider. Then connect your alert sources and tools.
On Kubernetes? Use the Kubernetes guide instead. The steps after the install are the same.
Prerequisites
Docker and Docker Compose
Docker with Docker Compose v2
LLM API key
For OpenAI, Anthropic, Google, OpenRouter, NVIDIA NIM, MiniMax or Ant Ling, or for your own OpenAI-compatible endpoint
Slack app (optional)
To post investigation results to channels and to approve actions in Slack
Installation
Download the release files
Download the Compose file and the nginx config of the latest release. You do not need to clone the repository or build anything:
mkdir akmatori && cd akmatori
curl -fsSLO https://github.com/akmatori/akmatori/releases/latest/download/docker-compose.yml
mkdir proxy && curl -fsSL -o proxy/nginx.conf \
https://github.com/akmatori/akmatori/releases/latest/download/nginx.confStart the services
Pull the pre-built images from GHCR and start them. Then check that all services show "Up":
docker compose pull
docker compose up -d
docker compose psThis starts PostgreSQL, the API server, the MCP Gateway, the agent worker, the web UI and an nginx proxy. On the first run Akmatori generates the database password and the JWT secret.
.env file. Create one only to change the port (HTTP_PORT, default 8080), pin a version (AKMATORI_VERSION, default latest), set an outbound proxy (HTTP_PROXY, HTTPS_PROXY), set JWT_SECRET yourself, or set ADMIN_PASSWORD to skip the setup wizard. Complete the setup wizard
Open the web UI:
http://localhost:8080 On the first visit a setup wizard opens. Set the password of the admin user (at least 8 characters).
Configure the LLM provider
After you log in, a welcome dialog asks for an API key. It saves the key to the active configuration, which on a new install is OpenAI with gpt-5.5. To use a different provider or model, go to Settings → AI models. The providers and some of their models:
- OpenAI (GPT-6, GPT-5.6 Terra / Sol / Luna, GPT-5.5, GPT-5.4, o4-mini)
- Anthropic (Claude Fable 5.1 / 5, Opus 5.5, Sonnet 5.5, Sonnet 5, Opus 4.8, Haiku 5.5 / 4.5)
- Google (Gemini 3.1 Pro Preview, Gemini 3.5 – 3.8 Flash, Gemini 2.5 Pro / Flash)
- OpenRouter (models of many providers through one API key)
- NVIDIA NIM (Llama 3.3 / 3.1 70B Instruct, Nemotron models)
- MiniMax (MiniMax-M3, M3.1-Flash, M2.7)
- Ant Ling (Ling-2.6-1T, Ling-2.6-flash, Ring-2.6-1T)
- Custom: any OpenAI-compatible endpoint, including Ollama, vLLM and LiteLLM (see Local models)