How to Deploy Coze Studio with Docker: Self-Hosting Tutorial

Before you start

Coze Studio ships as two containers: the server (backend API) and the web (frontend). Both images are published on Docker Hub and have hundreds of thousands of pulls, so they are production-proven.

System requirements

Step 1 — Clone the repository

git clone https://github.com/coze-dev/coze-studio.git
cd coze-studio

Step 2 — Configure environment

Copy the example environment file and set your database, Redis, and LLM provider credentials. Coze Studio supports OpenAI-compatible endpoints, so you can plug in virtually any model provider:

cp .env.example .env
# edit .env — set DB, REDIS, and LLM_API_KEY

Step 3 — Start with Docker Compose

docker compose up -d

This starts the server on port 8080 and the web frontend. Wait for the containers to become healthy, then open http://localhost:8080.

Step 4 — Build your first agent

Once the UI loads, create a new agent, pick a model, and start adding plugins or designing a workflow. Use the built-in debugger to test conversations, then publish your agent to a web app or API endpoint.

Troubleshooting

• Port already in use: change the host port mapping in docker-compose.yml.
• Agents not responding: verify your LLM API key and network access to the model endpoint.
• Slow first load: the web container builds static assets on first start; allow a few minutes.