This guide walks through a self-hosted installation of Coze Studio, ByteDance’s open-source AI agent development platform with visual workflow builder.
For Docker installation, see Docker.
Use Git to download the source code.
git clone https://github.com/coze-dev/coze-studio.git && cd coze-studio
Use the provided Makefile to start Coze Studio:
make web
Copy the environment file and start with Docker Compose:
cp ./docker/.env.example ./docker/.env
docker compose -f ./docker/docker-compose.yml up
Note: First-time deployment may take longer as images are pulled and built locally.
Open your browser and navigate to the registration page:
http://localhost:8888/sign
Create your admin account.
Access the model management interface to configure your LLM providers:
http://localhost:8888/admin/#model-management
Note: Model configuration is required for image versions ≥ 0.5.0 before you can build agents or workflows.
Open your browser and navigate to:
http://localhost:8888/
You can now start building AI agents and workflows!
localhost only.Prefer automation? See Coze Studio Ansible Setup for an example playbook.
Prefer containers? See Coze Studio Docker Setup for detailed Docker deployment instructions.
docker compose logs -fFor more troubleshooting, see the FAQ.
Any questions?
Feel free to contact us. Find all contact information on our contact page.