Operate Your AgentOS

Start, stop, and restart an AgentOS project with Docker Compose, and inspect what is running and connected.

agno up        # docker compose up -d --build
agno down      # docker compose down
agno restart   # down, then up
agno status    # what is running, how it is secured, what is connected

up, down, and restart shell out to docker compose against your project’s compose file. Run them from a project directory created with agno create, or point them at a compose file with --file.

Compose File Detection

Without --file, the CLI looks for the first match in the current directory, then in ./infra:

docker-compose.yml, docker-compose.yaml, compose.yml, compose.yaml

up, down, restart

agno up --pull            # pull newer images before starting
agno down --volumes       # also remove named volumes (destroys data)
agno restart --dry-run    # print the docker compose commands without running them

Flag	Commands	Description
--file, -f	all	Compose file to use
--pull, -p	up, restart	Always pull newer images
--volumes, -v	down	Also remove named volumes. Destroys data.
--dry-run, -dr	all	Print the command without running it
--json	all	Emit a single JSON document

restart runs down, waits two seconds, then runs up.

status

agno status

status discovers the AgentOS (see discovery order) and reports:

• Base URL and agno version
• MCP endpoint, or disabled when the MCP server is off
• Auth mode: none, security_key, or jwt
• MCP auth, when the MCP endpoint carries its own OAuth protection
• Each supported coding agent: not detected, detected but not connected, or configured (with the config location)

Flag	Default	Description
--url	autodiscover	AgentOS base URL
--server-name	derived from the AgentOS name	MCP server entry name to look for in client configs
--json	off	Emit a single JSON document

Next Steps

Task	Guide
Connect coding agents	agno connect
Manage tokens	agno tokens
Full flag and JSON reference	agnoctl reference