CLI Reference¶

All rune commands and options.

Agent¶

Manage deployed agents.

rune agent ls [--control-plane http://localhost:8081]
rune agent inspect <name> [--ns <namespace>] [--alias <alias>] [--control-plane ...]
rune agent sessions <name> [--ns <namespace>] [--alias <alias>] [--control-plane ...] [--database-url ...]
rune agent stop <name> [--ns <namespace>] [--alias <alias>] [--control-plane ...]
rune agent rm <name> [--ns <namespace>] [--alias <alias>] [--force] [--control-plane ...]

ls — List all deployed agents
inspect — Show deployment details and replica health for an agent
sessions — View conversation session history for an agent
stop — Scale an agent to 0 replicas (by name)
rm — Remove an agent deployment (by name); --force disables FK checks on cascade failure

Run¶

Deploy a single agent, or invoke one directly.

Deploy — registers the agent with the control plane and starts replicas:

rune run <agent_spec> [--namespace dev] [--alias stable] [--control-plane http://localhost:8081]

agent_spec: Path to a Runefile, an agent directory, or git://repo-url[#subdir]

Invoke — send a single prompt to a running agent and print the response:

rune run --agent <name> "<prompt>" [--stream] [--session-id <uuid>] [--gateway http://localhost:8080]

# Examples
rune run --agent chat "Hello, what can you do?"
rune run --agent web-researcher "What is Raft consensus?" --stream
rune run --agent memory-agent "My name is Alice." --session-id 550e8400-e29b-41d4-a716-446655440000

Workflow¶

Run a workflow.yaml pipeline across multiple agents.

rune workflow run --file workflow.yaml --input "your input" [--gateway http://localhost:8080]

--file — Path to workflow.yaml
--input — Input string, available as {{ input }} in step templates

Status¶

List deployments and replica health.

rune status [--control-plane http://localhost:8081]

Sessions¶

View conversation history for a session or deployment.

rune sessions <session_id|deployment_id> [--database-url <url>]

Accepts either a session ID or a deployment ID. When given a deployment ID, shows the most recent session for that deployment. Use rune status or rune agent ls to get IDs.

Default database-url: sqlite:~/.rune/rune.db

Daemon¶

Start or stop the Rune server.

rune daemon start [options]
rune daemon stop [--pid-file /tmp/rune.pid]
rune daemon status [--pid-file /tmp/rune.pid]

Daemon Start Options¶

Option	Default	Description
`--gateway-addr`	`0.0.0.0:8080`	Gateway HTTP listen
`--control-plane-addr`	`0.0.0.0:8081`	Control plane HTTP listen
`--database-url`	`sqlite:~/.rune/rune.db`	SQLite path
`--pid-file`	`/tmp/rune.pid`	PID file path
`--log-file`	`~/.rune/rune.log`	Daemon log file
`--foreground`	—	Don't daemonize
`--node-id`	—	Raft node ID (cluster mode)
`--raft-addr`	`0.0.0.0:9000`	Raft gRPC listen
`--peers`	—	Comma-separated `ID@HOST:PORT`
`--backend`	`wasm`	`wasm` / `docker` / `kubernetes`
`--k8s-namespace`	`rune`	K8s namespace (when backend=kubernetes)

Stop¶

Scale a deployment to 0 replicas (by deployment ID).

rune stop <deployment_id> [--control-plane http://localhost:8081]

Use rune agent stop <name> to stop by agent name instead.

Rm¶

Remove a deployment entirely (by deployment ID).

rune rm <deployment_id> [--force] [--control-plane http://localhost:8081]

--force — Force delete when normal cascade fails; disables foreign key checks (may leave orphan rows).

Use rune agent rm <name> to remove by agent name instead.

Compose¶

Deploy and manage multiple agents from a compose file.

rune compose up [-f rune-compose.yaml] [--control-plane http://localhost:8081]
rune compose down [-f rune-compose.yaml] [--control-plane http://localhost:8081]
rune compose ps [-f rune-compose.yaml] [--control-plane http://localhost:8081]

Default compose file name: rune-compose.yaml (also accepts rune-compose.yml).

Cluster¶

Manage Raft cluster membership.

rune cluster init [--addr http://localhost:9000]
rune cluster add-learner <node_id> <node_addr> [--leader-addr http://localhost:9000]
rune cluster change-membership <member_id,...> [--leader-addr http://localhost:9000]
rune cluster status [--addr http://localhost:9000]

Debug¶

Tail daemon logs or restart the daemon with a new log level.

rune debug [--restart] [--log-level info] [--pid-file /tmp/rune.pid] [--log-file ~/.rune/rune.log]

--restart — Send SIGTERM to the current daemon, then restart it before tailing
--log-level — Log level for the restarted daemon (trace / debug / info / warn / error)