clawops

Commands

Every clawops command, grouped by what you reach for it.

clawops <command> --help prints the flags for any of these.

Global flags

FlagEffect
--stackTarget a named stack (default comes from config)
--providerOverride provider: aws, gcp, azure, local
--jsonEmit JSON to stdout, for scripting
--quietSuppress non-error output
--profileAuth profile from ~/.clawops/config.json
--dry-runPreview without applying, on mutating commands
--yesSkip interactive confirmations (CI)

Getting started

CommandDoes
initChoose a provider, name the state backend after your account (or pass --state to name it yourself), generate an SSH key
setupInteractive wizard: configure and deploy in one pass
doctorCheck the local machine; with --stack, the deployment too; with --provider, one cloud's credentials and account setup whether or not a stack exists; with --instance-type, account checks ask about that size. Exits 1 on any failure

Provisioning

CommandDoes
upProvision and deploy a stack (--gateway-port for a non-default port)
planGenerate a deploy plan without applying it (--ssh-cidr, --gateway-cidr, --publish-gateway say who may connect; --instance-type picks the machine size)
applyApply a plan produced by clawops plan
downDestroy provisioned resources for a stack
destroyDestroy everything in a stack, irreversible
migrateMove an existing 1.x deployment onto the 2.0 runtime, keeping its state
stacksList or delete stacks
statusOutputs, region, provisioned time

Operating a deployment

CommandDoes
logsStream gateway logs. Says which source answered: the gateway, or container output when it is not reachable
monitorLive dashboard: health, container status, resources, log tail
sshOpen a session to the instance
tunnelForward the gateway port to localhost
gatewaystatus, restart, update [version]
agentslist, logs <name>: per-agent activity from the audit log. OpenClaw 2.0 has no per-agent restart
configget, set, unset on the gateway configuration
backupcreate, restore: restore expands into a staging directory, never in place

Security

CommandDoes
hardenApply hardening modules: ssh, ufw, fail2ban, unattended-upgrades, auditd
secretlist, set, delete, rotate, audit

Other

CommandDoes
mcpMCP server operations: see MCP
bugOpen a pre-filled GitHub issue with system context
helpAvailable commands and global flags

On this page