Commands

dvm sync <vm> | --all              create/start VM and run setup scripts
dvm sh <vm>                        interactive shell as DVM_USER
dvm ssh <vm> -- cmd...             run one command as DVM_USER
dvm cp <src> <dst>                 copy a file; one side is vm:path
dvm ls [--only-config] [<vm>]      list dvm-* Lima instances
dvm stop <vm> | --all [--only-config]   stop instances
dvm rm <vm> --yes [--config]       stop and delete the instance
dvm new <vm>                       write starter config and setup script
dvm version

dvm with no args runs dvm ls.

sync

Loads config, creates or starts dvm-<vm>, ensures DVM_USER and DVM_CODE_DIR exist, provisions subordinate uid/gid ranges, then runs the setup scripts (global, then per-VM). See config.md.

DVM_DRY_RUN=1 dvm sync app   # print Lima argv + setup order, no Lima contact
dvm sync app
dvm sync --all               # sync every VM with a config

sh / ssh

dvm sh opens DVM_USER’s login shell, starting in the project directory when it exists. dvm ssh runs one command as DVM_USER.

dvm sh app
dvm ssh app -- sudo journalctl -f

cp

One side must be vm:path. Relative VM paths resolve under /home/<DVM_USER>/code/<vm> and copy as DVM_USER; absolute VM paths are used as given. When the VM destination ends in / or names an existing directory, the file is copied into it under its own name; a bare vm: copies into the project root. Directory copies are not supported. Use ./tmp:notes.txt for a local path containing a colon.

dvm cp ./file app:/tmp/file      # to an exact path
dvm cp ./file app:notes.txt      # to code/app/notes.txt
dvm cp ./file app:               # into the project root, keeping the name
dvm cp app:notes.txt ./file

Lifecycle

dvm ls and dvm stop --all operate on dvm-* Lima instances. Add --only-config to limit them to instances that have ~/.config/dvm/vms/*/config.sh.

dvm rm <vm> --yes deletes the Lima instance. Add --config to also remove ~/.config/dvm/vms/<vm>.