getting started

bootstrapping mp on a fresh system, in one command:

sh mk/bs.sh

what that actually does, in order: builds mpx and mp (mk . b, or a plain sh mk/b.sh if you don't have the mk tool installed), installs both plus every mp.<command> and mplib/*.pm into MP_PREFIX (/usr/local by default, override the env var to install elsewhere), drops mp.conf.example and mpx.conf.example into /etc/mp.conf and /etc/mpx.conf if neither already exists, then runs "mp install meta/mp" so mp immediately tracks its own real dependencies (a c compiler, perl, and the mk build tool) the same way it tracks everything else.

after that, /etc/mp.conf is a real, editable config file, already commented with every key. the two you'll actually touch first:

REPO_main, the git url mp clones its ports tree from. the default already points at a real tree; change it (or add REPO_extra=... alongside it) to use your own.
PORTS_DIR, where every configured repo lives on disk. defaults to $WD/repos ($WD itself defaults to /usr/mp).

first real use:

mp search editor        # what's available
mp info gnu-nano        # one port's metadata, deps, and current status
mp install --dry-run gnu-nano   # resolve the whole dependency graph, install nothing
mp install gnu-nano      # the real thing

mp itself clones whatever repo doesn't exist yet the first time any command needs it, so the very first "mp search" or "mp install" is also what triggers the initial clone. no separate "mp init" step.

from here: guide/using-mp.html is the fuller command tour, guide/writing-a-port.html is for adding a package of your own, and mp.conf.example (repo root) is the exhaustive key-by-key config reference for everything this page only touched the surface of.