CLI

start reads braid.config.ts from the current directory by default — pass --config <path> to point elsewhere. Every other command just talks to the pidfile of whatever's already running, so it doesn't need the config again.

`braid start`

Forks every configured process, writes .braid/run.json (the pidfile), and detaches as a background daemon — the CLI process exits once every process has started; the daemon keeps running after.

braid start
braid start --config ./other.config.ts

Refuses to start if a pidfile from a still-running daemon already exists.

Pass --foreground to run attached to this terminal instead — start then blocks, streaming every process's combined output here, until Ctrl-C (or a braid stop from another terminal) stops everything. A config's own foreground: true makes this the default; --daemon forces the background daemon regardless of what the config says.

braid start --foreground
braid start --daemon

`braid status`

Lists each configured process's name, PID, and whether it's currently alive.

braid status

`braid logs`

Prints a process's persisted, rotated log — or every process, interleaved, if no name is given.

braid logs                    # every process, interleaved
braid logs api                # just "api"
braid logs api --follow       # keep streaming new output
braid logs api --lines 50     # only the last 50 lines

`braid stop`

With no name: kills every process recorded in the pidfile (and their descendants), then removes it. Works even if the daemon itself has crashed, since it kills PIDs directly rather than asking the daemon to do it.

braid stop

With a name: stops just that one process through the running daemon's control server, leaving everything else (and the daemon itself) running. Unlike the bare form, this needs the daemon to actually be reachable — there's no direct-PID fallback for a single process, since the daemon's own in-process state (dependents, in-flight guards) has to be involved.

braid stop api

`braid restart <name>`

Stops and respawns one named process, then runs its own onRestart hook (if any) and cascades to dependsOn dependents exactly as a watch-triggered restart would. A name is required — there's no bare whole-stack braid restart.

braid restart api

Can take a moment to return if onRestart keeps retrying. Prints an error and exits non-zero if the daemon isn't reachable.

Flags

--config <path> — start only; use a config file other than ./braid.config.ts.
--foreground — start only; run attached to this terminal instead of forking a background daemon.
--daemon — start only; force the background daemon, overriding a config's foreground: true.
--follow — logs only; keep the connection open and stream new output as it arrives.
--lines <n> — logs only; only print the last n lines.