Commands

Commands

TRON has two operator commands. Everything else runs itself.

tron seeder

Seed TRON into a target project. Run it from the project root. An interview that detects your repo, settles the knobs, and writes the instance into <agents>/tron/ — touching nothing else. Re-runnable (shows current values before overwriting; never clobbers your tuned workflow.yaml silently).

See Getting Started for the walkthrough.

tron start

Wake TRON. A short bootup — where to start (default: whole pipeline) and worker_count — then the live console. Installs the cron heartbeat so TRON keeps running after you close the console.

Inside the console

You type	Does
(any text)	talk to TRON — classified and routed; out-of-grammar is refused
status / fleet	the live fleet (workers, the architect + its queue)
pipeline	block statuses + cadence counters
log	recent events
attach <id>	a worker's recent activity
stop	release the fleet and end the session (guards unfinished work; stop --force to override)
help	the command list

Closing the console does not stop TRON — the heartbeat keeps ticking. Reattach any time with tron start (it replays recent events instead of re-bootstrapping).

Internal verbs (not the operator surface)

These exist behind tron but are cron-driven, automatic, or for debugging — you don't normally type them:

Verb	Used by
tron tick	the cron heartbeat (one bounded tick)
tron recover	reattach live workers after a crash
tron validate	blueprint-lint
tron doctor	environment check + lint
tron stop	same as stop inside the console
tron msg "<text>"	queue an operator line + tick (scripting)

See also: Configuration · Operations.