Skip to main content
The monte CLI is the only write path into the platform. This page states the conventions once, then lists every command.

​
Conventions

  • Default subverb. monte eval math means monte eval run math. Same for train. The env group has no default on purpose: list and install are different intents.
  • Confirm before launch. eval run and train run print the Run plan, then ask launch?. Skip the prompt with --yes. Without a TTY the prompt refuses (E_CONFIRM) instead of hanging. --dry-run prints the plan and launches nothing. The plan you approve is the exact object persisted as plan.json and hashed into the row’s provenance.
  • Typed confirm for promote. promote does not ask y/n. You must pass --confirm <measurement-name>, typed out. A mismatch is the same E_CONFIRM.
  • --json. Supported on eval run, train run, status, show, env list, and config only. Stdout carries pure JSON: the human plan summary moves to stderr. One trap: status --json omits the checkpoint ladder that the human view prints.
  • Refusals. Anything the platform declines exits 3 or higher with CODE: cause (fix) on stderr. The fix is part of the message.

​
Commands

​
monte init

Creates the Measurement: freezes the Splits, the eval settings, and the Environment fingerprint into measurement.json. Refuses an existing name (E_MEASUREMENT_EXISTS).

​
monte eval

Scores a model on a Split and appends the row. The default Split is dev. The default step is the Lineage’s newest Checkpoint. Test evals are budgeted: Baseline and final claim only.

​
monte train

Trains one chunk, chained from the Lineage’s newest Checkpoint. Refuses without a Baseline (E_NO_BASELINE). --smoke runs a capped micro-chunk that is never evidence. --chained only matters with --smoke: real runs always chain.

​
monte status

Prints the Run rows and the per-Lineage delta lines, paired when possible and labelled unpaired when not. Also prints Cycles with their strike verdicts and the checkpoint ladder.

​
monte logs, show, stop

logs prints the Run’s event file as JSON lines. show prints one full row with all provenance fields. stop terminates a running Run and the row reads as abandoned.

​
monte promote

Exports a succeeded training Run’s servable Checkpoint to --dest, sha256-verified. It then prints the file count and the destination disk usage.

​
monte env

list shows each Environment and its Gym benchmarks. install verifies frozen data against its manifest and registers the benchmark configs. Install is idempotent. A mock-only Environment prints a note instead.

​
monte config

Prints the resolved configuration: the data root and its source, the GPU check command, mock seconds, the base-model pin, the installed Environments, and the schema version.

​
Environment variables

​
Exit codes