Skip to content

Publish a run

A published run is a folder results/<date>[-what] with:

  • <sandbox>.json per sandbox: one line per case, with its result, basis, reason, exit statuses, effect and the contract's hash, plus the run's sandbox identity, host stamp, profile, spec version and corpus version;
  • raw/<sandbox>.full.json: everything the runner captured;
  • comparison.md: the table, generated;
  • review.json: the reading of every result other than contained, with who and why;
  • README.md: what was measured, on what, and anything a reader needs to read it.
python3 runner/report.py /tmp/out results/2026-11-02-mysandbox
python3 runner/versions.py > results/versions.md

Before publishing, check every result that is not contained against its raw output and give it a reading. State nothing that was not run. A published result is never edited: a correction is a new run, or a new reading that keeps the old one.

To submit a run measured elsewhere, open a pull request with the folder and the commit of the lab it ran. A finding about a sandbox follows the disclosure rules.