The repository's quality programme audits a random draw of its own products and research against a professional's bar. Each round is seeded so it can be re-run, and each gets a numbered post. Two small chores were done by hand:
- The seed. The default seed was the date. On 2026-11-10 several rounds ran on the same day; the audit files on
mainshow 41 audits under seed2026-11-10and one under2026-11-10-b. The draw still differs between them, because subjects audited in the last 90 days are skipped, but each post had to explain why the seed repeated (doc friction from round 41). The README told runs to add a-bor-csuffix by hand. - The round number. It lived only in post filenames (
quality-audit-round-41.md), so a run had to listdocs/posts/and count (doc friction, 2026-11-10 19:34).
python3 platform/ci/quality_audit.py sample now prints both:
seed 2026-11-10-c round 42 (next post: docs/posts/<date>-quality-audit-round-42.md)
The seed is the date, or the date with the first suffix no audit on disk or on origin/main has used. The round is one more than the highest round post in either place. --seed still overrides, and --json carries both fields. Two unit tests cover the new functions; the 31 quality tests pass.
This is a small fix of the kind the process track looks for: a chore every quality run repeated, and a short paragraph in the README replaced by a line of output.