[PiquelChips]

The October cleanup pass (#241–#249) is the only point where the codebase shrank while features kept landing, and the November 2 audit (#354, #355, #357–#359, #361–#365) found what the line counts alone don't show. Both were run by hand from long threads, so repeating them meant rebuilding the checks, the area split, the agent briefs and the review routing from scratch.

This adds a piqueld-cleanup skill that runs both the same way:

• Audit: scripts/stats.py measures the repo (hand-written, generated and lockfile lines, Rust source against tests, test share per area, files over 1,500 lines, panic and lint-suppression counts, mod.rs files with no children). references/audit.md lists the checks the November audit ran, what each found and the PR it led to, and how to brief a thread for each suggestion the owner picks.
• Split: the workspace is divided into eight areas, each with one owner. One Opus agent per area opens a draft refactor PR, and a docs lane runs alongside.
• Briefs: the briefs the original agents and reviewers followed, updated for the current repo, in references/.
• Reviews: each PR is reviewed by Sol (Opus if Sol is over its limit), with a focus list built from the PR's own "unchanged" claims. The skill also records what the first pass's reviews caught in each area.
• What went wrong last time: the first pass's fixes are built in. Every file has one owner, the docs lane doesn't edit doc comments, branch names are dated, there's a ledger that survives compaction, and the skill handles flaky tests and usage limits.
• Report and follow-ups: bugs are verified against main before the report goes out, and the stats are rerun after the merge. Cross-area PRs, a doc-comment PR, parity issues and fix rounds start only when you ask.

Run on main just after #354 (a77cb7ab), the script reproduces the November audit's numbers (35% of Rust is test code, 2% in piqueld-ui, 20% of the repo generated); it also runs cleanly on the current main.

Note

Written by Claude Opus 5.5 in Claude Code. The area pass parts were reviewed by GPT 6.1 Sol (via T3 Code)

🤖 Generated with Claude Code

[coderabbitai]

Important

Draft PR not reviewed

Draft PRs are not automatically reviewed by default.

• Trigger a manual review

To automatically review draft PRs, update your CodeRabbit configuration:

reviews:
  auto_review:
    drafts: true

• Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

Share

• X
• Mastodon
• Reddit
• LinkedIn