Repository navigation
Review-bottleneck guardrails: PR-size caps, auto-split, review-load meter #402
Description
From the evidence round (Faros 2026: +31% PRs merged with zero review, +157% review latency; LinearB: agentic PRs 2.6x larger, idle 5.3x longer). An orchestrator that parallelizes agents MANUFACTURES this bottleneck, so it should ship the guardrails as defaults:
- agent diffs over ~400 lines get auto-split (stacked PRs, Stacked PRs: position badges, ordered stack navigator, merge the safe prefix #240),
- a review-load meter on the planboard (human review capacity is the constraint),
- DORA's small-batches capability implemented as product defaults.
Activity
currentbits commentedon Aug 18, 2026
Implemented on branch
coder/review-bottleneck-guardrails-pr-size-cap-f0adad(commit 373334a), not yet pushed/PR'd.What shipped:
prDiffCapLinessetting (default 400 lines, DORA small batches; null disables).createPrmeasures the three-dot numstat vs the base branch and refuses oversized PRs before pushing, with a stablePR too large:message;allowOversizeis the explicit human override.--base <previous-slice-branch>, attribution bullet) or "Create anyway". True stacked-PR UX (position badges, safe-prefix merge) stays with Stacked PRs: position badges, ordered stack navigator, merge the safe prefix #240.Tests: new electron/test/pr-size-cap.test.js (9), reviewLoad/formatLineCount units, planboard meter rendering, oversize-bar flow, SettingsModal cap field. Renderer 1174/1174 green; electron 1481 pass with 16 failures that reproduce identically on the pristine tree (pre-existing/environmental). Review itinerary at .solenta/review-itinerary.json on the branch.
Merge origin/main into #402 branch (combine review itineraries)
currentbits commentedon Aug 22, 2026
Verified shipped on
main(audit 2026-08-22): implementation present in the working tree, not just on a worker branch. Closing — the planboard never auto-closed these (#632).