Context
Thread 5 of the code-review-suite forward programme (effectiveness, agreed 2026-06-17).
Gap
Re-review mode is lightweight today — it only checks that previously-identified findings were addressed. Fast, usually correct, and liked. But there's a window: between the original review and the re-review, the author could add a substantial amount of new code that rides through without proper scrutiny.
Fix
Add a threshold on the newly-added diff (the delta since the last reviewed commit). When the new content exceeds the threshold, escalate to a full review of that delta instead of the lightweight addressed-check.
Open design questions
- What to measure: raw added lines? added files? new findings-eligible surface?
- What threshold: hard number? proportional to original diff size?
- UX: does the reviewer see "escalated to full review because X lines were added", or is it silent?
Why this ordering
Real gap, but the failure mode is rare in practice (most re-review cycles are small fix-up commits). Correct but low urgency.
Status
Not started. Low urgency.
Context
Thread 5 of the code-review-suite forward programme (effectiveness, agreed 2026-06-17).
Gap
Re-review mode is lightweight today — it only checks that previously-identified findings were addressed. Fast, usually correct, and liked. But there's a window: between the original review and the re-review, the author could add a substantial amount of new code that rides through without proper scrutiny.
Fix
Add a threshold on the newly-added diff (the delta since the last reviewed commit). When the new content exceeds the threshold, escalate to a full review of that delta instead of the lightweight addressed-check.
Open design questions
Why this ordering
Real gap, but the failure mode is rare in practice (most re-review cycles are small fix-up commits). Correct but low urgency.
Status
Not started. Low urgency.