VeeReview: Measuring Build-Verify Pairs in Agent-Authored Pull Requests from a V-Model Perspective
Objective: We test whether the V-Model's build and verify pairs can measure verification evidence in pull requests and support comparisons between agent-authored and human-authored changes.
ProofPaper ↗
Key points
- Context: Reviewing a pull request requires verification evidence that addresses the proposed change.
- Coding agent benchmarks and studies of pull request outcomes do not directly assess this relationship.
- We evaluated VeeReview against a 25-pull request reference set labelled by two raters and applied it to 1,454 matched pairs of agent and human pull requests from 741 repositories.
- Conclusions: The rule-decided pair shows that a process measure judged by a language model partly reflects how a pull request is presented rather than the verification work behind it.
Sources (1)
- [1]VeeReview: Measuring Build-Verify Pairs in Agent-Authored Pull Requests from a V-Model PerspectivearXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 21, 03:25 AM
Objective: We test whether the V-Model's build and verify pairs can measure verification evidence in pull requests and support comparisons between agent-authored and human-authored changes.
Context: Reviewing a pull request requires verification evidence that addresses the proposed change.
Extractive summary: sentences quoted from the sources.
Before this
- Oct 21, 2026Sparsity Should Be ALLOTted by Output: Training-Free Allocation of Block-Sparse Attention for Video Diffusion Transformers
- Oct 20, 2026Build a semantic layer for agentic AI on AWS with Strategy Mosaic and Amazon Bedrock AgentCore
- Oct 20, 2026Exa 3.0 Search API Tops Perplexity, Brave and Parallel on 8 Benchmarks
- Oct 20, 2026Ask to Distill: Learning to Synthesize Data for On-Policy Distillation
- Oct 20, 2026RoboCasa-Shift: Stage-Aware Robustness Evaluation of Vision-Language-Action Policies in Long-Horizon Household Manipulation
- Oct 20, 2026Training the Whole Agent, Not Just the Model