# JET: Judge-Guided Evolution at Test Time for Agent Programs

arXiv: https://arxiv.org/abs/2609.34126 · HTML: https://mlsanity.com/paper/2609.34126/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 27 of its 30 references that are in the map.
- First posted on arXiv 28 Sep 2026.

## Builds on (most cited first, 27 of 27)

- [ReAct: Synergizing Reasoning and Acting in Language Models](https://mlsanity.com/paper/2210.03629/) (2022)
- [Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena](https://mlsanity.com/paper/2306.05685/) (2023)
- [Reflexion: Language Agents with Verbal Reinforcement Learning](https://mlsanity.com/paper/2303.11366/) (2023)
- [Self-Refine: Iterative Refinement with Self-Feedback](https://mlsanity.com/paper/2303.17651/) (2023)
- [Diffusion Policy: Visuomotor Policy Learning via Action Diffusion](https://mlsanity.com/paper/2303.04137/) (2023)
- [WebShop: Towards Scalable Real-World Web Interaction with Grounded Language Agents](https://mlsanity.com/paper/2207.01206/) (2022)
- [Scaling Laws for Reward Model Overoptimization](https://mlsanity.com/paper/2210.10760/) (2022)
- [AlphaEvolve: A coding agent for scientific and algorithmic discovery](https://mlsanity.com/paper/2506.13131/) (2025)
- [DSPy: Compiling Declarative Language Model Calls into Self-Improving Pipelines](https://mlsanity.com/paper/2310.03714/) (2023)
- [GEPA: Reflective Prompt Evolution Can Outperform Reinforcement Learning](https://mlsanity.com/paper/2507.19457/) (2025)
- [Promptbreeder: Self-Referential Self-Improvement Via Prompt Evolution](https://mlsanity.com/paper/2309.16797/) (2023)
- [Learning Robust Rewards with Adversarial Inverse Reinforcement Learning](https://mlsanity.com/paper/1710.11248/) (2017)
- [Generative Verifiers: Reward Modeling as Next-Token Prediction](https://mlsanity.com/paper/2408.15240/) (2024)
- [AFlow: Automating Agentic Workflow Generation](https://mlsanity.com/paper/2410.10762/) (2024)
- [Automated Design of Agentic Systems](https://mlsanity.com/paper/2408.08435/) (2024)
- [Meta-Harness: End-to-End Optimization of Model Harnesses](https://mlsanity.com/paper/2603.28052/) (2026)
- [Agent-as-a-Judge: Evaluate Agents with Agents](https://mlsanity.com/paper/2410.10934/) (2024)
- [Self-Taught Optimizer (STOP): Recursively Self-Improving Code Generation](https://mlsanity.com/paper/2310.02304/) (2023)
- [How to Evaluate Reward Models for RLHF](https://mlsanity.com/paper/2410.14872/) (2024)
- [Quantifying Differences in Reward Functions](https://mlsanity.com/paper/2006.13900/) (2020)
- [Rethinking the Evaluation of Harness Evolution for Agents](https://mlsanity.com/paper/2607.12227/) (2026)
- [Learning Human Objectives by Evaluating Hypothetical Behavior](https://mlsanity.com/paper/1912.05652/) (2019)
- [Rethinking Reward Model Evaluation: Are We Barking up the Wrong Tree?](https://mlsanity.com/paper/2410.05584/) (2024)
- [AgentRM: Enhancing Agent Generalization with Reward Modeling](https://mlsanity.com/paper/2502.18407/) (2025)
- [TTHE: Test-Time Harness Evolution](https://mlsanity.com/paper/2607.08124/) (2026)
- [JIT-Agent: Scaling Harness Intelligence via Just-in-Time Harness Evolution](https://mlsanity.com/paper/2608.25593/) (2026)
- [Boosting Virtual Agent Learning and Reasoning: A Step-Wise, Multi-Dimensional, and Generalist Reward Model with Benchmark](https://mlsanity.com/paper/2503.18665/) (2025)

## Abstract (arXiv)

An agent's executable program governs how it uses tools, processes observations, and responds to failures. Evolving this program at test time can help adaptation, but deciding which changes to retain is difficult when true rewards are unavailable. Execution traces provide evidence of agent behavior, yet interpreting that evidence requires a judge that remains useful as tasks and candidate programs change. We introduce Judge-Guided Evolution at Test Time (JET), which evolves an executable judge on labeled source trajectories, then freezes and transfers it to guide target-side program evolution. The judge supplies scores and diagnostic feedback without target evaluator access or model-weight updates. On unseen WebShop tasks, JET achieves approximately 13% higher mean reward than fixed-rubric guidance when evolution begins from an unevolved program (cold start) and 4% higher when it begins from one already optimized on source tasks (warm start), with a 36% relative improvement in cold-start exact success. An exact-judge control on PushT, where the judge reconstructs the scoring rule from observations, shows that without judge error, program search becomes the bottleneck. Analyses identify useful reward-prediction logic in the evolved code and show that better final selection alone cannot explain the gains. These results support executable judge transfer for program adaptation under evaluator-preserving task shifts.