Reflection in the Dark: Exposing and Escaping the Black Box in Reflective Prompt Optimization

Computer Science > Artificial Intelligence [Submitted on 19 Mar 2026] Reflection in the Dark: Exposing and Escaping the Black Box in Reflective Prompt Optimization Shiyan Liu, Qifeng Xia, Qiyun Xia, Yisheng Liu, Xinyu Yu, Rui Qu Automatic prompt optimization (APO) has emerged as a powerful paradigm for improving LLM performance without manual prompt engineering. Reflective APO methods such as GEPA iteratively refine prompts by diagnosing failure cases, but the optimization process remains black-box and label-free, leading to uninterpretable trajectories and systematic failure. We identify and empirically demonstrate four limitations: on GSM8K with a defective seed, GEPA degrades accuracy from 23.81% to 13.50%. We propose VISTA, a multi-agent APO framework that decouples hypothesis generation from prompt rewriting, enabling semantically labeled hypotheses, parallel minibatch verification, and interpretable optimization trace. A two-layer explore-exploit mechanism combining random restart and epsilon-greedy sampling further escapes local optima. VISTA recovers accuracy to 87.57% on the same defective seed and consistently outperforms baselines across all conditions on GSM8K and AIME2025. Subjects: Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA) Cite as: arXiv:2603.18388 [cs.AI]   (or arXiv:2603.18388v1 [cs.AI] for this version)   https://doi.org/10.48550/arXiv.2603.18388 Focus to learn more Submission history From: Shiyan Liu [view email] [v1] Thu, 19 Mar 2026 01:14:36 UTC (1,285 KB) Access Paper: view license Current browse context: cs.AI < prev   |   next > new | recent | 2026-03 Change to browse by: cs cs.MA References & Citations NASA ADS Google Scholar Semantic Scholar Export BibTeX Citation Bookmark Bibliographic Tools Bibliographic and Citation Tools Bibliographic Explorer Toggle Bibliographic Explorer (What is the Explorer?) Connected Papers Toggle Connected Papers (What is Connected Papers?) Litmaps Toggle Litmaps (What is Litmaps?) scite.ai Toggle scite Smart Citations (What are Smart Citations?) Code, Data, Media Demos Related Papers About arXivLabs Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?)