Prompt Optimization

Large Language Models exhibit mode collapse, producing homogeneous outputs that fail to explore valid solution spaces. We present QD-LLM, a framework for parameter-efficient neuroevolution that evolves prompt embeddings, compact neural interfaces (~32K parameters) that steer generation in frozen LLMs (70B+ parameters), within a Quality-Diversity (QD) optimization framework. Our contributions: (1) evolved prompt embeddings via gradient-free optimization enabling behavioral steering without model fine-tuning; (2) hybrid behavior characterization combining semantic and explicit features with formal coverage bounds (Theorem 1) under validated near-independence (NMI $= 0.08 \pm 0.02$); (3) co-evolutionary variation operators including targeted behavioral mutation via finite-difference gradient estimation. On HumanEval (164 problems), MBPP, and creative writing benchmarks, QD-LLM achieves 46.4% higher coverage and 41.4% higher QD-Score than QDAIF ($p<0.001$, 30 runs, Vargha-Delaney $A=0.94$). We demonstrate downstream utility: diverse archives improve test generation (34% more edge cases) and fine-tuning data quality (8.3% accuracy gain). We validate across open-source LLMs (Llama-3-70B, Mistral-Large) with full embedding access, establishing prompt embedding evolution as an effective paradigm bridging neuroevolution and modern LLMs.

LLM coding agents now generate code at an unprecedented scale, yet LLM-generated code introduces cybersecurity vulnerabilities into codebases without human involvement. Even when frontier models are explicitly asked to write secure production code with relevant weaknesses to avoid in context, we find that they still produce verifiable vulnerabilities on average 23% of the time across a corpus of 250 benign coding prompts. We introduce SecureForge, an automated pipeline that both audits security risks of frontier models and produces auditing-informed secure system prompts that reduce output security vulnerabilities while maintaining unit test performance. SecureForge first identifies benign prompts that produce statically detectable vulnerabilities, and then amplifies them into a large synthetic prompt corpus of diverse scenarios using a Markovian sampling technique to jointly maintain error rates and prompt diversity. This corpus is then used to iteratively optimize the system prompts to reduce output security vulnerabilities. On frontier models, SecureForge yields a statistically significant Pareto improvement in both unit test success and output security, with output vulnerabilities reduced by up to 48%. The resulting system prompts transfer zero-shot to in-the-wild coding agent prompts, without any exposure to real user prompt distributions during optimization.

Large language model (LLM)-based Multi-agent systems (MAS) have shown promise in tackling complex collaborative tasks, where agents are typically orchestrated via role-specific prompts. While the quality of these prompts is pivotal, jointly optimizing them across interacting agents remains a non-trivial challenge, primarily due to the misalignment between local agent objectives and holistic system goals. To address this, we introduce MASPO, a novel framework designed to automatically and iteratively refine prompts across the entire system. A core innovation of MASPO is its joint evaluation mechanism, which assesses prompts not merely by their local validity, but by their capacity to facilitate downstream success for successor agents. This effectively bridges the gap between local interactions and global outcomes without relying on ground-truth labels. Furthermore, MASPO employs a data-driven evolutionary beam search to efficiently navigate the high-dimensional prompt space. Extensive empirical evaluations across 6 diverse tasks demonstrate that MASPO consistently outperforms state-of-the-art prompt optimization methods, achieving an average accuracy improvement of 2.9. We release our code at https://github.com/wangzx1219/MASPO.

Many sequential decision-making problems exhibit hierarchical structure, where high-level semantic choices constrain downstream actions and feedback is delayed and ambiguous. Learning in such settings is challenging due to credit assignment: performance degradation may arise from flawed abstractions, suboptimal execution, or their interaction. We study this challenge through pair trading, a domain that naturally combines long-horizon semantic reasoning for asset pair selection with short-horizon execution under partial observability. We formulate pair trading as a hierarchical reinforcement learning problem and propose a language-driven optimization framework in which both high-level and low-level policies are parameterized by large language models (LLMs) and optimized exclusively through prompt updates. Our approach leverages pretrained LLMs as hierarchical policies and uses trajectory- and episode-level textual feedback to adapt abstractions and execution without gradient-based fine-tuning. By explicitly separating abstraction selection from execution, the framework reduces non-stationarity across hierarchical levels and enables targeted adaptation under delayed feedback. Experiments on real-world market data show consistent improvements over traditional and LLM-based baselines, demonstrating the effectiveness of language-driven hierarchical reinforcement learning.

Current Large Language Model (LLM) evaluation frameworks utilize the same static prompt template across all models under evaluation. This differs from the common industry practice of using prompt optimization (PO) techniques to optimize the prompt for each model to maximize application performance. In this paper, we investigate the effect of PO towards LLM evaluations. Our results on public academic and internal industry benchmarks show that PO greatly affects the final ranking of models. This highlights the importance of practitioners performing PO per model when conducting evaluations to choose the best model for a given task.

We introduce Skills-Coach, a novel automated framework designed to significantly enhance the self-evolution of skills within Large Language Model (LLM)-based agents. Addressing the current fragmentation of the skill ecosystem, Skills-Coach explores the boundaries of skill capabilities, thereby facilitating the comprehensive competency coverage essential for intelligent applications. The framework comprises four core modules: a Diverse Task Generation Module that systematically creates a comprehensive test suite for various skills; a Lightweight Optimization Module dedicated to optimizing skill prompts and their corresponding code; a Comparative Execution Module facilitating the execution and evaluation of both original and optimized skills; and a Traceable Evaluation Module, which rigorously evaluates performance against specified criteria. Skills-Coach offers flexible execution options through its virtual and real modes. To validate its efficacy, we introduce Skill-X, a comprehensive benchmark dataset consisting of 48 diverse skills. Experimental results demonstrate that Skills-Coach achieves significant performance improvements in skill capability across a wide range of categories, highlighting its potential to advance the development of more robust and adaptable LLM-based agents.

LLM workflows, which coordinate structured calls to individual LLMs/agents to achieve a particular goal, offer a promising path towards building powerful AI systems that can tackle diverse tasks. However, existing approaches for building such workflows generally rely on human-crafted pipelines and prompts, which presents a substantial bottleneck in real world deployment. How can we automatically induce LLM-based agents and workflows in a data-driven way? This paper describes a simple data-driven approach for automatically inducing agents and LLM workflows. We formulate workflow induction as a bilevel optimization problem: an outer loop which optimizes a high-level sketch of the workflow (in particular how the LLM calls should be structured), and an inner loop which optimizes each individual LLM call one-by one. Both loops are optimized with textual gradients'' where for the inner loop we optimize each component in a modular way through backpropagating'' textual gradients layer-by-layer. We find that LLM workflows discovered through our FlowBot (workflow induction through bilevel optimization and textual gradients) approach performs competitively against strong baselines that make use of human-crafted or generated workflows.

Chain-of-Thought (CoT) prompting has emerged as a simple and effective way to elicit step-by-step solutions from large language models (LLMs). However, CoT reasoning can be unstable across runs on long, multi-step problems, leading to inconsistent answers for unchanged task. Most prior work focuses on improving the forward reasoning chain within a single pass, with less attention to iterative and contrastive correction. To address this gap, we propose CAP-CoT, a Cycle Adversarial Prompt optimization framework designed to improve both CoT reasoning accuracy and stability of a single deployed solver. In each cycle, a forward solver generates candidate reasoning chains, an adversarial challenger constructs plausible but deliberately flawed chains using targeted error strategies, and a feedback agent contrasts the two chains and produces step-aligned structured feedback. This feedback closes the optimization loop in two directions, including updating the solver prompt based on errors exposed by the challenger, and updating the challenger prompt to generate increasingly targeted errors in subsequent cycles. Unlike safety-oriented adversarial prompting such as jailbreak or prompt-injection attacks, our adversarial component is task-semantic and aims to expose logical vulnerabilities in reasoning chains. Experiments across six benchmarks and four LLM backbones demonstrate that within two to three adversarial prompt optimization cycles, CAP-CoT consistently reduces variability across runs while improving reasoning accuracy and robustness to prompt perturbations.

Large language models (LLMs) are increasingly utilized in various complex reasoning tasks due to their excellent instruction following capability. However, the model's performance is highly dependent on the open-ended characteristics of the users' input prompt. Natural prompts often do not follow proper syntactic rules, which creates ambiguous queries that yield multiple interpretations. Such ambiguous prompts confuse the model in choosing the correct reasoning paths to answer questions. Prior works address this challenge by applying query editing during the LLM inference process without explicitly solving the root cause of the ambiguity. To address this limitation, we propose a pre-inference prompt optimization mechanism via explicit prompt disambiguation. Particularly, we identify semantic risks in the prompt, check their multi-perspective consistency, and resolve any semantic conflicts that arise. Finally, we organize the resolved ambiguities in a logically structured manner as a clean input to the LLM. By explicitly resolving semantic ambiguity, our method can produce a more focused attention distribution to the semantically essential tokens. We also leverage small language models (SLMs) as the main executor of prompt disambiguation to benefit from their efficient computation. Through comprehensive experiments on multiple benchmarks, we demonstrate that our method improves reasoning performance by 2.5 points at a cost of only $0.02. Our study promotes explicit prompt disambiguation as an effective prompt optimization method without disturbing the internal mechanism of LLM inference.

Training large foundation models for agentic tasks is increasingly impractical due to the high computational costs, long iteration cycles, and rapid obsolescence as new models are continuously released. Instead of post-training massive models for every new task or domain, we propose Supplement Generation Training (SGT), a more efficient and sustainable strategy. SGT trains a smaller LLM to generate useful supplemental text that, when appended to the original input, helps the larger LLM solve the task more effectively. These lightweight models can dynamically adapt supplements to task requirements, improving performance without modifying the underlying large models. This approach decouples task-specific optimization from large foundation models and enables more flexible, cost-effective deployment of LLM-powered agents in real-world applications.

Large language model (LLM) agents augmented with external tools often struggle as number of tools grow large and become domain-specific. In such settings, ambiguous tool descriptions and under-specified agent instructions frequently lead to tool mis-selection and incorrect slot/value instantiation. We hypothesize that this is due to two root causes: generic, one-size-fits-all prompts that ignore tool-specific nuances, and underspecified tool schemas that lack clear guidance on when and how to use each tool and how to format its parameters. We introduce Joint Tool-Prompt Reflective Optimization (JTPRO), a framework for improving tool-calling reliability in trace-supervised settings by iteratively using rollout-driven reflection to co-optimize global instructions and per-tool schema/argument descriptions for accurate tool selection and argument instantiation in large tool inventories. JTPRO is designed to preserve only tool-local cues needed for correct disambiguation and slot filling. We evaluate JTPRO across multi-tool benchmarks, which account for different number of tools using three metrics: Tool Selection Accuracy (TSA), Slot Filling Accuracy(SFA), and Overall Success Rate(OSR) (correct tool + correct slots + correct values). JTPRO consistently outperforms strong baselines, including CoT-style agents, and reflective prompt optimizers such as GEPA by 5%-20% (relative) on OSR. Ablations show that joint optimization of instructions and tool schemas is more effective and robust than optimizing either component in isolation.

LLM agents in markets present algorithmic collusion risks. While prior work shows LLM agents reach supracompetitive prices through tacit coordination, existing research focuses on hand-crafted prompts. The emerging paradigm of prompt optimization necessitates new methodologies for understanding autonomous agent behavior. We investigate whether prompt optimization leads to emergent collusive behaviors in market simulations. We propose a meta-learning loop where LLM agents participate in duopoly markets and an LLM meta-optimizer iteratively refines shared strategic guidance. Our experiments reveal that meta-prompt optimization enables agents to discover stable tacit collusion strategies with substantially improved coordination quality compared to baseline agents. These behaviors generalize to held-out test markets, indicating discovery of general coordination principles. Analysis of evolved prompts reveals systematic coordination mechanisms through stable shared strategies. Our findings call for further investigation into AI safety implications in autonomous multi-agent systems.

We introduce the Semantic Density Effect (SDE): the empirical finding that prompts carrying higher semantic information per token consistently produce more accurate, focused, and less hallucinated outputs across all major LLM families. SDE is defined as the ratio of semantically loaded tokens to total prompt tokens, adjusted for redundancy and concreteness. Unlike prior prompt optimization techniques that add tokens (Chain of Thought), duplicate the prompt (Prompt Repetition), or reorder components (Instruction Placement Effect), SDE improves performance by removing or replacing low-information tokens while preserving or sharpening the semantic signal. Evaluated across five frontier models and seven benchmarks, ultra-dense prompts (SDE > 0.80) outperform diluted counterparts by an average of +8.4 percentage points with 0 additional tokens and 0 latency overhead. Combined with Instruction Placement Effect (IPE), the gain reaches +11.7 percentage points

Manual annotation of high-quality visual question answering with grounding (VQA-G) datasets, which pair visual questions with evidential grounding, is crucial for advancing vision-language models (VLMs), but remains unscalable. Existing automated methods are often hindered by two key issues: (1) inconsistent data fidelity due to model hallucinations; (2) brittle verification mechanisms based on simple heuristics. To address these limitations, we introduce AutoVQA-G, a self-improving agentic framework for automated VQA-G annotation. AutoVQA-G employs an iterative refinement loop where a Consistency Evaluation module uses Chain-of-Thought (CoT) reasoning for fine-grained visual verification. Based on this feedback, a memory-augmented Prompt Optimization agent analyzes critiques from failed samples to progressively refine generation prompts. Our experiments show that AutoVQA-G generates VQA-G datasets with superior visual grounding accuracy compared to leading multimodal LLMs, offering a promising approach for creating high-fidelity data to facilitate more robust VLM training and evaluation. Code: https://github.com/rohnson1999/AutoVQA-G

Retrieval-Augmented Generation (RAG) is a key approach to mitigating the temporal staleness of large language models (LLMs) by grounding responses in up-to-date evidence. Within the RAG pipeline, re-rankers play a pivotal role in selecting the most useful documents from retrieved candidates. However, existing benchmarks predominantly evaluate re-rankers in static settings and do not adequately assess performance under evolving information -- a critical gap, as real-world systems often must choose among temporally different pieces of evidence. To address this limitation, we introduce FRESCO (Factual Recency and Evolving Semantic COnflict), a benchmark for evaluating re-rankers in temporally dynamic contexts. By pairing recency-seeking queries with historical Wikipedia revisions, FRESCO tests whether re-rankers can prioritize factually recent evidence while maintaining semantic relevance. Our evaluation reveals a consistent failure mode across existing re-rankers: a strong bias toward older, semantically rich documents, even when they are factually obsolete. We further investigate an instruction optimization framework to mitigate this issue. By identifying Pareto-optimal instructions that balance Evolving and Non-Evolving Knowledge tasks, we obtain gains of up to 27% on Evolving Knowledge tasks while maintaining competitive performance on Non-Evolving Knowledge tasks.

Knowledge graph embedding (KGE) models perform well on link prediction but struggle with unseen entities, relations, and especially literals, limiting their use in dynamic, heterogeneous graphs. In contrast, pretrained large language models (LLMs) generalize effectively through prompting. We reformulate link prediction as a prompt learning problem and introduce RALP, which learns string-based chain-of-thought (CoT) prompts as scoring functions for triples. Using Bayesian Optimization through MIPRO algorithm, RALP identifies effective prompts from fewer than 30 training examples without gradient access. At inference, RALP predicts missing entities, relations or whole triples and assigns confidence scores based on the learned prompt. We evaluate on transductive, numerical, and OWL instance retrieval benchmarks. RALP improves state-of-the-art KGE models by over 5% MRR across datasets and enhances generalization via high-quality inferred triples. On OWL reasoning tasks with complex class expressions (e.g., $\exists hasChild.Female$, $\geq 5 \; hasChild.Female$), it achieves over 88% Jaccard similarity. These results highlight prompt-based LLM reasoning as a flexible alternative to embedding-based methods. We release our implementation, training, and evaluation pipeline as open source: https://github.com/dice-group/RALP .

Visual Foundation Models (VFMs) such as the Segment Anything Model (SAM) have significantly advanced broad use of image segmentation. However, SAM and its variants necessitate substantial manual effort for prompt generation and additional training for specific applications. Recent approaches address these limitations by integrating SAM into in-context (one/few shot) segmentation, enabling auto-prompting through semantic alignment between query and support images. Despite these efforts, they still generate sub-optimal prompts that degrade segmentation quality due to visual inconsistencies between support and query images. To tackle this limitation, we introduce PR-MaGIC (Prompt Refinement via Mask Decoder Gradient Flow for In-Context Segmentation), a training-free test-time framework that refines prompts via gradient flow derived from SAM's mask decoder. PR-MaGIC seamlessly integrates into in-context segmentation frameworks, being theoretically grounded yet practically stabilized through a simple top-1 selection strategy that ensures robust performance across samples. Extensive evaluations demonstrate that PR-MaGIC consistently improves segmentation quality across various benchmarks, effectively mitigating inadequate prompts without requiring additional training or architectural modifications.

Can large language models (LLMs) generate continuous numerical features that improve reinforcement learning (RL) trading agents? We build a modular pipeline where a frozen LLM serves as a stateless feature extractor, transforming unstructured daily news and filings into a fixed-dimensional vector consumed by a downstream PPO agent. We introduce an automated prompt-optimization loop that treats the extraction prompt as a discrete hyperparameter and tunes it directly against the Information Coefficient - the Spearman rank correlation between predicted and realized returns - rather than NLP losses. The optimized prompt discovers genuinely predictive features (IC above 0.15 on held-out data). However, these valid intermediate representations do not automatically translate into downstream task performance: during a distribution shift caused by a macroeconomic shock, LLM-derived features add noise, and the augmented agent under-performs a price-only baseline. In a calmer test regime the agent recovers, yet macroeconomic state variables remain the most robust driver of policy improvement. Our findings highlight a gap between feature-level validity and policy-level robustness that parallels known challenges in transfer learning under distribution shift.

Ambiguity is an inherent property of natural-language agent specifications. When a system prompt leaves behaviour underdetermined, identical inputs follow divergent execution paths and produce inconsistent outcomes. The standard remedy is prompt optimisation: propose candidate prompts, run the agent to score them, and keep the best. This loop pays for the agent twice: once to generate candidates and again to rank them. On a tool-using agent whose rollouts cost dollars and minutes, the ranking cost dominates and budget-constrained optimisers routinely fail to find improvements. We present Semantic Feature Analysis (SFA), a pipeline that repairs agent specifications without running any search. SFA reads execution traces the agent has already produced, clusters the outputs of each workflow node, decomposes them into semantic feature classes using an extended subject-verb-object schema, ranks those features by their contribution to outcome separation using a decision tree, and injects the surviving features as corrective statements into the affected node's system prompt. Because it never ranks candidate prompts, it never spends a rollout on selection. We evaluate SFA against five prompt optimisers (GEPA, MIPROv2, SIMBA, BootstrapFewShot with random search, InferRules) and a single-reflection control, budget-matched in dollars at three budget levels across four benchmarks (IF-Bench, HotpotQA, HoVer, and GAIA). SFA consistently improves over the unmodified agent across benchmarks and budget levels, with the largest gains where rollouts are most expensive. On GAIA, where budget-constrained optimisers cannot afford to score even one candidate, SFA improves accuracy while other arms return their seed unchanged.

Text-to-image diffusion models exhibit strong generative performance but remain highly sensitive to prompt formulation, often requiring extensive manual trial and error to obtain satisfactory results. This motivates the development of automated, model-agnostic prompt optimization methods that can systematically explore the conditioning space beyond conventional text rewriting. This work investigates the use of a Genetic Algorithm (GA) for prompt optimization by directly evolving the token vectors employed by CLIP-based diffusion models. The GA optimizes a fitness function that combines aesthetic quality, measured by the LAION Aesthetic Predictor V2, with prompt-image alignment, assessed via CLIPScore. Experiments on 36 prompts from the Parti Prompts (P2) dataset show that the proposed approach outperforms the baseline methods, including Promptist and random search, achieving up to a 23.93% improvement in fitness. Overall, the method is adaptable to image generation models with tokenized text encoders and provides a modular framework for future extensions, the limitations and prospects of which are discussed.

This paper studies Automated Instruction Revision (AIR), a rule-induction-based method for adapting large language models (LLMs) to downstream tasks using limited task-specific examples. We position AIR within the broader landscape of adaptation strategies, including prompt optimization, retrieval-based methods, and fine-tuning. We then compare these approaches across a diverse benchmark suite designed to stress different task requirements, such as knowledge injection, structured extraction, label remapping, and logical reasoning. The paper argues that adaptation performance is strongly task-dependent: no single method dominates across all settings. Across five benchmarks, AIR was strongest or near-best on label-remapping classification, while KNN retrieval performed best on closed-book QA, and fine-tuning dominated structured extraction and event-order reasoning. AIR is most promising when task behavior can be captured by compact, interpretable instruction rules, while retrieval and fine-tuning remain stronger in tasks dominated by source-specific knowledge or dataset-specific annotation regularities.

Large language model agents often exhibit complementary strengths, making routing a promising approach for multi-agent question answering. However, existing routing methods remain limited in two important ways: they typically optimize over a fixed pool of agents without improving the agents themselves, and they often rely on rigid collaboration schemes that cannot adapt the number of participating agents to the query. We propose EvolveRouter, a trainable framework that addresses both limitations by jointly improving agent quality and collaboration structure. First, EvolveRouter couples graph-based query routing with targeted instruction refinement in a closed-loop co-evolution process, allowing router diagnostics to guide agent improvement while refined agents provide cleaner supervision for routing. Second, it introduces an adaptive inference strategy that dynamically determines the effective collaboration size for each query through router-weighted answer agreement. Together, these designs enable more capable and more efficient multi-agent reasoning. Experiments on five question answering benchmarks show that EvolveRouter consistently outperforms SOTA routing baselines in both F1 and exact match, while further analysis confirms the benefits of closed-loop refinement and adaptive collaboration.

Large Language Models (LLMs) have shown strong performance across a wide range of natural language processing tasks; however, their effectiveness is highly dependent on prompt design, structure, and embedded reasoning signals. Conventional prompt engineering methods largely rely on heuristic trial-and-error processes, which limits scalability, reproducibility, and generalization across tasks. DSPy, a declarative framework for optimizing text-processing pipelines, offers an alternative approach by enabling automated, modular, and learnable prompt construction for LLM-based systems.This paper presents a systematic study of DSPy-based declarative learning for prompt optimization, with emphasis on prompt synthesis, correction, calibration, and adaptive reasoning control. We introduce a unified DSPy LLM architecture that combines symbolic planning, gradient free optimization, and automated module rewriting to reduce hallucinations, improve factual grounding, and avoid unnecessary prompt complexity. Experimental evaluations conducted on reasoning tasks, retrieval-augmented generation, and multi-step chain-of-thought benchmarks demonstrate consistent gains in output reliability, efficiency, and generalization across models. The results show improvements of up to 30 to 45% in factual accuracy and a reduction of approximately 25% in hallucination rates. Finally, we outline key limitations and discuss future research directions for declarative prompt optimization frameworks.

Masked diffusion language models (dLMs) generate text through bidirectional denoising, yet this capability remains locked for infilling prompts. This limitation is an artifact of the current supervised finetuning (SFT) convention of applying response-only masking. To unlock this capability, we extend full-sequence masking during SFT, where both prompts and responses are masked jointly. Once unlocked, the model infills masked portions of a prompt template conditioned on few-shot examples. We show that such model-infilled prompts match or surpass manually designed templates, transfer effectively across models, and are complementary to existing prompt optimization methods. Our results suggest that training practices, not architectural limitations, are the primary bottleneck preventing masked diffusion language models from infilling effective prompts

Given a user's complex information need, a multi-agent Deep Research system iteratively plans, retrieves, and synthesizes evidence across hundreds of documents to produce a high-quality answer. In one possible architecture, an orchestrator agent coordinates the process, while parallel worker agents execute tasks. Current Deep Research systems, however, often rely on hand-engineered prompts and static architectures, making improvement brittle, expensive, and time-consuming. We therefore explore various multi-agent optimization methods to show that enabling agents to self-play and explore different prompt combinations can produce high-quality Deep Research systems that match or outperform expert-crafted prompts.

Multi-agent Retrieval-Augmented Generation (RAG), wherein each agent takes on a specific role, supports hard queries that require multiple steps and sources, or complex reasoning. Existing approaches, however, rely on static agent behaviors and fixed orchestration strategies, leading to brittle performance on diverse, multi-hop tasks. We identify two key limitations: the lack of continuously adaptive orchestration mechanisms and the absence of behavior-level learning for individual agents. To this end, we propose HERA, a hierarchical framework that jointly evolves multi-agent orchestration and role-specific agent prompts. At the global level, HERA optimizes query-specific agent topologies through reward-guided sampling and experience accumulation. At the local level, Role-Aware Prompt Evolution refines agent behaviors via credit assignment and dual-axes adaptation along operational and behavioral principles, enabling targeted, role-conditioned improvements. On six knowledge-intensive benchmarks, HERA achieves an average improvement of 38.69% over recent baselines while maintaining robust generalization and token efficiency. Topological analyses reveal emergent self-organization, where sparse exploration yields compact, high-utility multi-agent networks, demonstrating both efficient coordination and robust reasoning.

The large-scale adoption of Large Language Models (LLMs) forces a trade-off between operational cost (OpEx) and data privacy. Current routing frameworks reduce costs but ignore prompt sensitivity, exposing users and institutions to leakage risks towards third-party cloud providers. We formalise the "Inseparability Paradigm": advanced context management intrinsically coincides with privacy management. We propose a local "Privacy Guard" -- a holistic contextual observer powered by an on-premise Small Language Model (SLM) -- that performs abstractive summarisation and Automatic Prompt Optimisation (APO) to decompose prompts into focused sub-tasks, re-routing high-risk queries to Zero-Trust or NDA-covered models. This dual mechanism simultaneously eliminates sensitive inference vectors (Zero Leakage) and reduces cloud token payloads (OpEx Reduction). A LIFO-based context compacting mechanism further bounds working memory, limiting the emergent leakage surface. We validate the framework through a 2x2 benchmark (Lazy vs. Expert users; Personal vs. Institutional secrets) on a 1,000-sample dataset, achieving a 45% blended OpEx reduction, 100% redaction success on personal secrets, and -- via LLM-as-a-Judge evaluation -- an 85% preference rate for APO-compressed responses over raw baselines. Our results demonstrate that Token Parsimony and Zero Leakage are mathematically dual projections of the same contextual compression operator.

As Large Language Models (LLMs) achieve increasingly sophisticated performance on complex reasoning tasks, current architectures serve as critical proxies for the internal heuristics of frontier models. Characterizing emergent reasoning is vital for long-term interpretability and safety. Furthermore, understanding how prompting modulates these processes is essential, as natural language will likely be the primary interface for interacting with AGI systems. In this work, we use a custom variant of Genetic Pareto (GEPA) to systematically optimize prompts for scientific reasoning tasks, and analyze how prompting can affect reasoning behavior. We investigate the structural patterns and logical heuristics inherent in GEPA-optimized prompts, and evaluate their transferability and brittleness. Our findings reveal that gains in scientific reasoning often correspond to model-specific heuristics that fail to generalize across systems, which we call "local" logic. By framing prompt optimization as a tool for model interpretability, we argue that mapping these preferred reasoning structures for LLMs is an important prerequisite for effectively collaborating with superhuman intelligence.

Multi-Modal LLMs (MLLMs) demonstrate strong visual grounding capabilities on popular object detection benchmarks like OdinW-13 and RefCOCO. However, state-of-the-art models still struggle to generalize to out-of-distribution classes, tasks and imaging modalities not typically found in their pre-training. While in-context prompting is a common strategy to improve performance across diverse tasks, we find that it often yields lower detection accuracy than prompting with class names alone. This suggests that current MLLMs cannot yet effectively leverage few-shot visual examples and rich textual descriptions for object detection. Since frontier MLLMs are typically only accessible via APIs, and state-of-the-art open-weights models are prohibitively expensive to fine-tune on consumer-grade hardware, we instead explore black-box prompt optimization for few-shot object detection. To this end, we propose Detection Prompt Optimization (DetPO), a gradient-free test-time optimization approach that refines text-only prompts by maximizing detection accuracy on few-shot visual training examples while calibrating prediction confidence. Our proposed approach yields consistent improvements across generalist MLLMs on Roboflow20-VL and LVIS, outperforming prior black-box approaches by up to 9.7 mAP. Our code and optimized prompts are available at https://ggare-cmu.github.io/DetPO/

Reinforcement Learning with Verifiable Rewards (RLVR) enhances Large Language Model (LLM) reasoning but is suffer from advantage collapse: when all rollouts of a query receive identical rewards, the group variance vanishes, most damagingly on hard samples, where scaling rollout budgets yields little. We introduce Joint Policy and Prompt Optimization (P O) to mitigate this collapse by alternating continuous policy updates with discrete prompt evolution. P O mines hard samples with a success-rate threshold, evolves reasoning prompts for them with GEPA, and internalizes the elicited trajectories via context distillation, which optimizes each trajectory under the original query and thus removes inference-time prompting, with a Context Ratio Mask (CRM) filtering out extreme likelihood ratios. P O restores critical advantage signals and surpasses the GRPO baseline by up to 8.2 points in average accuracy on six held-out benchmarks across all training datasets and backbones, while also outperforming DAPO and other baselines. The gains are especially pronounced on hard benchmarks, reaching up to 16.3 points above GRPO on average across AIME24 and AIME25. Our findings expose the limits of standard exploration in sparse-reward environments, illuminating the potential of unifying evolutionary algorithms with reinforcement learning. This integration of discrete semantic search and continuous parameter updates provides a self-reinforcing framework that facilitates more effective LLM alignment.