Papers

Large software systems often suffer from time inefficiencies that cause excessive execution time despite functional correctness. Localizing their fix locations is difficult because, unlike functional bugs, they produce neither test failures nor stack-trace clues, making traditional and recent LLM-based fault localization methods unsuitable. Runtime profiling provides alternative evidence but faces three challenges in repository-level settings: single-run profiling cannot reliably distinguish inefficiency hotspots from execution noise; existing profilers struggle to extract relevant execution paths from extensive background execution; and a semantic gap remains between observed hotspots and actual fix locations. We propose EffiHolmes, an LLM-based framework for repository-level time inefficiency fix localization. EffiHolmes uses differential profiling under default and scaled workloads to identify inefficiency hotspots, extracts compact execution paths connecting these hotspots to the reported inefficient function, and employs domain-guided LLM reasoning to locate the underlying inefficiency logic. We also introduce RepoEffi-Bench, the first benchmark for repository-level inefficiency localization, containing 140 high-quality issues collected from popular Python repositories. Experiments show that EffiHolmes consistently outperforms state-of-the-art retrieval-, agent-, and profiling-based baselines, improving file-level Acc@3 by 4.29 percentage points with GPT-5.1 and function-level Acc@5 by 15.00 percentage points with qwen3-4b. It also remains robust across model capacities.

Large Language Models are increasingly evaluated for code generation using test-based benchmarks. The validity of such evaluations depends on the reliability of their references and tests, while test-based correctness captures only part of the observable properties of generated code. We present CodeAssay, a taxonomy-first benchmark of 185 Python tasks across ten software-engineering categories. It combines audited ground truth, public tests for generation and repair, hidden tests for grading, mutation-based test-suite validation, and selected code-property measures. Regrading fixed model outputs after the audit changed 170 of 1,890 correctness labels (9.0%) and increased the measured best-to-worst model spread from 11.9 to 23.7 percentage points, although aggregate correctness remained nearly unchanged. The complete and hidden test suites achieved mutation scores of 82.6% and 74.8%, respectively. Across seven proprietary LLMs, standard-prompt correctness ranged from 77.3% to 98.9%, with significant differences in 12 of 21 model pairs. On the 120 tasks solved by all 14 model-prompt configurations, no model performed best across all selected code properties. A security-focused prompt produced no significant change in correctness or consistent reduction in the selected static-analysis findings, while increasing program length and cyclomatic complexity across all models. These findings show that reliable evaluation of LLM-generated code requires validated ground truth, protected tests, and multiple explicitly interpreted measures. CodeAssay provides a reproducible basis for evidence-based model evaluation in AI-augmented software development.

Combining Bayesian learning and quantitative verification is a powerful toolset for analysing key quantitative properties of software systems, like reliability and response time. However, the accuracy and robustness of verification results strongly depend on the prior knowledge (PK) underlying Bayesian inference. This knowledge reflects original beliefs about the probability of events and typically depends on domain expertise. Using inaccurate or uninformative PK can negatively affect quantitative analysis, yielding incorrect verification results. Our EPIK approach tackles this important challenge by eliciting and embedding PK in quantitative verification equipped with Bayesian estimators. Unlike existing approaches that require PK on formal model transition parameters, EPIK leverages system-level properties that are directly observable and are linked to real-world semantics. EPIK formulates a twofold optimisation problem to derive the distributions of unknown transition parameters and then embeds these distributions to verify new or difficult-to-measure (elusive) properties. The detailed experimental evaluation using multiple variants of real-world case studies and diverse EPIK instantiations shows its effectiveness, flexibility and generality.

Context: The growing adoption of AI-assisted development tools is changing how software teams collaborate, share knowledge, and coordinate, yet its consequences for team social dynamics remain largely unexplored. Gap: It is unclear whether AI adoption is associated with an increase or reduction in community smells,socio-technical anti-patterns reflecting coordination and communication breakdowns,and through which mechanisms. Method: Grounded in Transactive Memory Systems (TMS) theory, we validate instruments for HumanAI and HumanHuman interaction along two TMS dimensions, Specialization and Coordination, and test five PLS-SEM models on survey data from 152 software professionals using AI tools. Community smell constructs were derived from the literature and validated through expert surveys and factor analysis. Results: AI adoption relates to community smells not in a single way, but through mechanisms depending on the work. In specialization work, AI is associated with higher knowledge-sharing peer interaction, which is in turn associated with fewer smells. In coordination work, AI is directly associated with higher communication quality, complementing rather than replacing human interaction. Contributions: We provide an empirically validated, TMS-grounded model showing that the AIcommunity-smell relationship is contingent on the type of collaboration, with a reusable instrument and evidence-based implications for research and practice.

Large language models are increasingly embedded in software engineering workflows as coding agents that can inspect repositories, invoke tools, execute tests, debug failures, and generate patches. Yet most existing coding agents remain largely static after deployment, even though software development is a dynamic, feedback-rich process in which repositories evolve, dependencies change, tests fail, and repair attempts leave reusable experience. This tension has motivated a growing body of work on self-evolving coding agents, where the agent improves its future behavior by persistently updating its framework, memory, skills and tools, components, workflow and topology, or environment and context from prior coding interactions. In this survey, we provide a structured synthesis of this emerging area. We first define the concept of self-evolving coding agents and distinguish it from conventional coding agents and general self-evolving agents. We then develop a taxonomy centered on the targets of evolution, complemented by two orthogonal perspectives: when evolution occurs and which code-specific signals drive it. We further examine the benchmarks used to measure the effect of evolution and related coding products. Across the literature, we find that executable feedback, repository-level context, and coding trajectories make software engineering a natural domain for agent self-evolution, but also introduce challenges in feedback reliability, benchmark overfitting, reversibility, system complexity, safety, cost, and generalization. By organizing existing work around these dimensions, this survey aims to clarify the conceptual boundaries of self-evolving coding agents and provide a foundation for designing more adaptive, reliable, and software-aware agentic systems.

Large language models (LLMs) have substantially improved code generation, yet achieving strong functional correctness remains difficult, especially for heterogeneous programming tasks where a single prompting strategy and a single directly generated output are often insufficient. In this paper, we present RAV, a lightweight and modular framework that improves code generation with a fixed backbone model through three coordinated stages: Route, which applies task-aware prompt routing before generation; Align, which reduces the mismatch between fine-tuning prompts and inference-time prompts through aligned LoRA adaptation; and Verify, which selects the final output by executing multiple candidates against visible public tests. We evaluate RAV on the MBPP benchmark under both the sanitized and full settings. The complete RAV pipeline achieves the best performance among all evaluated configurations, reaching 0.8911 on MBPP Sanitized and 0.8520 on MBPP Full. Compared with the base model, these results represent improvements of 6.35 and 9.92 percentage points, respectively. Component-wise ablation experiments further show that task-aware routing and aligned adaptation become substantially more effective when combined with execution-based verification. Additional robustness and contamination analyses support the reliability of the observed improvements. Overall, the results indicate that functional correctness in code generation can be meaningfully improved without modifying the backbone architecture, by jointly optimizing how tasks are prompted, how the model is adapted, and how final outputs are selected.

UI component libraries are commonly assessed using execution-based metrics such as statement and branch coverage, yet these metrics provide limited insight into whether tests verify the behavioral relations implied by component APIs and documentation. This paper presents an MR-based framework that uses inferred metamorphic relations (MRs) as an empirical behavioral reference, rather than a complete specification, for assessing UI component test suites. Given a component's source, documentation, and tests, the framework infers component-specific MRs using a UI-specific taxonomy, aligns tests with the inferred relations through hybrid deterministic and semantic analysis, and computes relation-level MR coverage metrics. We manually validate both the inferred MR space and the test–MR alignment. Our evaluation shows that existing test suites exercise substantially more behavioral relations than they explicitly validate: MR Cover remains between 42.5% and 47.6% across three LLM configurations and consistently below MR Touch. Most uncovered relations are weak-oracle cases, where behaviors are exercised but lack explicit behavioral validation. MR coverage also complements execution-based coverage by revealing behavioral gaps not reflected by statement or branch coverage alone. We further assess practical relevance through issue-description mapping, oracle strengthening, and MR-relevant injected faults. Most reported issue descriptions can be mapped to inferred MR relation types; weak-oracle relations often expose missing validation evidence; and MR labels show a trend in MR-relevant fault detection. Overall, MR coverage provides a complementary relation-level perspective for assessing behavioral validation in modern UI component testing.

Generative AI is rapidly reshaping Open Source Software (OSS) software development,prompting projects to introduce policies governing AI-assisted contributions. However, little is known about how these policies differ or whether they influence developer experience. We present the first large-scale empirical study of AI governance policies in OSS. Analyzing 29,624 GitHub repositories, we identify 385 projects that adopted AI policies and derive TRACE, a framework capturing five governance dimensions: Transparency, Responsibility, Attribution, Constraints, and Enforcement. We further classify policies into five governance families and estimate their effects using propensity-score matching and longitudinal difference-in-differences analysis. Our results show that AI governance primarily regulates rather than prohibits AI-assisted development. Policy adoption brings maintainer engagement, increased AI disclosure, richer review interactions, and improved code quality while AI-assisted contributions continue to grow. Governance design matters: policies emphasizing transparency and responsibility produced stronger community and quality outcomes than restrictive approaches alone. Our findings show how different AI governance strategies shape developer experience and provide evidence to help OSS communities design effective AI policies.

Enterprise compliance management requires rapid adaptation to evolving regulatory frameworks (e.g., DORA, AI RMF, FedRAMP) and tight remediation SLAs. Traditional static orchestrators often fail in hybrid cloud environments where event-driven assessments demand that automation code adapt to runtime context in seconds. This paper presents lessons learned from evaluating six large language models for AI-driven workflow generation in a production enterprise platform, benchmarked across 29 real-world IT automation scenarios, two generation pipeline architectures, and eight independent runs per prompt-model-pipeline configuration (2,784 runs total). Our initial pipeline used monolithic workflow generation, achieving 31.5-82.8% structural success rates (JSON schema validity and correct UI rendering), with most models struggling on complex JSON generation. We developed a redesigned piecewise pipeline that decomposes workflow construction into variable scaffolding, base block assembly, and nested block generation, raising structural success to 74.1-97.8% across all models. We analyze production tradeoffs including cost (USD 0.008-0.20 per workflow), latency (under 50s for interactive use), and model selection. Piecewise decomposition enables smaller models (e.g., mistral-small at 95.7% structural success and USD 0.01 per workflow) to reach production viability, removing dependency on expensive frontier models. While mistral-medium-2505 and gpt-oss-120b achieved the highest structural success (96.1% and 97.8%), mistral-medium-2505 carries a 19x cost premium versus mistral-small. Our deployment lessons highlight the need to separate structural validity from semantic correctness (logical fulfillment of user intent) and provide a solution for model-agnostic, scalable automation in cloud engineering.

With WebAssembly widely supported in browsers, containers, IoT devices, and serverless platforms and increasingly adopted as a universal low-level bytecode standard, auditing its hidden vulnerabilities and malicious intentions has become critical. Decompiling existing WebAssembly modules can help security researchers and end users understand binary behavior, but current tools suffer from verbose result, poor readability, and limited type recovery. We present NotDec, an advanced WebAssembly decompilation framework. NotDec extends the WebAssembly type checking algorithm to lift bytecode into an SSA-based IR, applies the inter-procedural type recovery algorithm Retypd with pointer and numeric value differentiation methods to recover complex data structures, and leverages Memory SSA alongside semantics-preserving structured control-flow analysis to emit readable, semantically consistent C code. NotDec achieves 100% recompilation success rate on all 5,241 Juliet samples and all Howard dataset programs, significantly outperforming baselines including Ghidra (45.95% success rate). On type recovery accuracy, NotDec recovers 85.33% of struct member accesses in real-world programs, vastly exceeding Ghidra's 9.24%. While the full inter-procedural version faces scalability challenges on large binaries, the intra-procedural variant NotDec_F demonstrates superior efficiency, consuming less than half of Ghidra's memory and up to 97% less execution time on unoptimized binaries.

Software engineering (SWE) agents resolve repository-level issues through long trajectories that grow increasingly expensive as context accumulates. Failed runs tend to be longer and exhibit redundant exploration or looping, suggesting that some failures may be detectable before completion. Early termination, however, risks interrupting trajectories that would otherwise succeed; conversely, an unsuccessful trajectory may still contain useful repository edits. We present FailFast-RestartSmart, a two-stage controller for a single active trajectory. FailFast is a lightweight 0.6B monitor trained with terminal and dense fail-to-pass supervision to predict failure from observable prefixes without policy logits or hidden states. Upon an alarm, RestartSmart launches a fresh same-policy rollout without prior prompt history and offers the interrupted repository diff as an optional overlay that the agent may inspect, apply, or discard. On SWE-bench Verified, a monitor trained solely on Qwen3.6-27B trajectories transfers to three other policies, including a closed-API model, and saves 14.6%-20.4% of execution tokens at a target 5% false-positive rate; on Qwen3.6-27B, its 20.4% saving exceeds the 12.5% achieved by our per-step AgentStop adaptation. At a target 25% false-positive rate, RestartSmart raises Qwen3.6-27B resolution from 66.6% to 71.8%, whereas cold restart reaches only 66.8%. Together, these results support early termination with sequential same-policy recovery.

Quantum software engineering faces unique challenges due to the interaction of classical and quantum components, which produce complex and often poorly understood bug patterns. Characterizing these bugs is essential for advancing testing, debugging, and quality assurance in quantum ecosystems. This paper presents a comparative study of 12,910 issues from Qiskit and 4,613 issues from 11 additional repositories, including Cirq and PyQuil. Using a rule-based classification framework, we analyze bugs by type, category, severity, quality attributes, and quantum-specific subtypes. Results show that classical bugs consistently dominate (67%) across ecosystems, while quantum-specific bugs account for 27-30%. Ecosystem-specific trends emerge: Qiskit repositories exhibit more compatibility related bugs, whereas other ecosystems show higher syntax and quantum-specific bug rates. Across both ecosystems, gate and circuit issues dominate quantum-specific bugs, though non-Qiskit projects reveal broader diversity, including algorithmic, resource, and hybrid-interface issues. Statistical validation confirms that the framework generalizes at the bug-type level while detecting significant variations at finer levels. Benchmarking against four supervised machine-learning baselines further shows that the rule-based framework consistently outperforms data-driven models, particularly for fine-grained quantum-specific subtypes, while longitudinal analysis (2017-2025) indicates that quantum- specific bugs remain relatively stable over time rather than exhibiting a steady increase. This study provides the first cross- ecosystem comparison of bug distributions in quantum software, demonstrating the utility of an interpretable, automation-ready, rule-based framework for guiding testing, debugging, and quality assurance.

Purpose: Developers choose how much a language-model agent reasons before it acts. Some pick a high level for fear that a low one is not enough, and pay for it in tokens, time and overthinking. We tested whether a low level is enough for routine office work. Methods: Two tiers of GPT-5.6 did 14 routine office tasks at the low and the max reasoning level, 840 runs in all. To make the tasks harder, each one sets a target that the rules make impossible to reach and offers a forbidden tool that would reach it. Some versions also tell the agent that the forbidden tool counts. A program checked every run from its final state. Results: At both levels the agent always followed the rules and never used the forbidden tool, including 30 runs in which it was told that the tool counts while it could still have used it. The low level used about 43% fewer output tokens and 20% less time. Conclusion: For routine office work, a low reasoning level is enough.

Detecting source code vulnerabilities is increasingly difficult as modern security flaws are rooted in complex causal dependencies between execution flows, control conditions, and program states. Despite recent advances in Large Language Models (LLMs) and multi-agent frameworks, existing approaches primarily address superficial similarities between benign and vulnerable functions while failing to capture the complex causal dependencies inherent in security flaws. To address these limitations, we propose Causal Context-based Agentic Reasoning (CLEAR), a novel multi-agent vulnerability detection framework integrated with a causal knowledge graph. CLEAR systematically constructs a Vulnerability Causal Knowledge Graph (VCKG) that models the causal chains between entrypoints, preconditions, root causes, and fix intents across vulnerability instances. Leveraging this structured knowledge, four specialized agents, including the Collector, Claim, Critic, and Judge, collaboratively verify vulnerability hypotheses through retrieved causal contexts. Experimental results on C/C++ and Java vulnerability benchmarks demonstrate that CLEAR improves Pair-Correct (P-C) performance by 130.7% and 71.56% over state-of-the-art approaches, demonstrating the effectiveness of causal knowledge graph-guided reasoning for automated vulnerability detection.

LLMs are widely used to generate structured output like source code or JSON. Grammar-constrained decoding (GCD) can guarantee the syntactic validity of the generated output, by masking out tokens that violate rules specified by a context-free grammar. However, the online computational overhead of existing GCD methods, with latency typically scaling linearly with vocabulary size, limits the throughput of LLMs, especially for models with large vocabularies. To address this issue, we propose PSC, a novel grammar-constrained decoding method. By combining acceptance conditions of all vocabulary tokens into a single classifier of the parser stack during preprocessing, PSC can compute the complete vocabulary mask by checking the parser stack exactly once per decoding step, with time complexity independent of the vocabulary size. Experiments show that PSC computes masks up to 700$\times$ faster than baselines on complex programming language grammars, and up to 30$\times$ faster for schema-conformant JSON; end-to-end LLM throughput with PSC approaches that of unconstrained decoding. We analyze the preprocessing overhead for preprocessing providers and decoding users, and provide a break-even point analysis to help users decide whether to do preprocessing by themselves.

LLM-based CAD agents produce executable parametric programs, but their correction loops may lose evidence about satisfied requirements, faulty operations, and prior repairs. We introduce TraceCAD, a recovery layer that links requested features, modeling steps, failure evidence, and candidate outcomes as persistent state. TraceCAD diagnoses likely faulty operations, searches bounded edits in their dependency regions, validates candidates through execution and preservation checks, and retains successful and failed repair outcomes in reusable skill memory. On DeepCAD-derived benchmarks with 200-model ablations and a 1K-model comparison, TraceCAD achieves competitive geometric quality in terms of IoU, Chamfer distance, and Hausdorff distance. Removing persistent state nearly halves recovery score; removing localized search more than doubles geometric regression and doubles code-agent invocations. Initializing the skill store on disjoint training models further reduces retries, token cost, and latency. These results demonstrate that persistent, localized, and reusable recovery improves final CAD quality and repair reliability.

Self-Sovereign Identity (SSI) promises to decentralize digital identity, but widespread adoption remains hindered by integration complexity and tooling immaturity. This paper investigates the Developer Experience (DX) of open-source SSI tooling through an exploratory probe study. Nine developers with prior knowledge of decentralized identity concepts, representing early integrators building SSI applications, attempted core credential lifecycle tasks using Walt.id, Traction, and MetaMask. Our goal was to surface recurring integration barriers through qualitative thematic analysis of open-ended developer reports, complemented by task-level difficulty ratings. Our findings reveal a critical abstraction gap: while passive operations like credential receipt are relatively mature, active construction tasks, particularly schema customization, expose significant architectural friction. We identify that these barriers stem from inadequate API abstractions, brittle environment configurations, and documentation that fails to track the ecosystem's rapid evolution. These issues reflect structural design decisions in current frameworks. This study characterizes the structural integration barriers in the current SSI open-source ecosystem. To address the identified gaps, we propose three architectural shifts for developer tooling: web-based sandboxes, AI-assisted schema generators, and executable documentation strategies.

Large Language Models (LLMs) are integrated into software systems and AI services, making efficient LLM serving a concern for software engineering. Serving LLMs is challenging because inference requires computation, memory, GPU resources, and execution while maintaining latency and throughput. Although prior research has proposed LLM inference, optimization, and serving techniques and frameworks, little is known about how they are adopted in practice. In this study, we investigate the use of LLM serving frameworks and serving methods in open-source software systems. We identify and analyze five LLM-specific frameworks: vLLM, SGLang, TensorRT-LLM, LMDeploy, and FlashInfer. We examine how these frameworks and techniques are adopted individually and in combination, how adoption varies across categories of LLMs, and how repositories differ in intent, focus, use case, and architectural design. Our results show that vLLM is the most visible framework in popularity and adoption, while parallel computation, memory management, and network pruning are the most frequently used serving-method categories. Multi-framework usage is limited, suggesting that developers rely on a single serving framework; however, combined frameworks connect complementary capabilities across the serving stack. Framework adoption varies across model families, modalities, model sizes, domain specializations, and deployment settings. Repository-level analysis shows that LLM serving frameworks support applications and architectures, including Reinforcement Learning (RL)-based reasoning, multimodal generation and understanding, microservices, and cloud infrastructure. Overall, this study provides a large-scale empirical characterization of LLM serving framework adoption in practice and offers insights for researchers, framework maintainers, and practitioners working on LLM systems.

For most of the past six decades, software measurement relied on labor-intensive manual collection of proprietary data, which hampered progress. The shift to repurposing traces from version control and related tools dramatically expanded data availability$\unicode{x2014}$especially with the rise of open-source software$\unicode{x2014}$but hinged on an often unstated assumption: that these tools are used by professional developers to build genuine software systems. However, as trace-generating tools, data types and scale, and empirical methods have all evolved, it has become clear that changes in data generation and analytical approaches affect many prior findings about software development, maintenance, and evolution. With AI agents now actively using these same tools, the resulting traces frequently violate the original assumption of human origin. To preserve the relevance of software measurement research, immediate action is needed: We must detect when foundational assumptions are violated in contemporary data and develop new methodologies that remain valid under changed circumstances. To this end, we propose a systematic AI-assisted replication program that revisits key findings using modern techniques, aiming for methods that yield consistent results on current data to keep software measurement meaningful.

Localizing concurrent bugs from bug reports alone is challenging due to incomplete information, misleading program-entity mentions, and complex cross-thread interactions, causing existing LLM-based approaches to suffer from unstable reasoning and limited explainability. We propose ConFL, an explainable concurrent fault localization framework that augments LLM reasoning with structured concurrency knowledge. ConFL constructs a Concurrent Knowledge Base (CKB) from source code and performs LLM-guided hierarchical retrieval to progressively narrow the search space from components to interaction-level concurrency contexts. An interaction-level DSL explicitly encodes cross-thread interactions over shared resources, enabling focused reasoning without traversing deep call chains. Experiments on real-world concurrent bugs from eight large-scale Java projects show that ConFL significantly outperforms state-of-the-art IR-based and LLM-based baselines, achieving an MRR of 0.503 and a MAP of 0.486, while remaining robust to noisy bug reports, unseen bugs, and different LLM backbones.

Software fault localization identifies the code locations responsible for reported issues and is a fundamental step toward automated debugging and program repair. Recent retrieval-based approaches formulate fault localization as a dense retrieval task by learning a shared embedding space between issue reports and source code. However, these methods encode all issue reports using a fixed query representation, despite the substantial diversity of real-world issue reports in length, structure, and debugging information. To address this limitation, we propose HyperFL, a query-adaptive representation learning framework for software fault localization. HyperFL employs a lightweight hypernetwork to generate query-specific LoRA parameters for the query encoder, enabling dynamic query adaptation while keeping the code encoder fixed and reusable. Experiments on a real-world issue localization benchmark demonstrate that HyperFL consistently improves retrieval performance across multiple embedding backbones, achieving up to 13.3% relative improvement in function-level MRR@10 and 16.7% relative improvement in Hit@1 over the state-of-the-art method SweRank. Further analysis shows that HyperFL learns distinct adaptation patterns for different issue characteristics, highlighting the effectiveness of query-adaptive representations for software issue localization.

Domain-specific Instruction Set Architecture eXtensions (ISAX) are widely adopted in the RISC-V ecosystem to accelerate emerging workloads, but implementing and validating ISAXes across different cores remains slow and fragmented. Existing frameworks still require per-core interface adaptation, and differential testing often breaks once either the microarchitecture or the ISAX changes. We present LACE, an LLM-aided multi-agent workflow that translates natural-language ISAX intents into a compact two-level IR (operation-level and HDL task-level), performs retrieval-guided localized RTL edits over large repositories, and closes the loop with a compiler-agnostic riscv-formal checking flow (assuming RVFI availability or instrumentation). Across four embedded RISC-V cores, LACE raises pass@1 generation accuracy from near-zero to 72.8% within our evaluation setup, while improving code localization and reducing integration rework. The code of LACE is available at https://github.com/UMN-ZhaoLab/LACE.

Kernel generation for hardware accelerators such as GPUs and NPUs has become a proving ground for large language models (LLMs), and state-of-the-art systems raise correctness through pipelines that couple LLMs with agentic reinforcement learning and evolutionary search. Such pipelines generate, compile, and execute large numbers of candidate kernels, discarding most of them and forgoing the opportunity to distill failures into reusable knowledge. Many discarded candidates are near-miss operators that compile and run but fail numerical validation; each embodies genuine domain knowledge and a nontrivial investment in LLM inference, cross-compilation, and hardware execution. We argue for a paradigm shift: rather than regenerate, debug. Debugging is far more constrained than generating from scratch: the search space is small and feedback is dense. We present a domain-specific debug agent that addresses three core challenges in autonomous repair: mitigating knowledge scarcity through retrieved patterns and diagnostic instrumentation, ensuring integrity through anti-cheat detection and full-coverage evaluation, and controlling cost via convergence guards and bounded iteration. Debugging serves two complementary roles: it extends the capability frontier by recovering operators that repeated regeneration fails to produce, and it lowers cost per deliverable operator. Debug Pass@1 achieves 66.7% versus Regenerate Avg Pass@1's 25.9% and Regenerate Pass@3's 40.7%, while consuming 92.8% fewer tokens per success than three-trial regeneration. Component ablations show that the knowledge base drives recovery, while integrity gates reject 12.5-33.3% of the successes the workflow itself accepted.

Software supply-chain attacks increasingly exploit an identity gap where compromised maintainer accounts authorize malicious changes. This work evaluates patch-level authorship verification as a behavioral defense layer, showing that stylometric analysis can operate not only on full source files but also on patch-level commits. We fine-tune a cross-modal transformer on more than 20 years of Linux kernel commit history to embed code diffs and commit messages into a unified stylometric space, achieving ROC AUC of 0.93 for open-world authorship verification. We then use these representations in a streaming anomaly detector suited to continuous integration and deployment (CI/CD) settings. We validate the pipeline on two retrospective supply-chain incidents involving different patch characteristics: the 2021 PHP backdoor and the 2026 ForceMemo/GlassWorm campaign. Without retraining, the proposed detector surfaces both PHP forged commits within approximately 1% of the maintainer audit queue and ranks the 28 scoreable ForceMemo spoofs with a median per-repository review burden of 0.8%. These results indicate that cross-modal patch-level embeddings can support behavioral triage against author impersonation in real-world repositories.

LLM-based code auditors are increasingly integrated into pull-request (PR) workflows, yet their reliability against adversarial changes distributed across repository evolution remains poorly understood. We introduce PRWeaver, a benchmark of 208 execution-validated attacks from ten real-world repositories, each instantiated under four matched review renderings (832 renderings in total). We evaluate three PR-auditing agents across six auditor-model systems. Across all systems, decomposing an attack changes detection by at most five percentage points, showing that commit boundaries alone do not explain evasion. In contrast, per-PR interleaving at $N=16$ and coherent carrier fusion reduce detection by 5-13 and 10-18 points, respectively. Under whole-window review at $N=24$, detection falls to 16-22%, compared with 50-60% under per-PR review. These results show that access to repository history is insufficient: concealment becomes most effective when benign and malicious changes jointly occupy the auditor's active review context or when the stated purpose plausibly accounts for the attack-bearing diff.

AI coding agents accept free-form natural language prompts that may inadvertently contain credentials, personally identifiable information (PII), or proprietary business data. Existing data loss prevention (DLP) solutions rely on rigid regex patterns, model fine-tuning, or vendor-managed classifiers with limited customizability. We present PolicyGuard, a pre-model interception framework that classifies user prompts using an LLM guided by a natural language policy file. Our key contributions are: (1) the policy-as-prompt paradigm, where DLP classification criteria are defined entirely in a plaintext policy document editable by non-engineers without code changes or model retraining; (2) a sealed evaluation protocol with template-family-level data splits, hidden holdouts, and frozen test sets to rigorously assess generalization; and (3) a comprehensive empirical evaluation across 2,000 multilingual prompts demonstrating 96.5% effective block rate (EBR) with only 3.0% false positive rate (FPR) on a frozen test set of 927 prompts, and perfect 100% accuracy on a 217-prompt hidden holdout. Information-matched baseline experiments show that PolicyGuard's natural language format significantly outperforms equivalent content in JSON format (McNemar chi-squared = 31.58, p < 0.001) and dramatically outperforms zero-shot classification (Cohen's h = 0.915). Cross-model portability experiments demonstrate that the same policy achieves 86.4-96.5% EBR across four different LLMs without modification.

Coding-agent benchmarks increasingly cover long-horizon, end-to-end, and interactive development, but typically retain one requested outcome or a fixed change sequence. Sequential policies can process a pull-request (PR) queue one candidate at a time, but when queued PRs interact, maximizing safe delivery can require jointly deciding which changes to merge and in what order. We introduce BulkPR-Bench, an executable benchmark in which an agent must recover consequential PR relations and return a large safe subset in executable order under a rolling-release protocol. The suite contains 581 newly authored candidate PRs on frozen snapshots of 18 real repositories. Registered state-by-state repository execution, including hidden safety checks, validates the gold relation graph; an exact oracle then computes the largest safe subset. Our primary metric, Relational Delivery Score (RDS), scores safe delivery and correct rejection over relation groups from the realized merge trace; Global Safety-Gated Yield (Global-SGY) separately measures strict delivery of the realized whole-queue plan. Under the buffered primary protocol with batch size $K=32$, the three highest RDS estimates among the six models are 66.6%, 62.0%, and 57.9%, compared with 53.1% for the strongest sequential baseline. Only 8 of 324 model runs complete a queue exactly. Critical-relation recall ranges from 35.2% to 57.7%, and diagnostic runs supplied with the gold relations show substantial remaining headroom. Gains on relation groups therefore do not yet translate into dependable whole-queue governance.

Continuous Integration (CI) is central to modern software development, yet developers often struggle to choose the most suitable CI service. Prior work has identified barriers to CI adoption but offers little empirical evidence on how developers select CI services or whether adoption decisions are driven by genuine project needs versus social influence. This paper presents an exploratory survey study addressing that gap. We aim to contact about 5,000 active GitHub developers, including both CI users and non-users. The study investigates: (1) what drives CI adoption and service selection, distinguishing need-driven from socially influenced motivations; (2) whether developers consider CI universally necessary or context-dependent and what barriers hinder adoption; and (3) developers' perceptions of automated CI recommendation systems. Our findings will inform researchers developing CI recommendation systems and practitioners aiming to streamline CI adoption in open-source projects.

Tool-using language-model agents repeatedly rediscover procedures they have already executed, producing traces that mix reusable structure with retries, exploration, accidental ordering, and repeated lookups. We present TraceCompiler, a skill-guided system that mines clusters of noisy agent traces and compiles them into executable, mostly deterministic workflows. It admits an inter-tool dependency only when a consumer argument contains a value attributable uniquely to an earlier producer; every hard edge carries an auditable evidence tuple, and ambiguous relations are marked suspected and impose no ordering constraint. Bindings are classified as constants, user inputs, copied outputs, transforms, or residual LLM decisions. On T1, a mechanized form of the rule recovers producer-consumer dependencies at 0.928 precision and 0.943 recall over 15,775 def-use edges of its training split, against 0.711 F1 for adjacency and 0.712 for a frequency-thresholded directly-follows measure on identical data; the compiler skill run blind reaches 0.992 on 250 of those edges. On AppWorld we replay released trajectories in the deterministic simulator to recover masked return values and measure the rule against 563 token edges at 0.993 precision - a self-consistency check, since replay injects tokens by a related heuristic. We compile two recurring intents: a Venmo money-request intent reduces 34 observed API calls to 11 runtime calls and, under leave-one-out execution against the benchmark's own state tests, passes 15 of 21, the failing fold escalating rather than acting because its required branch was never observed; and a Spotify/Todoist intent the compiler correctly refuses to compile, because an irreversible side effect is under-determined. We measure call reduction but not offline compilation cost, so we claim no net efficiency result.

Traditional software cost estimation models, such as COCOMO II, Function Points, and Story Points, assume that development effort is primarily driven by human labor in design, coding, and testing. Agentic software engineering, where autonomous AI agents perform substantial implementation work and humans focus on planning, specification, and validation, challenges this assumption. New cost dimensions arise: large language model (LLM) token consumption across agent actions, Human-in-the-Loop (HITL) oversight effort, and infrastructure costs for agent orchestration and tooling. These costs are nondeterministic: identical tasks may consume different tokens, follow divergent reasoning paths, and require varying human correction, phenomena absent in traditional development. A new framework is needed to bridge standard sizing metrics with this cost structure. This paper proposes ACEM (Agentic Cost Estimation Model), which decomposes total agentic development cost into three additive dimensions: LLM, HITL, and infrastructure cost. ACEM introduces three constructs for agentic dynamics: the Revision Factor (RF), modeling token overhead from output rejection and retries; the Context Factor (CF), capturing rising token consumption as context accumulates; and the HITL Intensity Score (HIS), a four-level oversight classification scheme. It further maps Use Case Points, Story Points, and Function Points to estimated token consumption, enabling organizations to reuse existing project-scoping data for agentic cost forecasting. ACEM is presented as a fully specified model structure and calibration methodology, with constants left symbolic pending empirical grounding. As an early-stage proposal, it invites the research community to calibrate, test, and extend the model through real project data.