AX·NEWS
Sign in

Software Engineering

Software Engineering news — Tuesday 27 October 2026

150 stories from 6 sources, every one in English. Each links to its publisher.

Open in the live feedRSS feedOn a TV

  1. Coders’ Rights Project Reverse Engineering FAQ (2017)
    USlobste.rs

    Comments

  2. Early progress on accelerating drug development with Claude
    USHacker News Front Page

    Article URL: https://www.anthropic.com/research/early-progress-drug-development Comments URL: https://news.ycombinator.com/item?id=50265609 Points: 310 # Comments: 306

  3. Raspberry Pi Zero 3W and Pi 4 Model A spotted in device trees
    USHacker News Front Page

    Article URL: https://bret.dk/raspberry-pi-zero-3-w-pi-4-model-a/ Comments URL: https://news.ycombinator.com/item?id=50265594 Points: 196 # Comments: 124

  4. Fixing KIO copy to USB stick and network filesystems
    USlobste.rs

    Comments

  5. The next circle of security hell
    USlobste.rs

    Comments

  6. Tailscale SSH: Investigation and remediation update
    USHacker News Front Page

    Article URL: https://tailscale.com/ssh-investigation-update Comments URL: https://news.ycombinator.com/item?id=50264857 Points: 361 # Comments: 186

  7. Survival LLM: AI assistant and emergency library that runs on your own computer
    USHacker News Front Page

    Article URL: https://www.brangus.io/survival-llm Comments URL: https://news.ycombinator.com/item?id=50264853 Points: 495 # Comments: 188

  8. Fuzzers vs. AI: Comparing Fuzzilli and BigSleep on JavaScript Engines
    USlobste.rs

    Comments

  9. An Evening with Berferd: In Which a Cracker is Lured, Endured, and Studied (1992)
    USlobste.rs

    Comments

  10. Four Months of Zig
    USlobste.rs

    Comments

  11. Discord will soon let anyone find public server messages through Google
    USHacker News Front Page

    Article URL: https://discord.com/blog/making-content-from-discoverable-servers-easier-to-discover Comments URL: https://news.ycombinator.com/item?id=50263854 Points: 251 # Comments: 135

  12. Breaking the open web won't stop the bad bots
    USHacker News Front Page

    Article URL: https://www.techpolicy.press/breaking-the-open-web-wont-stop-the-bad-bots/ Comments URL: https://news.ycombinator.com/item?id=50263837 Points: 310 # Comments: 134

  13. 1,447 emoji? We'll do it live
    USlobste.rs

    Comments

  14. Gilgamesh by Simon Armitage review – an ancient epic is brilliantly translated
    USHacker News Front Page

    Article URL: https://www.theguardian.com/books/2026/oct/20/gilgamesh-by-simon-armitage-review-an-ancient-epic-is-brilliantly-translated-for-our-times Comments URL: https://news.ycombinator.com/item?id=50263127 Points: 167 # Comments: 51

  15. The Shift of Theseus — Toy Bit Provenance Abstract Domain
    USlobste.rs

    Comments

  16. Introducing LFM3: Every Modality for Specialized Intelligence
    USHacker News Front Page

    Article URL: https://www.liquid.ai/blog/lfm3 Comments URL: https://news.ycombinator.com/item?id=50262229 Points: 128 # Comments: 51

  17. A fast and secure tar parser for the Rust ecosystem
    USlobste.rs

    Comments

  18. Ghostty 1.4.0
    USHacker News Front Page

    Article URL: https://ghostty.org/docs/install/release-notes/1-4-0 Comments URL: https://news.ycombinator.com/item?id=50261998 Points: 1294 # Comments: 412

  19. Finding panics with Call Graph Analyzer
    USlobste.rs

    Comments

  20. Filian, Fil-C binaries on Debian
    USlobste.rs

    Comments

  21. A 28-Year-Old DOGE Operative Gutted USAID. HR Had No Record of Him Working There
    USHacker News Front Page

    Article URL: https://www.wired.com/story/a-28-year-old-doge-operative-gutted-usaid-hr-had-no-record-of-him-working-there/ Comments URL: https://news.ycombinator.com/item?id=50261784 Points: 109 # Comments: 35

  22. Classification of Finite Simple Groups Formalized in Lean
    USHacker News Front Page

    Article URL: https://cfsg.axiommath.ai/ Comments URL: https://news.ycombinator.com/item?id=50261647 Points: 221 # Comments: 68

  23. Amazon Is Gutting Its AI Division After Sustained Failure
    USHacker News Front Page

    Article URL: https://futurism.com/artificial-intelligence/amazon-ai-division-failure-nova Comments URL: https://news.ycombinator.com/item?id=50261589 Points: 269 # Comments: 185

  24. Apple introduces the new 14‑inch MacBook Pro with M6
    USHacker News Front Page

    Article URL: https://www.apple.com/in/newsroom/2026/10/apple-introduces-the-new-14-inch-macbook-pro-with-m6/ Comments URL: https://news.ycombinator.com/item?id=50261352 Points: 404 # Comments: 557

  25. Anthropic trained Claude on illegally obtained Stanford professors' books
    USHacker News Front Page

    Article URL: https://stanforddaily.com/2026/10/27/professors-books-pirated-to-train-claude/ Comments URL: https://news.ycombinator.com/item?id=50260814 Points: 371 # Comments: 208

  26. A $300B 'Debt Wall' Is Looming over Private-Equity Backed Companies
    USHacker News Front Page

    Article URL: https://www.wsj.com/finance/investing/a-300-billion-debt-wall-is-looming-over-private-equity-backed-companies-d6359497 Comments URL: https://news.ycombinator.com/item?id=50260526 Points: 101 # Comments: 30

  27. What’s !important #19: CSS Refraction Effects, Scroll State Queries, @custom-media, and More
    USCSS-Tricks

    We’re back! Don't worry, this isn’t a trick (despite what our name suggests!). However, due to the hiatus, we're also a little behind. That’s why this issue of What’s !important is absolutely stacked. All killer, no filler. Here's what you…

  28. A stackless floodfill automation (2023)
    USlobste.rs

    Comments

  29. Compile-Time Dependent Types
    USlobste.rs

    Comments

  30. The US Government Is Hiding Secret Court Ruling About How It Spies on Americans
    USHacker News Front Page

    Article URL: https://www.wired.com/story/the-us-government-is-hiding-a-secret-court-ruling-about-how-it-spies-on-americans/ Comments URL: https://news.ycombinator.com/item?id=50259604 Points: 998 # Comments: 381

  31. Is my data used for model training?
    USHacker News Front Page

    Article URL: https://privacy.claude.com/en/articles/7996868-is-my-data-used-for-model-training Comments URL: https://news.ycombinator.com/item?id=50259406 Points: 185 # Comments: 81

  32. PHP2Shell: Exploitation attempts and remediation trends
    USHacker News Front Page

    Article URL: https://www.wiz.io/blog/php2shell-exploitation-so-far Comments URL: https://news.ycombinator.com/item?id=50259343 Points: 165 # Comments: 55

  33. Here be monsters: What do you still need to know when agents do the work?
    USHacker News Front Page

    Article URL: https://blog.adrq.dev/here-be-monsters-what-do-you-still-need-to-know-when-agents-do-the-work Comments URL: https://news.ycombinator.com/item?id=50258868 Points: 132 # Comments: 69

  34. slughorn — GPU Vector Graphics
    USlobste.rs

    Comments

  35. New iMac with M6
    USHacker News Front Page

    Article URL: https://www.apple.com/imac/ Comments URL: https://news.ycombinator.com/item?id=50258538 Points: 363 # Comments: 464

  36. New 14‑inch MacBook Pro with M6
    USHacker News Front Page

    Article URL: https://www.apple.com/newsroom/2026/10/apple-introduces-the-new-14-inch-macbook-pro-with-m6/ Comments URL: https://news.ycombinator.com/item?id=50258521 Points: 140 # Comments: 163

  37. EnsoJS - native-first microframework for building declarative Web Components
    USlobste.rs

    Comments

  38. A select should tell you something
    USlobste.rs

    Comments

  39. Software may be getting worse, but it used to be horrible
    USHacker News Front Page

    Article URL: https://www.bitlog.com/2026/10/27/software-may-be-getting-worse-but-it-used-to-be-horrible/ Comments URL: https://news.ycombinator.com/item?id=50257548 Points: 254 # Comments: 232

  40. AI promised less work. I've never worked more
    USHacker News Front Page

    Article URL: https://oioannou.com/blog/ai-promised-less-work/ Comments URL: https://news.ycombinator.com/item?id=50257325 Points: 393 # Comments: 347

  41. Roblox downloaded 145GB models from us (including adult content) without asking
    USHacker News Front Page

    Article URL: https://open3dlab.com/news/roblox-downloaded-145-gb-of-models-from-our-sites/ Comments URL: https://news.ycombinator.com/item?id=50256985 Points: 407 # Comments: 234

  42. Full Haskell-like Type Class resolution in Java
    USlobste.rs

    Comments

  43. TypeScript and Haskell: Unions ＆ Singletons (2024)
    USlobste.rs

    Comments

  44. Code.org's Excellent Beginner Programming Courses Gets Lost in the AI Hype
    USlobste.rs

    I’ve often recommended Code.org, with its memorable name. Their "Introduction to Computer Science" courses are among the best beginner programming courses I know. But since the platform was rebranded as CodeAI, they've been getting lost in…

  45. I will no longer upload to NEW
    USHacker News Front Page

    Article URL: https://lists.debian.org/debian-devel/2026/10/msg00263.html Comments URL: https://news.ycombinator.com/item?id=50256738 Points: 148 # Comments: 65

  46. Keyless Entry: Hacking SwitchBot Smart Locks
    USlobste.rs

    Comments

  47. Elevators, toasters, and the birth of Java
    USStack Overflow Blog

    Ryan sits down with James Gosling, creator of Java at Sun Microsystems, to talk about the early networking friction that drove the need for interconnectedness, the stringent reliability requirements of consumer electronics companies that…

  48. Cuis: The Modern Smalltalk-80
    USHacker News Front Page

    Article URL: https://cuis.st/ Comments URL: https://news.ycombinator.com/item?id=50255343 Points: 159 # Comments: 19

  49. The Mythical Man-Month in the Age of AI
    USHacker News Front Page

    Article URL: https://j11g.com/the-mythical-man-month-in-the-age-of-ai Comments URL: https://news.ycombinator.com/item?id=50255298 Points: 380 # Comments: 263

  50. Python's experimental JIT to be removed
    USlobste.rs

    Comments

  51. Navigating Epistemic Parity in LLM Agents: A Benchmark for Cross-Source Conflict Resolution Between Memory and Procedural Skills
    USarXiv cs.SE (Software)

    arXiv:2610.33128v1 Announce Type: new Abstract: Production LLM agents increasingly run episodic memory pipelines and procedural skill modules (e.g., SKILL.md files) in tandem. Both are injected into the same context region with no defined…

  52. EVM-Anchored Provenance for AI-Assisted Code Evolution: A Smoke-Scale Deployment Report
    USarXiv cs.SE (Software)

    arXiv:2610.33350v1 Announce Type: new Abstract: We report a smoke-scale deployment of an EVM-anchored provenance framework for the state stream produced by AI-assisted code generation. The system comprises codebase-state-manager, an…

  53. How Do Small Coding Agents Use Code Graphs and Embeddings?
    USarXiv cs.SE (Software)

    arXiv:2610.33958v1 Announce Type: new Abstract: Coding agents are given embedding search and code-graph tools on the premise that a repository's structure is something the model lacks and the tool supplies. We test that premise on…

  54. Bug Severity Classification with Fine-Tuned Open LLMs: A Paired Comparison with TF-IDF and CodeBERT Baselines on Eclipse Reports
    USarXiv cs.SE (Software)

    arXiv:2610.34127v1 Announce Type: new Abstract: Open large language models (LLMs) small enough to run on a laptop are an obvious candidate for automatic bug severity prediction. We measure what they add over cheaper classifiers on one…

  55. Honesty Testing of LLMs and AI Agents: A Vision for Software Engineering
    USarXiv cs.SE (Software)

    arXiv:2610.34239v1 Announce Type: new Abstract: Human oversight of large language models (LLMs) and artificial intelligence (AI) agents increasingly depends on their reports. Claims that conflict with a system's own judgments can conceal…

  56. Beyond Benchmarks: Can Small Open-Weight Models Get Real Coding Work Done?
    USarXiv cs.SE (Software)

    arXiv:2610.34241v1 Announce Type: new Abstract: Making an open-weight language model generate text is not the same as making it finish coding work. This retrospective empirical systems experience report follows my engineering work from…

  57. CodeMeepo: A Multi-Agent Workflow for Repository-Level Code Migration and Optimization
    USarXiv cs.SE (Software)

    arXiv:2610.34633v1 Announce Type: new Abstract: Repository-level code migration rewrites an entire project in a new programming language while preserving its observable behavior. It is challenging because behavior spans interdependent…

  58. Thinking-in-Bets at Test Time Reduces Sycophantic Revisions in LLM Agents
    USarXiv cs.SE (Software)

    arXiv:2610.34654v1 Announce Type: new Abstract: Language model agents often abandon correct work when a user challenges it, yet simply telling them to stand firm makes them ignore real mistakes. Once an agent reads a complaint, it tends to…

  59. SemWeaver: Refining LLM-Generated Static Vulnerability Checkers with Analyzer-Internal Evidence
    USarXiv cs.SE (Software)

    arXiv:2610.34928v1 Announce Type: new Abstract: Recent work shows that LLMs can synthesize static-analysis checkers from security patches, but a compilable checker is not necessarily a correct one. A patch reveals what changed, not why the…

  60. Reasoning Patterns for Efficient AI Coding: An Empirical Study
    USarXiv cs.SE (Software)

    arXiv:2610.34943v1 Announce Type: new Abstract: Large Reasoning Models (LRMs) generate long reasoning traces that expose intermediate inference behavior, offering opportunities to study and control reasoning in AI coding. However, the…

  61. Making Small Code LLMs More Efficient for Secure Code Review with Neuro-Symbolic Fine-Tuning
    USarXiv cs.SE (Software)

    arXiv:2610.35477v1 Announce Type: new Abstract: As AI coding assistants generate more code, the need to review it for security flaws grows. Existing approaches leave a gap: static analyzers identify program sites but offer little…

  62. Web Agents Should Adopt the Compile-and-Execute Paradigm
    USarXiv cs.SE (Software)

    arXiv:2605.14290v2 Announce Type: replace-cross Abstract: ReAct has become the default architecture across LLM agents, and many existing web agents follow this paradigm. We argue that it is the wrong default for web agents. Instead, web…

  63. Why Does Structured Output Fail? An Empirical Study of Prompt-Enforced Pydantic Schemas Across Nine LLMs
    USarXiv cs.SE (Software)

    arXiv:2610.33949v1 Announce Type: new Abstract: Pipelines that require structured outputs from hosted language models often specify a Pydantic schema in the prompt, validate the response, and retry when validation fails. Although native…

  64. Software Engineering for AI-Driven Science
    USarXiv cs.SE (Software)

    arXiv:2610.34573v1 Announce Type: new Abstract: AI has made significant contributions to scientific discovery. Autonomous agents can now independently analyse complex problems, design experiments, write and execute code, analyse findings…

  65. Numina Fuse: Interactive Autoformalization in Lean 4
    USarXiv cs.SE (Software)

    arXiv:2610.33860v1 Announce Type: cross Abstract: Numina Fuse is an agent-native application for formalizing mathematics in Lean 4. Since Lean can compile while not representing its intended informal meaning, and agents now write Lean…

  66. Repairing Broken Access Control in Web Applications with LLMs: An Empirical Benchmark Study
    USarXiv cs.SE (Software)

    arXiv:2610.34290v1 Announce Type: cross Abstract: Broken access control (BAC) ranks among the most severe web-application vulnerability classes, yet its automated repair remains largely unexplored. Unlike memory-safety vulnerabilities, BAC…

  67. Tiny Machine-Learning Operations within Cyber-Physical Systems: a Field Study
    USarXiv cs.SE (Software)

    arXiv:2606.20952v2 Announce Type: replace Abstract: Machine-Learning Operations (MLOps) is maturing into a software-engineering discipline, yet its tiny-scale variant (TinyMLOps)-targeting the resource-constrained microcontrollers embedded…

  68. Toward Methods and Tools for Critical Examination of Disparate Systems for Discovery of Hidden Dependencies in Engineering of Autonomous and Semi-Autonomous Systems
    USarXiv cs.SE (Software)

    arXiv:2610.34974v1 Announce Type: new Abstract: We present a methodology, an initial AI Skill package, and emerging results, for enhancing software engineers' acquisition and application of domain knowledge and system-wide engineering…

  69. Then there is prompt injection, a malicious email, ...: The Wild Wide West of OpenClaw Agentic System Security
    USarXiv cs.SE (Software)

    arXiv:2610.35451v1 Announce Type: cross Abstract: AI-powered personal assistant frameworks such as OpenClaw combine LLM reasoning, persistent memory, tool invocation, local execution, and web-based workflows. These capabilities enable…

  70. Connecting Quantum Proof Systems and Hoare Logic
    USarXiv cs.PL (Programming)

    arXiv:2610.34723v1 Announce Type: cross Abstract: Verifying quantum programs today requires specialized proof systems built on interactive theorem provers such as Rocq, Lean, Isabelle, or Why3, and even modest quantum algorithms cost…

  71. Operationalizing Governed Agentic Workflows in Regulated Finance: A Customer Zero Case Study using Zafin AIOS
    USarXiv cs.SE (Software)

    arXiv:2610.33253v1 Announce Type: new Abstract: Regulated industries such as finance and healthcare face critical governance challenges that hamper the adoption of AI-assisted software development and hinder the transition from…

  72. Toward Evidence-Based Agentic Software Engineering
    USarXiv cs.SE (Software)

    arXiv:2610.35348v1 Announce Type: new Abstract: Task success rates and execution costs of agentic coding tools depend on both model and harness. The harness manages context, provides tools, and runs the agent loop. In empirical software…

  73. FLOWSTEP: A Framework for Learning to Predict GitHub Actions Workflow STEPs
    USarXiv cs.SE (Software)

    arXiv:2610.35650v1 Announce Type: new Abstract: Continuous Integration and Delivery (CI/CD) are fundamental to modern software development, yet configuring pipelines such as GitHub Actions (GHA) workflows remains complex and error-prone…

  74. Decision Model Factory: Building Specialized Sovereign Decision Models
    USarXiv cs.SE (Software)

    arXiv:2610.34956v1 Announce Type: cross Abstract: Decision needs vary with each application, language, and deployment environment. The Decision Model Factory turns those needs into specialized, sovereign models on demand. A brief recipe…

  75. FlashScript: Compiling Recurring Agent Workflows into Reusable Neuro-Symbolic Programs
    USarXiv cs.SE (Software)

    arXiv:2610.33876v1 Announce Type: cross Abstract: Many agent tasks recur with new data or small changes, yet agents regenerate their reasoning and actions on each run, producing different results. Distilling a prior run into a skill helps…

  76. Change2Task: From Repository Changes to Executable Coding Agent Tasks and Environments
    USarXiv cs.SE (Software)

    arXiv:2607.28591v2 Announce Type: replace Abstract: Scaling coding agents requires a continuing supply of executable data for training, benchmarking, and continuous evaluation. Each task must couple a realistic software state with a…

  77. MOFjson: An Explicit Topological Representation of Metal-Organic Frameworks for Language Models
    USarXiv cs.SE (Software)

    arXiv:2610.34549v1 Announce Type: cross Abstract: Language models for metal-organic frameworks (MOFs) rely on formats that separate atomic geometry from structural topology, forcing models to choose between unstructured coordinate lists…

  78. Agentic AI Orchestration Framework for Spatial Intelligence: Supply Chain Deployment
    USarXiv cs.SE (Software)

    arXiv:2610.34569v1 Announce Type: cross Abstract: Industrial spatial decisions are assembled across systems that share neither a reference frame nor an owner, and analysts spend substantial effort reconciling those systems rather than…

  79. Covered Is Not Checked: Which Branch Outcomes Would Your Tests Notice?
    USarXiv cs.SE (Software)

    arXiv:2610.33708v1 Announce Type: new Abstract: Code coverage reports what tests execute, not whether tests detect faults. Checked coverage addresses this gap by tracing code that influences test oracles through dynamic slicing, but…

  80. Firefly: Illuminating Large-Scale Verified Tool-Call Data Generation from Real APIs
    USarXiv cs.SE (Software)

    arXiv:2605.17558v2 Announce Type: replace Abstract: Training tool-calling agents requires large-scale trajectory data with verifiable labels, yet existing approaches either synthesize environments that diverge from real API behavior or…

  81. Rivet: Utility-Preserving Context Control for Repository-Level Coding Agents
    USarXiv cs.SE (Software)

    arXiv:2610.34967v1 Announce Type: new Abstract: Repository-level code generation asks an agent to implement a task requirement inside an existing repository. To produce code that works in the repository, the agent needs repository…

  82. CyberDual: AI Attackers versus AI Code Reviewers in Real Software Maintenance
    USarXiv cs.SE (Software)

    arXiv:2610.34435v1 Announce Type: cross Abstract: Open-source projects now let AI agents review incoming code, but the code under review may itself come from an AI attacker. We present CyberDual, an arena for this dual role of AI: attacker…

  83. Terminus-4B: Can a Smaller Model Replace Frontier LLMs at Agentic Execution Tasks?
    USarXiv cs.SE (Software)

    arXiv:2605.03195v3 Announce Type: replace-cross Abstract: Modern coding agents increasingly delegate specialized subtasks to subagents, which are smaller, focused agentic loops that handle narrow responsibilities like search, debugging or…

  84. Can Agentic Code Review Catch Vulnerabilities Before They Are Merged?
    USarXiv cs.PL (Programming)

    arXiv:2610.33509v1 Announce Type: cross Abstract: LLM agents have been increasingly adopted for security-focused code review, but existing benchmarks mostly derive the reviewed changes from their later fixes instead of the real-world…

  85. PastaLean: Deterministic transpilation from Python to Lean for Verification
    USarXiv cs.PL (Programming)

    arXiv:2610.34205v1 Announce Type: new Abstract: We present PastaLean, an extensible system for deterministic translation from Python to Lean via the Abstract Syntax Tree of Python. This allows for verification of Python code, especially…

  86. MicroAgent: Context-Augmented Multi-Agent Framework for Automatic Microservice Decomposition
    USarXiv cs.SE (Software)

    arXiv:2606.29742v2 Announce Type: replace Abstract: The adoption of Microservice Architecture (MSA) has revolutionized software engineering by enhancing scalability, agility, and maintainability over traditional monolithic applications. As…

  87. SymLift: Neural Semantic Lifting for Compile-Free Vulnerability Detection in Context-Incomplete Code Fragments
    USarXiv cs.SE (Software)

    arXiv:2610.34431v1 Announce Type: new Abstract: Missing project context can prevent build-dependent analyzers from constructing program relations required for vulnerability detection, while neural predictions alone do not establish…

  88. RepoTrace: Improving Requirement Tracing in Multi-Round Agent Interactions
    USarXiv cs.SE (Software)

    arXiv:2610.35333v1 Announce Type: new Abstract: Code agents have become increasingly capable of performing software engineering tasks through interaction with software environments. However, agents tend to degrade as the number of…

  89. Compaction as Epistemic Failure: How Agentic LLM Tools Fabricate Confirmed Results from Killed Processes
    USarXiv cs.SE (Software)

    arXiv:2607.13071v2 Announce Type: replace Abstract: Agentic LLM coding tools compress long session histories into compaction summaries that subsequent sessions inherit as ground truth. This paper documents a failure mode in Claude Code…

  90. BoardSafety: AI-Ready Data from Manufacturer Documentation for Development-Board Safety
    USarXiv cs.SE (Software)

    arXiv:2610.33448v1 Announce Type: cross Abstract: The development of large language models (LLMs) are capable to assist with hardware experiments. In this scenario, this raises a practical question: Can LLMs determine whether a proposed…

  91. FeatureLoop: Benchmarking Coding Agents under Evolving Requirements
    USarXiv cs.SE (Software)

    arXiv:2610.35488v1 Announce Type: new Abstract: Software development with coding agents often begins with a rough request that users refine after trying the implementation. Final correctness alone does not capture the repeated guidance…

  92. Developer Task Delegation to LLMs and the Trust Factor: A Mixed Methods Study
    USarXiv cs.SE (Software)

    arXiv:2610.34317v1 Announce Type: new Abstract: With the rise of Large Language Models (LLMs), software developers are presented with a powerful new approach: delegating software development tasks to LLMs. Existing studies have explored…

  93. The Physical Cost of Context: Quality-Constrained Resource Accounting for Coding Agents
    USarXiv cs.SE (Software)

    arXiv:2610.34763v1 Announce Type: new Abstract: Coding agents resend a growing context on every model call, and context policies are usually judged by the prompt tokens they save. Executed work also depends on call count and prefix reuse…

  94. Software Engineering LLMs: Why We Need Them and How Software Engineering Helps Build Them
    USarXiv cs.SE (Software)

    arXiv:2610.33851v1 Announce Type: new Abstract: Software engineering (SE) expertise can shape how coding large language models (LLMs) are trained and which engineering capabilities they acquire. We examine this role through an industrial…

  95. Chart the Invisible: Interactive Perception for Hidden-Target Exploration in GUI Agents
    USarXiv cs.SE (Software)

    arXiv:2610.35237v1 Announce Type: new Abstract: Despite rapid progress in command-line interface (CLI) tasks, recent LLM-based agents still struggle with reliable task execution through graphical user interfaces (GUIs). A key challenge is…

  96. SWE-Ladder: Learning to Localize, Reproduce, and Patch via Progressive Privileged Self-Distillation
    USarXiv cs.SE (Software)

    arXiv:2610.33667v1 Announce Type: cross Abstract: Resolving a repository-level issue typically requires a software-engineering (SWE) agent to localize the faulty code, reproduce the failure, and patch the code. Reinforcement learning…

  97. PBFuzz: Agentic Directed Fuzzing for PoV Generation
    USarXiv cs.SE (Software)

    arXiv:2512.04611v3 Announce Type: replace-cross Abstract: Proof-of-Vulnerability (PoV) input generation is a critical task in software security and supports downstream applications such as bug reproduction and patch validation. Generating…

  98. GPU-Accelerated Belief Propagation for Program Analysis
    USarXiv cs.SE (Software)

    arXiv:2509.22337v3 Announce Type: replace Abstract: Belief Propagation (BP) is a widely used approximate inference algorithm in probabilistic graphical models (PGMs), but is computationally expensive when applied to large-scale program…

  99. Human-AI Synergy in Agentic Code Review
    USarXiv cs.SE (Software)

    arXiv:2603.15911v2 Announce Type: replace Abstract: Code review is a critical software engineering practice where developers review code changes before integration to ensure code quality, detect defects, and improve maintainability. In…

  100. ComBeautify: Leveraging Code Change Summaries for Conventional Commit Generation
    USarXiv cs.SE (Software)

    arXiv:2610.34374v1 Announce Type: new Abstract: Modern software development relies on version control systems such as Git to track changes and facilitate collaboration. While commit messages are crucial for conveying developer intent, they…

  101. A Causal Perspective on the Role of Training Dynamics for Interpreting Privacy Risks in Code Models
    USarXiv cs.SE (Software)

    arXiv:2512.07814v3 Announce Type: replace Abstract: Large language models for code (LLM4Code) have greatly improved developer productivity but also raise privacy concerns due to their reliance on open-source repositories containing…

  102. Testing with Coding Agents: A Large-Scale Comparison of Agent-Supported and Human-Authored Pull Requests
    USarXiv cs.SE (Software)

    arXiv:2610.35297v1 Announce Type: new Abstract: Context. Coding agents increasingly contribute tests through pull requests that developers then maintain, so how testing work is divided affects long-term maintainability. Objective. We…

  103. The Computational Assessment Stack: An Architecture for AI-Native Measurement
    USarXiv cs.SE (Software)

    arXiv:2610.33901v1 Announce Type: cross Abstract: Test development is shifting toward automation and agentic artificial intelligence (AI), in which autonomous systems draft items at scale, assemble test forms, administer adaptive…

  104. Cross-Platform Application Migration to HarmonyOS: An Agentic Pipeline with Artifact-Derived Verification
    USarXiv cs.SE (Software)

    arXiv:2610.33197v1 Announce Type: new Abstract: The migration of applications written for a different platforms like Android or iOS to HarmonyOS NEXT requires a complete rewrite in ArkTS. Although LLMs automate the translation of source…

  105. Inductive Grounded Arithmetic: Goodstein, the Hydra, and $\Gamma_0$
    USarXiv cs.PL (Programming)

    arXiv:2610.34548v1 Announce Type: cross Abstract: What does a recursive definition mean when the logic is grounded, so a sentence is true only if a terminating process makes it so? Kripke's answer for the special case of truth was the…

  106. Ownership Refinement Types for Pointer Arithmetic and Nested Arrays
    USarXiv cs.PL (Programming)

    arXiv:2604.22361v2 Announce Type: replace Abstract: Tanaka et al. proposed a type system for verifying functional correctness properties of programs that use arrays and pointer arithmetic. Their system extends ConSORT--a type system…

  107. How Do Coding-Agent Harnesses Self-Evolve? An Empirical Study of Retrospective Adaptation
    USarXiv cs.SE (Software)

    arXiv:2610.34938v1 Announce Type: new Abstract: Coding agents can adapt their harnesses, the software, configuration, and instructions used to run them, based on experience from earlier tasks, but whether these adaptations improve their…

  108. Built to Remember, Shipped to Forget: Operational Contracts, Failures, and Repairs in Containerized LLM Agent Deployments
    USarXiv cs.SE (Software)

    arXiv:2610.34620v1 Announce Type: new Abstract: LLM agent systems reach production as container images run by platforms such as Docker, Compose, and Kubernetes. Framework and platform document opposite expectations: to the platform a…

  109. WebUniverse: Scaling Verified Training Data for Agentic Web Development
    USarXiv cs.SE (Software)

    arXiv:2610.33149v1 Announce Type: new Abstract: Training agents for front-end web development requires demonstrations of code editing, execution, browser inspection and revision, yet public trajectories and reusable task environments…

  110. In the Driver's Seat: A Multi-Company Study on the Reality of Autonomous Driving System Testing
    USarXiv cs.SE (Software)

    arXiv:2607.15820v2 Announce Type: replace Abstract: Autonomous driving systems (ADS) are increasingly deployed in complex real-world environments, creating strong demands for rigorous testing to establish their functionality and safety…

  111. ToolContractGate: Versioned Trusted Contracts for Drift-Safe Agent Tool Invocation
    USarXiv cs.SE (Software)

    arXiv:2610.33108v1 Announce Type: new Abstract: Agent tool approval and input validation do not, by themselves, bind a request to an approved source artifact. ToolContractGate provides a deterministic runtime contract evaluator for…

  112. Enhancing Software Engineering Through Closed-Loop Memory Optimization
    USarXiv cs.SE (Software)

    arXiv:2606.05646v2 Announce Type: replace Abstract: Large language models (LLMs) have enabled powerful software engineering (SE) agents capable of navigating complex codebases and resolving real-world issues. However, these agents remain…

  113. The Policies of Programming Language Ecosystems: A "Qualitative" Analysis
    USarXiv cs.SE (Software)

    arXiv:2610.34187v1 Announce Type: new Abstract: Programming language ecosystems (PLEs), such as CRAN for R, have greatly accelerated software reuse. At the same time, they have increased software supply-chain risks and incidents. However…

  114. Build or Adopt? Custom-Built vs. General-Purpose LLM Agent Harnesses for Automotive GUI Testing
    USarXiv cs.SE (Software)

    arXiv:2610.33636v1 Announce Type: new Abstract: As general-purpose LLM agent harnesses mature, organizations must decide whether to continue developing in-house systems or adopt existing alternatives. We study this decision at Volvo Group…

  115. DRILS: Dependency-based Requirement Change Impact Analysis Using LLMs
    USarXiv cs.SE (Software)

    arXiv:2610.33920v1 Announce Type: new Abstract: Change is inevitable in software systems, where requirements must continuously adapt to evolving stakeholder needs, regulations, and project scopes. Managing such changes is a fundamental…

  116. A Multi-Domain Study on the Effectiveness of LLMs at Writing Property-Based Tests
    USarXiv cs.SE (Software)

    arXiv:2610.34487v1 Announce Type: new Abstract: A property-based test (PBT) checks a property of a program on many generated inputs, but writing one takes a property, a generator of valid inputs, and an oracle. LLM-based PBT writers…

  117. Broken on Arrival: Non-Functional Quantized LLM Artifacts in Public Model Registries and How to Catch Them
    USarXiv cs.SE (Software)

    arXiv:2609.05881v2 Announce Type: replace Abstract: Developers download quantized large language models (LLMs) from public registries to run them locally. None of the registries we examined documented functional testing of their files. We…

  118. EviACT: An Evidence-to-Action Framework for Agentic Program Repair
    USarXiv cs.SE (Software)

    arXiv:2605.27238v2 Announce Type: replace Abstract: LLM-based agents have moved automated program repair (APR) from fixed-context patch generation to interactive repository-level repair. However, existing agentic APR systems still struggle…

  119. A Difference Without a Distinction: A Measurement Study of Release Trust Evidence in Package Registries
    USarXiv cs.SE (Software)

    arXiv:2606.22593v3 Announce Type: replace Abstract: A stolen npm token or account lets an attacker publish a malicious version of a popular package. Such uploads skip the project's CI workflow and can lack its provenance, so since November…

  120. Interaction Improvement
    USarXiv cs.PL (Programming)

    arXiv:2601.01638v3 Announce Type: replace-cross Abstract: The relational semantics of linear logic is a powerful framework for defining resource-aware models of the $\lambda$-calculus. However, its quantitative aspects are not reflected in…

  121. How Quickly Do Development Teams Update Their Vulnerable Dependencies?
    USarXiv cs.SE (Software)

    arXiv:2403.17382v5 Announce Type: replace Abstract: Industry practitioners are increasingly concerned with software that contains vulnerable versions of third-party dependencies that are included both directly and transitively. To address…

  122. The Agent in the Middle: How AI-Mediated Access Reshapes Web Vulnerability Exploitation
    USarXiv cs.SE (Software)

    arXiv:2610.35541v1 Announce Type: cross Abstract: Web services increasingly expose backend interfaces that let users delegate tasks to AI agents, which invoke service operations directly without navigating websites. This shift changes how…

  123. Do AI Coding Agents Write Infrastructure as Code Like Humans? An Empirical Study of GitHub Pull Requests
    USarXiv cs.SE (Software)

    arXiv:2610.34613v1 Announce Type: new Abstract: Infrastructure-as-Code (IaC) configurations define cloud resources, networking, and access control, and AI coding agents increasingly write them. Little is known about how agent contributions…

  124. Silent and Costly: Characterizing and Detecting Prompt-Caching Mis-Integration in LLM Agents
    USarXiv cs.SE (Software)

    arXiv:2610.34817v1 Announce Type: new Abstract: Large language model (LLM) services process massive volumes of user requests, resulting in substantial inference costs. To reduce both cost and latency, providers have introduced prompt…

  125. MulFCoder: Framework-conditioned Multi-agent for MLLM-based Multi-framework Front-end Code Generation
    USarXiv cs.SE (Software)

    arXiv:2610.34418v1 Announce Type: new Abstract: Generating runnable front-end code from UI screenshots is a long-standing goal in automated software engineering. Existing MLLM-based methods predominantly focused on HTML/CSS, leaving…

  126. The Hidden Costs of Forgetting: A Study of Trajectory Deletion in Software Engineering Agents
    USarXiv cs.SE (Software)

    arXiv:2610.35496v1 Announce Type: new Abstract: Software engineering agents resend their entire trajectory to an LLM at every turn, motivating online trajectory reduction. Prior work shows that randomly deleting 75% of tokens in old turns…

  127. V$^3$AST-Bench: Verifiable, Valid, and Versatile Safety Evaluation for Coding Agents
    USarXiv cs.SE (Software)

    arXiv:2610.35188v1 Announce Type: new Abstract: Large language models increasingly act through code, modifying repositories, executing programs, and interacting with services in agent form. Their actions introduce risks that cannot be…

  128. SpeC2Rust: Specification-Driven Repository-Level C-to-Rust Translation
    USarXiv cs.SE (Software)

    arXiv:2610.34141v1 Announce Type: new Abstract: Legacy C codebases remain central to system software and performance-critical infrastructure, but their weak memory safety imposes security and maintenance risks. Recent C-to-Rust translation…

  129. Securing Agent-Generated Code with Its Own Development History
    USarXiv cs.SE (Software)

    arXiv:2610.34098v1 Announce Type: new Abstract: Pull requests written by coding agents can pass a project's test suite while omitting a guard or a validation step. To find the missing condition, a reviewer currently relies on static…

  130. ArchGuard-APR: Architecture-Constraint-Aware Multi-Agent Program Repair Using Repository Graphs and Execution Feedback
    USarXiv cs.SE (Software)

    arXiv:2610.33853v1 Announce Type: new Abstract: Repository-level automated program repair must reason about more than a localized failing statement. A candidate change can satisfy a narrow test while violating dependency boundaries…

  131. Special Delivery: Programming with Mailbox Types (Extended Version)
    USarXiv cs.PL (Programming)

    arXiv:2306.12935v5 Announce Type: replace Abstract: The asynchronous and unidirectional communication model supported by mailboxes is a key reason for the success of actor languages like Erlang and Elixir for implementing reliable and…

  132. Speculative Forking: Turning Idle Wait into Parallel Exploration for Repository Environment Setup
    USarXiv cs.SE (Software)

    arXiv:2610.33129v1 Announce Type: new Abstract: Before developers can use a software project, they must first get it to run. This often requires installing the right language runtime, system packages, build tools, and test commands. Recent…

  133. OneCodeReviewBench: Typed Decisions for Code Review and Search
    USarXiv cs.SE (Software)

    arXiv:2610.33389v1 Announce Type: new Abstract: Code review requires finding functional bugs, memory-safety errors, security weaknesses, and performance problems across languages and software stacks. Tools can express bounded parts of this…

  134. SpecAgent: Empowering Program Verification with Agentic Synthesis of Formal Program Specifications
    USarXiv cs.SE (Software)

    arXiv:2610.05132v2 Announce Type: replace Abstract: Formal specifications are essential for deductive program verification, providing semantic abstractions for compositional verification of complex software. However, manually constructing…

  135. A Rowhammer attack on Fil-C's software capabilities
    USarXiv cs.PL (Programming)

    arXiv:2610.33748v1 Announce Type: new Abstract: Rowhammer is a hardware vulnerability in dynamic random-access memory (DRAM) in which repeated accesses to one or more aggressor rows can induce bit-flips in nearby victim rows. This…

  136. Responsibility Boundaries for Agentic Systems: A Comparative Evaluation of Probabilistic Agentic Programming
    USarXiv cs.PL (Programming)

    arXiv:2610.33312v1 Announce Type: cross Abstract: Language-model agents combine reasoning, tools, memory, delegation, and external effects. Authority is often distributed across prompts, host code, configuration, and logs. Probabilistic…

  137. Developers as Managers of AI Agents: A Workplace Shadowing Study
    USarXiv cs.SE (Software)

    arXiv:2610.33588v1 Announce Type: new Abstract: As AI coding agents become more autonomous, developers increasingly direct, coordinate, and check their work. Yet little is known about these practices in everyday professional development…

  138. Ethics by Configuration: AI Ethics Principles in Coding Agent Instructions
    USarXiv cs.SE (Software)

    arXiv:2610.34847v1 Announce Type: new Abstract: As AI coding agents become integral to software development, projects increasingly utilise configuration files, such as context files and skills, to govern agent behaviour. While AI ethics…

  139. What Will Replace Pull Requests? Rethinking Software Collaboration for AI Agents
    USarXiv cs.SE (Software)

    arXiv:2610.34933v1 Announce Type: new Abstract: The emergence of artificial intelligence (AI) coding agents has put pull requests under pressure: agents open millions of pull requests per month, while some open-source projects have begun…

  140. PMU Foundation Model Bridge: A Tool for Connecting Synchrophasor Streams to Time Series Foundation Models
    USarXiv cs.SE (Software)

    arXiv:2610.33387v1 Announce Type: cross Abstract: This paper presents PMU Foundation Model Bridge, an open source tool that connects online phasor measurement unit (PMU) or phasor data concentrator (PDC) streams to publicly available time…

  141. Secure and Functional Code Generation with Self-Evaluation
    USarXiv cs.SE (Software)

    arXiv:2610.33450v1 Announce Type: cross Abstract: Large Language Models (LLMs) can generate functionally correct code that remains insecure. Security-focused fine-tuning can be expensive and is not always available, motivating…

  142. AgentProofBench: Benchmarking Reusable Verifiers for Web Agents
    USarXiv cs.SE (Software)

    arXiv:2610.34634v1 Announce Type: cross Abstract: Reusable Web-agent verification requires a rule to follow changing requirements and support subsequent execution. We introduce AgentProofBench, which connects paired tests of a frozen…

  143. Beyond Correctness: Attribution of Natural-Language Documentation for Formal Libraries
    USarXiv cs.SE (Software)

    arXiv:2610.33094v1 Announce Type: new Abstract: Formal languages are widely used in software engineering to specify and verify program properties. Because their mathematical notation and logic are hard to read, developers often rely on…

  144. AIPC: Evidence-Guided Agent Workflows for PyTorch Project Deployment
    USarXiv cs.SE (Software)

    arXiv:2610.34890v1 Announce Type: new Abstract: Deploying and optimizing a PyTorch inference project on a dedicated accelerator, especially an on-device NPU, requires more than using a framework device API as with CUDA. The project must be…

  145. Position, Role or Label? A Controlled Study of Instruction Conflicts in LLM Agents
    USarXiv cs.SE (Software)

    arXiv:2610.33201v1 Announce Type: cross Abstract: LLM agents take instructions from the operator, repository configuration, user, tool results and saved memory, which often conflict. Instruction-hierarchy benchmarks ask whether the…

  146. GEAR: Feature-Based Configuration and Code Generation for Agentic Systems
    USarXiv cs.SE (Software)

    arXiv:2610.35592v1 Announce Type: cross Abstract: Engineering multi-agent systems (MAS) requires choosing agents, orchestration strategies, communication mechanisms, reasoning approaches, and memory configurations. These choices depend on…

  147. What's in a Package? Getting Visibility Into Dependencies Using Security-Sensitive API Calls
    USarXiv cs.SE (Software)

    arXiv:2408.02846v3 Announce Type: replace-cross Abstract: The Log4Shell vulnerability occurred because the JNDI (Java Naming and Directory Interface) lookup feature was used, even though it was not needed for log4j's core functionality…

  148. Arya Forge: Auditing Validation and Release Approval for Financial Instruction Data
    USarXiv cs.SE (Software)

    arXiv:2610.34778v1 Announce Type: cross Abstract: Financial instruction-data pipelines must distinguish content validity from permission to release a particular dataset. Arya Forge combines typed records, current-content checks, and…

  149. Understanding Python Linters through a Comprehensive and Systematic Empirical Study
    USarXiv cs.SE (Software)

    arXiv:2610.35485v1 Announce Type: new Abstract: Code linters, tools that automatically detect code issues, help developers accelerate development and improve code quality by identifying errors early. Unlike previous studies that focused…

  150. VulDec: Vulnerability Detection in Binaries via Neural Decompilation
    USarXiv cs.SE (Software)

    arXiv:2610.33716v1 Announce Type: cross Abstract: Automated vulnerability detection in binary executables is challenging due to the lack of source code and the limited semantic information available at the assembly level. Despite…