Automated Program Repair

Follow this topic Ask about this topic

Automated program repair studies how to find and correct bugs while preserving a program’s intended behavior. Approaches range from search-based patch generation and program analysis to language models that use fault localization, repository context, retrieved examples, and iterative agent workflows. Tests, runtime traces, static analysis, and formal methods can guide or verify repairs, with benchmarks spanning small functions, repositories such as those tested by SWE-bench, and security or hardware code. A central challenge is that compiling or passing available tests does not guarantee a secure, semantically correct, minimal fix, so stronger oracles must be balanced against context limits, cost, and generalization.

Latest papers

159 papers33 papers in the last four weeks, up 74% on the four weeks before. 0.2% of all new papers.

Falsification, Not Exposure: An Internally Preregistered Placebo-Controlled Decomposition of Self-Repair Feedback in Frozen Small Code Models
Jun 30, 2026Mehmet IscanLLM Self-Correction Automated Program Repair
PythaLab, Yıldız Technical University, Istanbul, Turkey
Loc2Repair: A Framework for Evaluating the Impact of File-Level Issue Localization in Repo-Level LLM Repair
Jun 29, 2026Mohammad Nour Al Awad, Sergey IvanovAutomated Program Repair Software Engineering Benchmarks
ITMO University
Diff-Based Code Corruption using LLMs for Large-Scale Bugfix Benchmarking
Jun 27, 2026Balázs Szalontai, Ábel Szauter, Balázs Márton +3Software Engineering Automated Program Repair
Eötvös Loránd University · Faculty of Informatics
Dockerless: Environment-Free Program Verifier for Coding Agents
Jun 26, 2026Wenhao Zeng, Yuling Shi, Xiaodong Gu +10Automated Program Repair AI Coding Agents
Shanghai Jiao Tong University · Douyin Group
Empirical Software Engineering TerraProbe: A Layered-Oracle Framework for Detecting Deceptive Fixes in LLM-Assisted Terraform
Jun 25, 2026Manar Alsaid, Chimdumebi Nebolisa, Faris AbbasLLM Evaluation Automated Program Repair
Department of Computer Science and Information Systems, East Texas A&M University, 2600 W. Neal Street, Commerce, 75428, Texas, United States · Department of Dual Enrollment and P-16 Programs, Texas Woman’s University, 3042026 Administration Drive, Denton, 76204, Texas, United States
Evidence-Driven LLM Agent for C-to-Synthesizable-C Conversion and Verification
Jun 25, 2026Zhe Zhao, Hongbing Lang, Zhihan Xiao +3Automated Program Repair Code Generation
Shenzhen International Graduate School, Tsinghua University, Shenzhen 518055, China · Shenzhen Belon Technology Co., Ltd., Shenzhen 518055, China
CyberChainBench: Can AI Agents Secure Smart Contracts Against Real-World On-Chain Vulnerabilities?
Jun 24, 2026Jintao Huang, Fengqing Jiang, Radha Poovendran +1Automated Program Repair Software Vulnerability Detection
1The Ohio State University · University of Washington
Helpful or Harmful? Evaluating LLM-Assisted Vulnerability Patching via a Human Study
Jun 24, 2026Giulian Biolo, Michael Tezza, Yuanjun Gong +1Automated Program Repair Human-in-the-Loop Evaluation
University of Trento, IT · Vrije Universiteit Amsterdam, NL
VeriPilot: An LLM-Powered Verilog Debugging Framework
Jun 22, 2026Yihan Wang, Cheng Liu, Jiazheng Zhang +4Automated Program Repair Electronic Design Automation
State Key Lab of Processors, Institute of Computing Technology, Chinese Academy of Sciences, and the University of Chinese Academy of Sciences, Beijing, China · North China Electric Power University, Beijing 102206, China
RAVEN: Agentic RAG for Automated Vulnerability Repair
Jun 21, 2026Varun Gadey, Zijie Liu, Alexandra DmitrienkoRetrieval-Augmented Generation Automated Program Repair
University of Duisburg-Essen, Germany
TraceView: Interactive Visualization of Agentic Program Repair Trajectories
Jun 20, 2026Amirali Sajadi, Tu Nguyen, Kimmie Huynh +2Agent Failure Analysis Software Engineering Agents
Drexel University, Philadelphia, PA, USA · Belmont University, Nashville, TN, USA
Library-Aware Doubles and Iterative Repair for Large Language Model-Generated Unit Tests in OpenSIL Firmware
Jun 18, 2026Ma Toan Bach, Yuchi Zheng, Haingo Razafindranto +4Automated Software Testing Automated Program Repair
School of Software Design and Data Science Seneca Polytechnic Canada · Advanced Micro Devices Canada
Beyond Problem Solving: UOJ-Bench for Evaluating Code Generation, Hacking, and Repair in Competitive Programming
Jun 11, 2026Tingqiang Xu, Hangrui Zhou, Tianle Cai +2Automated Program Repair Competitive Programming
Institute for Interdisciplinary Information Sciences, Tsinghua University · Universal Online Judge · ByteDance Seed +1
CyberGym-E2E: Scalable Real-World Benchmark for AI Agents' End-to-End Cybersecurity Capabilities
Jun 3, 2026Tianneng Shi, Robin Rheem, Dongwei Jiang +13Automated Program Repair Software Vulnerability Detection
1UC Berkeley · 2Johns Hopkins University · 3UC Santa Cruz +1
FLARE: Fine-Grained Diagnostic Feedback for LLM Code Refinement
Jun 2, 2026Yinsheng Yao, Hongxiang Zhang, Weixi Tong +1LLM Self-Refinement Automated Program Repair
Purdue University West Lafayette, IN, USA
Knowledge boundary probing and demand-guided intervention for LLM-based power system code generation
May 29, 2026Hui Wu, Xiaoyang Wang, Zhong FanAutomated Program Repair Code Generation
Department of Engineering, University of Exeter · Department of Computer Science, University of Exeter
HTMLCure: Turning Browser Experience into State Guided Repair for Interactive HTML
May 26, 2026Jiajun Wu, Jian Yang, Tuney Zheng +4Automated Software Testing Automated Program Repair
1Beihang University · 2IQuest Research · 3Peking University
ContraFix: Skill-Enhanced Contrastive Runtime Analysis for Vulnerability Repair
May 17, 2026Simiao Liu, Fang Liu, Peiding Wang +4Automated Program Repair
Beihang University Beijing, China · The University of Hong Kong Hong Kong, China
MemRepair: Hierarchical Memory for Agentic Repository-Level Vulnerability Repair
May 17, 2026Simiao Liu, Li Zhang, Fang Liu +3Automated Program Repair Software Security
Beihang University Beijing, China · The University of Hong Kong Hong Kong, China
Hydra: Efficient, Correct Code Generation via Checkpoint-and-Rollback Support
May 14, 2026Alexander Du, Jianjun Ou, Danyang Zhuo +1Automated Program Repair Code Generation
Duke University Durham, NC, USA
BoostAPR: Boosting Automated Program Repair via Execution-Grounded Reinforcement Learning with Dual Reward Models
May 9, 2026Yuanhao Li, Hongbo Wang, Xiaotang Shang +3Automated Program Repair Process Reward Models
State Key Laboratory of Networking and Switching Technology, Beijing University of Posts and Telecommunications, Beijing 100876, China · University of Luxembourg
Debugging the Debuggers: Failure-Anchored Structured Recovery for Software Engineering Agents
May 9, 2026Chenyu Zhao, Shenglin Zhang, Yihang Lin +7Agent Failure Analysis Software Engineering Agents
Nankai University Tianjin, China · Tsinghua University Beijing, China · Microsoft Redmond, USA
CUDABeaver: Benchmarking LLM-Based Automated CUDA Debugging
May 8, 2026Shiyang Li, Haoyang Chen, Mattia Fazzini +1LLM Evaluation Automated Program Repair
University of Minnesota
TraceFix: Repairing Agent Coordination Protocols with TLA+ Counterexamples
May 8, 2026Shuren Xia, Qiwei Li, Taqiya Ehsan +1Multi-Agent LLM Systems Multi-Agent Coordination
Rutgers University Piscataway, NJ, USA
Signal Reshaping for GRPO in Weak-Feedback Agentic Code Repair
May 8, 2026Jia Li, Yuxin Su, Ting Peng +3RL for Code Generation Reward Shaping
The Chinese University of Hong Kong · Sun Yat-sen University · Tencent
SmellBench: Evaluating LLM Agents on Architectural Code Smell Repair
May 7, 2026Ion George Dinu, Marian Cristian Mihăescu, Traian RebedeaAutomated Program Repair LLM Agent Evaluation
University of Craiova Craiova, Romania · NVIDIA University Politehnica of Bucharest Bucharest, Romania
Bridging the Last Mile of Circuit Design: PostEDA-Bench, a Hierarchical Benchmark for PPA Convergence and DRC Fixing
May 7, 2026Pengju Liu, Nuo Xu, Jinwei Tang +2Automated Program Repair Electronic Design Automation
University of Minnesota
ARISE: A Repository-level Graph Representation and Toolset for Agentic Program Repair and Fault Localization
May 4, 2026Shahd Seddik, Fahd Seddik, Amirrezza Esmaeili +2Automated Program Repair Static Code Analysis
FARD Lab, University of British Columbia, Kelowna, Canada
GeoContra: From Fluent GIS Code to Verifiable Spatial Analysis with Geography-Grounded Repair
May 1, 2026Yinhao Xiao, Rongbo Xiao, Yihan ZhangLLM Self-Correction Automated Program Repair
School of Big Data and Artificial Intelligence, Guangdong University of Finance and Economics · School of Geography and Environment Economics, Guangdong University of Finance & Economics · Guangdong Engineering Research Center of Low-Altitude Remote Sensing Intelligent Monitoring +1
FGDM: Reasoning Aware Multi-Agentic Framework for Software Bug Detection using Chain of Thought and Tree of Thought Prompting
Apr 27, 2026Srita Padmanabhuni, Bhargavi Karuturi, Jerusha Karen Indupalli +2Multi-Agent LLM Systems Automated Program Repair
SRM University, Andhra Pradesh, Amaravati, India · Corresponding Author

Jun 30, 2026J/K move · Enter open · S save

Falsification, Not Exposure: An Internally Preregistered Placebo-Controlled Decomposition of Self-Repair Feedback in Frozen Small Code Models

Mehmet Iscan

PythaLab, Yıldız Technical University, Istanbul, Turkey

LLM Self-Correction Automated Program Repair Execution-Guided Code Generation Code Generation Evaluation

Open paper AskPDF arXiv

In deployment settings where retraining is infeasible, small frozen code models are routinely asked to repair a failed program after seeing their own failing output, usually treated as a retry mechanism. From a Popperian view, a generated program is a conjecture and a test-execution violation is an oracle-relative, executable counterexample, so feedback's value should be attributed not to re-exposure to failing code but to whether the conjecture is opened to external, executable criticism. As the third stage of a falsification-centered measurement program, this study builds a placebo-controlled instrument that decomposes the feedback packet against a blind-resampling baseline at matched output-generation budget and against content-free, shape-matched placebos. The contribution is not a new repair algorithm but a reflexive methodology (packet decomposition, placebo mirroring, matched-budget discordant-pair tests, fresh-generation confirmation, executable audits) that makes both the model's program conjecture and the researcher's "feedback content works" claim falsifiable. Across six HumanEval+/MBPP+ cells with three 0.5B-1.5B frozen models, 290 dead task-cell units (no best-of-8 candidate passing the public tier) were evaluated; the main run produced 7,000 fresh generations and a preregistered follow-up 1,400 more. Blind resampling exceeded bare-code retry by +18 net unlocks (25/7, Holm p=0.0021). Code-plus-facts recovered +18 over bare code (21/3, p=0.00042) and +15 over a generic-bullet placebo (p=0.0041). An instruction-only effect was not distinguishable (+3, p=0.36). Code-plus-facts and blind resampling tied at 26 unlocks each (not equivalence). Six external-controller follow-ups tied a content-free shape placebo. In this regime, falsification helped not as vocabulary or self-critique, but as comparison with external, executable counterexamples.