AutoCodeSherpa: Symbolic Explanations in AI Coding Agents

• URL: http://arxiv.org/abs/2507.22414v1
• Date: Wed, 30 Jul 2025 06:34:02 GMT
• Title: AutoCodeSherpa: Symbolic Explanations in AI Coding Agents
• Authors: Sungmin Kang, Haifeng Ruan, Abhik Roychoudhury,
• Abstract summary: Large Language Model (LLM) agents autonomously use external tools on top of one or more LLMs to accomplish specific tasks.<n> Lately LLM agents for software engineering tasks have become popular.<n>This is demonstrated by existing agentic AI solutions such as AutoCodeRover or SpecRover which perform automated program repair.
• Score: 10.706082274730734
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Large Language Model (LLM) agents autonomously use external tools on top of one or more LLMs to accomplish specific tasks. Lately LLM agents for software engineering tasks have become popular. These agents can benefit from the use of program analysis tools working on program representations. This is demonstrated by existing agentic AI solutions such as AutoCodeRover or SpecRover which perform automated program repair. Specifically the goal of these works is to use program analysis to improve the patch quality. These agents are currently being used to automatically fix static analysis issues from the widely used SonarQube static analyzer. Nevertheless, for the agents to be deployed in a production environment, agents need to suggest software artifacts, such as patches, with evidence and with high confidence. In this work, we provide a workflow where an agent provides explanations of the bug in the form of symbolic formulae. The explanations are in the form of input conditions, infection conditions and output conditions, implemented as property based tests (PBT) and program-internal symbolic expressions. These can help in human developer cognition of the agent outputs as well as in achieving completely automated agentic workflows for software. The human developer can benefit from the input condition, represented as a PBT, to generate various concrete inputs showing a given issue. Furthermore, since the PBTs are executable, our explanations are executable as well. We can thus also use the explanations in a completely automated issue resolution environment for accepting or rejecting the patches that are suggested by patching agents such as AutoCodeRover. Finally, as agentic AI approaches continue to develop, the program analysis driven explanations can be provided to other LLM-based repair techniques such as Agentless to improve their output.

Related papers

• RepoLaunch: Automating Build&Test Pipeline of Code Repositories on ANY Language and ANY Platform [49.43594274832262]
  We introduce RepoLaunch, the first agent capable of automatically resolving dependencies, compiling source code, and extracting test results for repositories across arbitrary programming languages and operating systems.<n>RepoLaunch automates the remaining steps, enabling scalable benchmarking and training of coding agents and LLMs.
  arXiv  Detail & Related papers  (2026-03-05T10:15:13Z)
• Simple Agents Outperform Experts in Biomedical Imaging Workflow Optimization [69.36509281190662]
  Adapting production-level computer vision tools to bespoke scientific datasets is a critical "last mile" bottleneck.<n>We consider using AI agents to automate this manual coding, and focus on the open question of optimal agent design.<n>We demonstrate that a simple agent framework consistently generates adaptation code that outperforms human-expert solutions.
  arXiv  Detail & Related papers  (2025-12-02T18:42:26Z)
• Are Agents Just Automata? On the Formal Equivalence Between Agentic AI and the Chomsky Hierarchy [4.245979127318219]
  This paper establishes a formal equivalence between the architectural classes of modern agentic AI systems and the abstract machines of the hierarchy.<n>We demonstrate that simple reflex agents are equivalent to Finite Automata, hierarchical task-decomposition agents are equivalent to Pushdown Automata, and agents employing readable/writable memory for reflection are equivalent to TMs.
  arXiv  Detail & Related papers  (2025-10-27T16:22:02Z)
• How Can Input Reformulation Improve Tool Usage Accuracy in a Complex Dynamic Environment? A Study on $τ$-bench [58.114899897566964]
  In a multi-turn conversational environment, large language models (LLMs) often struggle with consistent reasoning and adherence to domain-specific policies.<n>We propose the Input-Reformulation Multi-Agent (IRMA) framework, which automatically reformulates user queries augmented with relevant domain rules.<n>IRMA significantly outperforms ReAct, Function Calling, and Self-Reflection by 16.1%, 12.7%, and 19.1%, respectively.
  arXiv  Detail & Related papers  (2025-08-28T15:57:33Z)
• Do AI models help produce verified bug fixes? [62.985237003585674]
  Large Language Models are used to produce corrections to software bugs.<n>This paper investigates how programmers use Large Language Models to complement their own skills.<n>The results are a first step towards a proper role for AI and LLMs in providing guaranteed-correct fixes to program bugs.
  arXiv  Detail & Related papers  (2025-07-21T17:30:16Z)
• Unified Software Engineering agent as AI Software Engineer [14.733475669942276]
  Large Language Model (LLM) technology has raised expectations for automated coding.<n>In this paper, we seek to understand this question by developing a Unified Software Engineering agent or USEagent.<n>We envision USEagent as the first draft of a future AI Software Engineer which can be a team member in future software development teams involving both AI and humans.
  arXiv  Detail & Related papers  (2025-06-17T16:19:13Z)
• ThinkGeo: Evaluating Tool-Augmented Agents for Remote Sensing Tasks [54.52092001110694]
  ThinkGeo is a benchmark designed to evaluate tool-augmented agents on remote sensing tasks via structured tool use and multi-step planning.<n>Inspired by tool-interaction paradigms, ThinkGeo includes human-curated queries spanning a wide range of real-world applications.<n>Our analysis reveals notable disparities in tool accuracy and planning consistency across models.
  arXiv  Detail & Related papers  (2025-05-29T17:59:38Z)
• LLM Agents Making Agent Tools [2.5529148902034637]
  Tool use has turned large language models (LLMs) into powerful agents that can perform complex multi-step tasks.<n>But these tools must be implemented in advance by human developers.<n>We propose ToolMaker, an agentic framework that autonomously transforms papers with code into LLM-compatible tools.
  arXiv  Detail & Related papers  (2025-02-17T11:44:11Z)
• Defining and Detecting the Defects of the Large Language Model-based Autonomous Agents [31.126001253902416]
  We present the first study focused on identifying and detecting defects in LLM Agents.<n>We collected and analyzed 6,854 relevant posts from StackOverflow to define 8 types of agent defects.<n>Our results show that Agentable achieved an overall accuracy of 88.79% and a recall rate of 91.03%.
  arXiv  Detail & Related papers  (2024-12-24T11:54:14Z)
• Agentless: Demystifying LLM-based Software Engineering Agents [12.19683999553113]
  We build Agentless -- an agentless approach to automatically solve software development problems. Compared to the verbose and complex setup of agent-based approaches, Agentless employs a simplistic three-phase process of localization, repair, and patch validation. Our results on the popular SWE-bench Lite benchmark show that surprisingly the simplistic Agentless is able to achieve both the highest performance and low cost.
  arXiv  Detail & Related papers  (2024-07-01T17:24:45Z)
• Alibaba LingmaAgent: Improving Automated Issue Resolution via Comprehensive Repository Exploration [64.19431011897515]
  This paper presents Alibaba LingmaAgent, a novel Automated Software Engineering method designed to comprehensively understand and utilize whole software repositories for issue resolution.<n>Our approach introduces a top-down method to condense critical repository information into a knowledge graph, reducing complexity, and employs a Monte Carlo tree search based strategy.<n>In production deployment and evaluation at Alibaba Cloud, LingmaAgent automatically resolved 16.9% of in-house issues faced by development engineers, and solved 43.3% of problems after manual intervention.
  arXiv  Detail & Related papers  (2024-06-03T15:20:06Z)
• SWE-agent: Agent-Computer Interfaces Enable Automated Software Engineering [79.07755560048388]
  SWE-agent is a system that facilitates LM agents to autonomously use computers to solve software engineering tasks. SWE-agent's custom agent-computer interface (ACI) significantly enhances an agent's ability to create and edit code files, navigate entire repositories, and execute tests and other programs. We evaluate SWE-agent on SWE-bench and HumanEvalFix, achieving state-of-the-art performance on both with a pass@1 rate of 12.5% and 87.7%, respectively.
  arXiv  Detail & Related papers  (2024-05-06T17:41:33Z)
• NExT: Teaching Large Language Models to Reason about Code Execution [50.93581376646064]
  Large language models (LLMs) of code are typically trained on the surface textual form of programs. We propose NExT, a method to teach LLMs to inspect the execution traces of programs and reason about their run-time behavior.
  arXiv  Detail & Related papers  (2024-04-23T01:46:32Z)
• A Preliminary Study on Using Large Language Models in Software Pentesting [2.0551676463612636]
  Large language models (LLM) are perceived to offer promising potentials for automating security tasks. We investigate the use of LLMs in software pentesting, where the main task is to automatically identify software security vulnerabilities in source code.
  arXiv  Detail & Related papers  (2024-01-30T21:42:59Z)
• ProAgent: From Robotic Process Automation to Agentic Process Automation [87.0555252338361]
  Large Language Models (LLMs) have emerged human-like intelligence. This paper introduces Agentic Process Automation (APA), a groundbreaking automation paradigm using LLM-based agents for advanced automation. We then instantiate ProAgent, an agent designed to craft from human instructions and make intricate decisions by coordinating specialized agents.
  arXiv  Detail & Related papers  (2023-11-02T14:32:16Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.