This paper introduces MM-IssueLoc, a benchmark and evaluation protocol for repository-level issue localization with visual evidence.

Phoenix is a multi-agent system that uses seven safety controls and a test evaluation strategy to resolve GitHub issues, achieving 75% oracle-resolution with no regressions on a curated benchmark and…

The paper introduces Kumushi, a root-cause-driven patching agent that significantly improves automated vulnerability repair by focusing LLMs on the true source of bugs, outperforming existing methods…

This paper evaluates the improvement of statistical fault localization by augmenting it with execution features.

Proposed ProjAgent, a repository-level code generation system using procedural similarity for retrieval and conservative static-analysis feedback.

This paper explores the causes of inconsistent vulnerability scanner findings in the open-source ecosystem.

This paper proposes a framework to transform historical real-bug mechanisms into executable feedback targets for LLM-based unit test generation, improving real-bug detection over various baselines.

This paper investigates the relationship between bug report features and repair success for automated program repair agents, finding that concrete, executable, and well-localized information is most b…

Antaeus is a framework that uses repository-level code context to detect logic vulnerabilities using LLM reasoning, reducing calls, cost, and triage effort.

The paper introduces Code-QA-Bench, a novel framework that rigorously separates genuine code reasoning from mere documentation memorization in repository-level code understanding benchmarks.

The paper introduces CrossCommitVuln-Bench, a benchmark dataset demonstrating that many real-world Python vulnerabilities are introduced across multiple commits, making them invisible to standard per-…

This study provides an ecosystem-scale measurement of commit signing on GitHub, finding that current signing adoption rates are misleading and that developers struggle to maintain consistent, long-ter…

The paper introduces Heimdall, an automated pipeline that uses LLMs and formal verification to safely and automatically migrate legacy, potentially buggy eBPF programs written in C to memory-safe Rust…

This paper introduces CORE-Bench, a comprehensive benchmark for code retrieval in agentic coding.

The paper reframes LLM reliability from an impossible universal problem to a manageable, local patch-based problem, showing that sufficient interventions can be found by focusing on recurring failure…

The paper critiques current AI benchmarking practices for low-resource settings, arguing that evaluation must shift focus from isolated model performance to the holistic performance of the deployed sy…

This paper presents AgentLocate, a framework for failure localization in large language model-based multi-agent systems using an LLM-based judging mechanism and multi-perspective verification.