Intel / article

XAgent: eXecution-guided Agentic AI for Effective Localization and Resolution of GitHub Issues

Source: arxiv.org
Author: Hieu Huynh, Patanamon Thongtanunam, Michael Fu, Bach Le, Kla Tantithamthavorn
Date: Published Sep 10, 2026

Context

XAgent uses information about a program's behavior during execution to help find and repair software defects. The researchers' diagnosis is that an relying on an issue description can search too narrowly, identify the wrong location or produce an incomplete repair. Their framework adds observed behavior and program to both localization and validation.

In the authors' -lite evaluation, XAgent resolved 62% of issues and achieved 72.8% accuracy at identifying the relevant function. These measure different steps: finding where a problem lives and successfully fixing it. The reported result supports execution-guided investigation as a promising repair strategy within that .

Terms in this piece · Glossary

AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
context window — The maximum amount of text a model can consider at once — its working memory for the current conversation or task.
SWE-bench — The standard benchmark for AI coding agents: real GitHub issues from real repositories, scored by whether the agent's patch passes the project's own tests.
benchmark — A standard public test set for comparing AI models — the shared scoreboards behind every "model X beats model Y" claim.

Key quotes

“Solely relying on this information can bias LLM reasoning toward the narrow scope of the issue description, leading to incomplete patches that fail to address the underlying issue.”

“XAgent successfully resolves 7 additional issues that the top existing baselines fail to address.”

Read the source arxiv.org