Research

Paper

AI LLM March 18, 2026

FailureMem: A Failure-Aware Multimodal Framework for Autonomous Software Repair

Authors

Ruize Ma, Yilei Jiang, Shilin Zhang, Zheng Ma, Yi Feng, Vincent Ng, Zhi Wang, Xiangyu Yue, Chuanyi Li, Lewei Lu

Abstract

Multimodal Automated Program Repair (MAPR) extends traditional program repair by requiring models to jointly reason over source code, textual issue descriptions, and visual artifacts such as GUI screenshots. While recent LLM-based repair systems have shown promising results, existing approaches face several limitations: rigid workflow pipelines restrict exploration during debugging, visual reasoning is often performed over full-page screenshots without localized grounding, and failed repair attempts are rarely transformed into reusable knowledge. To address these challenges, we propose FailureMem, a multimodal repair framework that integrates three key mechanisms: a hybrid workflow-agent architecture that balances structured localization with flexible reasoning, active perception tools that enable region-level visual grounding, and a Failure Memory Bank that converts past repair attempts into reusable guidance. Experiments on SWE-bench Multimodal demonstrate FailureMem improves the resolved rate over GUIRepair by 3.7%.

Metadata

arXiv ID: 2603.17826
Provider: ARXIV
Primary Category: cs.SE
Published: 2026-03-18
Fetched: 2026-03-19 06:01

Related papers

Raw Data (Debug)
{
  "raw_xml": "<entry>\n    <id>http://arxiv.org/abs/2603.17826v1</id>\n    <title>FailureMem: A Failure-Aware Multimodal Framework for Autonomous Software Repair</title>\n    <updated>2026-03-18T15:24:22Z</updated>\n    <link href='https://arxiv.org/abs/2603.17826v1' rel='alternate' type='text/html'/>\n    <link href='https://arxiv.org/pdf/2603.17826v1' rel='related' title='pdf' type='application/pdf'/>\n    <summary>Multimodal Automated Program Repair (MAPR) extends traditional program repair by requiring models to jointly reason over source code, textual issue descriptions, and visual artifacts such as GUI screenshots. While recent LLM-based repair systems have shown promising results, existing approaches face several limitations: rigid workflow pipelines restrict exploration during debugging, visual reasoning is often performed over full-page screenshots without localized grounding, and failed repair attempts are rarely transformed into reusable knowledge. To address these challenges, we propose FailureMem, a multimodal repair framework that integrates three key mechanisms: a hybrid workflow-agent architecture that balances structured localization with flexible reasoning, active perception tools that enable region-level visual grounding, and a Failure Memory Bank that converts past repair attempts into reusable guidance. Experiments on SWE-bench Multimodal demonstrate FailureMem improves the resolved rate over GUIRepair by 3.7%.</summary>\n    <category scheme='http://arxiv.org/schemas/atom' term='cs.SE'/>\n    <category scheme='http://arxiv.org/schemas/atom' term='cs.AI'/>\n    <published>2026-03-18T15:24:22Z</published>\n    <arxiv:primary_category term='cs.SE'/>\n    <author>\n      <name>Ruize Ma</name>\n    </author>\n    <author>\n      <name>Yilei Jiang</name>\n    </author>\n    <author>\n      <name>Shilin Zhang</name>\n    </author>\n    <author>\n      <name>Zheng Ma</name>\n    </author>\n    <author>\n      <name>Yi Feng</name>\n    </author>\n    <author>\n      <name>Vincent Ng</name>\n    </author>\n    <author>\n      <name>Zhi Wang</name>\n    </author>\n    <author>\n      <name>Xiangyu Yue</name>\n    </author>\n    <author>\n      <name>Chuanyi Li</name>\n    </author>\n    <author>\n      <name>Lewei Lu</name>\n    </author>\n  </entry>"
}