Prometheus: Unified Knowledge Graphs for Issue Resolution in Multilingual Codebases

Yue Pan, Zimin Chen, Siyu Lu, Xiang Li, Zhaoyang Chu, JIAYI XU, Claire Le Goues, Martin Monperrus, He Ye

Keywords: Language models, Software engineering, Agent, Issue Resolution, Knowledge Graph

TL;DR: PROMETHEUS is a multi-agent system that builds a code knowledge graph for multilingual repositories, enabling scalable context retrieval and resolving real-world issues beyond benchmarks.

Abstract:

Language model (LM) agents, such as SWE-agent and OpenHands, have made progress toward automated issue resolution. However, existing approaches are often limited to Python-only issues and rely on pre-constructed containers in SWE-bench with reproduced issues, restricting their applicability to real-world and work for multi-language repositories. We present PROMETHEUS , designed to resolve real-world issues beyond benchmark settings. PROMETHEUS is a multi- agent system that transforms an entire code repository into a unified knowl- edge graph to guide context retrieval for issue resolution. PROMETHEUS en- codes files, abstract syntax trees, and natural language text into a graph of typed nodes and five general edge types to support multiple programming languages. PROMETHEUS uses Neo4j for graph persistence, enabling scalable and struc- tured reasoning over large codebases. Integrated by the DeepSeek-V3 model, PROMETHEUS resolves 35.33% and 25.7% of issues on SWE-bench Lite and SWE-bench Multilingual, respectively, with an average API cost of $0.23 and $0.38 per issue. PROMETHEUS resolves 10 unique issues not addressed by prior work and is the first to demonstrate effectiveness across seven programming lan- guages. Moreover, it shows the ability to resolve real-world GitHub issues in the LangChain and OpenHands repositories. We have open-sourced PROMETHEUS
at: https://anonymous.4open.science/r/Prometheus-E8B1.

Primary Area: foundation or frontier models, including LLMs

Submission Number: 17642