In the field of Automated Program Repair (APR), Large Language Models (LLMs) have shown significant improvements, yet two major limitations persist.
First, raw execution traces are often too large and repetitive to serve as effective model context. Second, repeated patch sampling may yield different implementations without providing distinct root-cause hypotheses or repair strategies.
To address these issues, we present CT-Repair, an agentic APR framework that represents static and dynamic evidence as queryable Code Property Graphs (CPG) and Temporal Execution Graphs (TEG). CT-Repair employs a three-stage filtering pipeline to construct compact TEGs. Three finite-state-machine-guided agents analyze each bug from static, dynamic, and hybrid perspectives, independently producing evidence-grounded repair strategies. A strategy-guided generation procedure instantiates these strategies as candidate patches and utilizes validation feedback to refine the most promising strategy.
We evaluated CT-Repair on 854 Java bugs from Defects4J v3.0. In the mixed-model configuration, CT-Repair correctly repairs 489 bugs. Under a controlled GPT-5.4-mini configuration, it repairs 388 bugs, 19 and 30 more than ReinFix and RepairAgent, respectively. The union of the three evidence perspectives repairs 99 more bugs than the strongest individual perspective. The filtering pipeline also compacts runtime evidence, with execution filtering narrowing the candidate method scope by 94.85% on average and behavior filtering further reducing retained runtime records by 55.97%. These results show that structured runtime evidence and multi-perspective reasoning can enhance repair effectiveness without solely relying on a larger patch-generation budget.
Blogger's Review: The innovation of CT-Repair lies in its multi-perspective analysis and evidence-driven repair strategies, significantly increasing the success rate of program repairs, especially when facing complex bugs. By effectively leveraging both static and dynamic information, CT-Repair showcases the potential for future advancements in the APR domain, warranting deeper exploration and application by developers.