Multi-SWE-bench

Scaling data for SWE-agents (NeurIPS 2025 D&B Spotlight). A toolkit for automatically generating large-scale training datasets for software engineering agents.

SWE-bench is a benchmark for evaluating language models on real-world GitHub issue resolution, featuring genuine problems from popular Python repositories, now a core standard for measuring AI coding agent capabilities.

Augment SWE-bench Agent is the number one open-source SWE-bench Verified implementation, demonstrating how to build high-performance software engineering agents to automatically resolve GitHub issues.

Trae Agent is an LLM-based agent for general purpose software engineering tasks.