SWE-CI: Evaluating Agent Capabilities in Maintaining Codebases via Continuous Integration
This addresses the need for better evaluation of AI agents in real-world software maintenance, though it is incremental as it builds on existing benchmarks like SWE-bench.
The authors tackled the problem of evaluating LLM-powered agents in maintaining codebases over time by introducing SWE-CI, a benchmark based on Continuous Integration that shifts focus from static bug fixing to dynamic maintainability, comprising 100 tasks with an average evolution history of 233 days and 71 commits.
Large language model (LLM)-powered agents have demonstrated strong capabilities in automating software engineering tasks such as static bug fixing, as evidenced by benchmarks like SWE-bench. However, in the real world, the development of mature software is typically predicated on complex requirement changes and long-term feature iterations -- a process that static, one-shot repair paradigms fail to capture. To bridge this gap, we propose \textbf{SWE-CI}, the first repository-level benchmark built upon the Continuous Integration loop, aiming to shift the evaluation paradigm for code generation from static, short-term \textit{functional correctness} toward dynamic, long-term \textit{maintainability}. The benchmark comprises 100 tasks, each corresponding on average to an evolution history spanning 233 days and 71 consecutive commits in a real-world code repository. SWE-CI requires agents to systematically resolve these tasks through dozens of rounds of analysis and coding iterations. SWE-CI provides valuable insights into how well agents can sustain code quality throughout long-term evolution.