Autonomous program improvement typically involves automatically producing bug fixes and feature additions. Such program improvement can be accomplished by a combination of large language model (LLM) and program analysis capabilities, in the form of an LLM agent. Since program repair or program improvement typically requires a specification of intended behavior - specification inference can be useful for producing high quality program patches. In this work, we examine efficient and low-cost workflows for iterative specification inference within an LLM agent. Given a GitHub issue to be resolved in a software project, our goal is to conduct iterative code search accompanied by specification inference - thereby inferring intent from both the project structure and behavior. The intent thus captured is examined by a reviewer agent with the goal of vetting the patches as well as providing a measure of confidence in the vetted patches. Our approach SpecRover is built on the open-source LLM agent AutoCodeRover. In an evaluation on the full SWE-Bench consisting of 2294 GitHub issues, it shows more than 50% improvement in efficacy over AutoCodeRover. Compared to the open-source agents available, our work shows modest cost ($0.65 per issue) in resolving an average GitHub issue in SWE-Bench lite. The production of explanation by SpecRover allows for a better “signal” to be given to the developer, on when the suggested patches can be accepted with confidence. SpecRover also seeks to demonstrate the continued importance of specification inference in automated program repair, even as program repair technologies enter the LLM era.

Automated program repair (APR) allows for au-tonomous software protection and improvement. Many proposed repair techniques rely on available test suites, since tests are available in real-world settings. Tests are incomplete specifications, however. As a result, repairs generated based on tests may suffer from the test overfitting problem. The patches produced by APR techniques may pass the given tests and thus be plausible, and yet be an incorrect patch. This hints towards more extensive test suites to guide program repair. Generating additional tests to improve the test suite quality is generally difficult because the oracle or expected observable behavior of the generated tests is unknown. In our work, we first construct additional oracles by instrumenting buggy programs from the Defects4j benchmark with the knowledge obtained from the available bug reports. Then, we formulate a coevolution approach that generates tests and repairs in a unified workflow. The complete workflow is implemented as an extension of the well-known Java testing framework Evosuite. This includes re-purposing the search in Evosuite to search for repairs (instead of searching for tests) and enables an easy adoption for developers who are already familiar with Evosuite for test suite generation. The evaluation of our tool EvoREPAIR shows that coevolution positively impacts the quality of patches and tests. In the future, we hope that such coevolution can inspire new repair tools and techniques.

Researchers have made significant progress in automating the software development process in the past decades. Automated techniques for issue summarization, bug reproduction, fault localization, and program repair have been built to ease the workload of developers. Recent progress in Large Language Models (LLMs) has significantly impacted the development process, where developers can use LLM-based programming assistants to achieve automated coding. Nevertheless, software engineering involves the process of program improvement apart from coding, specifically to enable software maintenance (e.g. program repair to fix bugs) and software evolution (e.g. feature additions). In this paper, we propose an automated approach for solving Github issues to autonomously achieve program improvement. In our approach called AutoCodeRover, LLMs are combined with sophisticated code search capabilities, ultimately leading to a program modification or patch. In contrast to recent LLM agent approaches from AI researchers and practitioners, our outlook is more software engineering oriented. We work on a program representation (abstract syntax tree) as opposed to viewing a software project as a mere collection of files. Our code search exploits the program structure in the form of classes/methods to enhance LLM’s understanding of the issue’s root cause, and effectively retrieve a context via iterative search. The use of spectrum-based fault localization using tests, further sharpens the context, as long as a test-suite is available. Experiments on the recently proposed SWE-bench-lite (300 real-life Github issues) show increased efficacy in solving Github issues (19% on SWE-bench-lite), which is higher than the efficacy of the recently reported Swe-agent. Interestingly, our approach resolved 57 GitHub issues in about 4 minutes each (pass@1), whereas developers spent more than 2.68 days on average. In addition, AutoCodeRover achieved this efficacy with significantly lower cost (on average, $0.43 USD), compared to other baselines. We posit that our workflow enables autonomous software engineering, where, in future, auto-generated code from LLMs can be autonomously improved.

While large language models (LLMs) have shown significant advancements in code generation, their susceptibility to producing incorrect code poses a significant challenge to the adoption of LLM-generated programs. This issue largely stems from the reliance on natural language descriptions as informal oracles in code generation. Current strategies to mitigate this involve selecting the best program from multiple LLM-generated alternatives, judged by criteria like the consistency of their execution results on an LLM-generated test suite. However, this approach has crucial limitations: (1) LLMs often generate redundant tests or tests that cannot distinguish between correct and incorrect solutions, (2) the used consistency criteria, such as the majority vote, fail to foster developer trust due to the absence of transparent rationale behind the made choices. In this work, we propose a new perspective on increasing the quality of LLM-generated code via program selection using the LLM as a test oracle. Our method is based on our experimentally confirmed observation that LLMs serve more effectively as oracles when tasked with selecting the correct output from multiple choices. Leveraging this insight, we first generate distinguishing inputs that capture semantic discrepancies of programs sampled from an LLM, and record outputs produced by the programs on these inputs. An LLM then selects the most likely to be correct output from these, guided by the natural language problem description. We implemented this idea in a tool LLMCodeChoice and evaluated its accuracy in generating and selecting standalone programs. Our experiments demonstrated its effectiveness in improving pass@1 by 3.6-7% on HumanEval and MBPP benchmarks compared to the state-of-art CodeT. Most interestingly, the selected input-output specifications helped us to uncover incompleteness and ambiguities in task descriptions and also identify incorrect ground-truth implementations in the benchmarks.

Side-channel vulnerability detection has gained prominence recently due to Spectre and Meltdown attacks. Techniques for side-channel detection range from fuzz testing to program analysis and program composition. Existing side-channel mitigation techniques repair the vulnerability at the IR/binary level or use runtime monitoring solutions. In both cases, the source code itself is not modified, can evolve while keeping the vulnerability, and the developer would get no feedback on how to develop secure applications in the first place. Thus, these solutions do not help the developer understand the side-channel risks in her code and do not provide guidance to avoid code patterns with side-channel risks. In this article, we present Pendulum , the first approach for automatically locating and repairing side-channel vulnerabilities in the source code, specifically for timing side channels. Our approach uses a quantitative estimation of found vulnerabilities to guide the fix localization, which goes hand-in-hand with a pattern-guided repair. Our evaluation shows that Pendulum can repair a large number of side-channel vulnerabilities in real-world applications. Overall, our approach integrates vulnerability detection, quantization, localization, and repair into one unified process. This also enhances the possibility of our side-channel mitigation approach being adopted into programmingenvironments.