Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

RepoLaunch: Automating Build&Test Pipeline of Code Repositories on ANY Language and ANY Platform

Kenan Li, Rongzhi Li, Linghao Zhang, Qirui Jin, Liao Zhu, Xiaosong Huang, Geng Zhang, Yikai Zhang, Shilin He, Chengxing Xie, Xin Zhang, Zijian Jin, Bowen Li, Chaoyun Zhang, Yu Kang, Yufan Huang, Elsie Nallipogu, Saravan Rajmohan, Qingwei Lin, Dongmei Zhang

TL;DR

RepoLaunch is introduced, the first agent capable of automatically resolving dependencies, compiling source code, and extracting test results for repositories across arbitrary programming languages and operating systems, enabling scalable benchmarking and training of coding agents and LLMs.

Abstract

Building software repositories typically requires significant manual effort. Recent advances in large language model (LLM) agents have accelerated automation in software engineering (SWE). We introduce RepoLaunch, the first agent capable of automatically resolving dependencies, compiling source code, and extracting test results for repositories across arbitrary programming languages and operating systems. To demonstrate its utility, we further propose a fully automated pipeline for SWE dataset creation, where task design is the only human intervention. RepoLaunch automates the remaining steps, enabling scalable benchmarking and training of coding agents and LLMs. Notably, several works on agentic benchmarking and training have recently adopted RepoLaunch for automated task generation.

RepoLaunch: Automating Build&Test Pipeline of Code Repositories on ANY Language and ANY Platform

TL;DR

RepoLaunch is introduced, the first agent capable of automatically resolving dependencies, compiling source code, and extracting test results for repositories across arbitrary programming languages and operating systems, enabling scalable benchmarking and training of coding agents and LLMs.

Abstract

Building software repositories typically requires significant manual effort. Recent advances in large language model (LLM) agents have accelerated automation in software engineering (SWE). We introduce RepoLaunch, the first agent capable of automatically resolving dependencies, compiling source code, and extracting test results for repositories across arbitrary programming languages and operating systems. To demonstrate its utility, we further propose a fully automated pipeline for SWE dataset creation, where task design is the only human intervention. RepoLaunch automates the remaining steps, enabling scalable benchmarking and training of coding agents and LLMs. Notably, several works on agentic benchmarking and training have recently adopted RepoLaunch for automated task generation.
Paper Structure (33 sections, 10 equations, 4 figures, 5 tables)

This paper contains 33 sections, 10 equations, 4 figures, 5 tables.

Table of Contents

  1. Introduction
  2. Related Works
  3. RepoLaunch
  4. Problem Formulation
  5. Preparation Stage
  6. Build Stage
  7. Release Stage
  8. RepoLaunch-Automated SWE Dataset Creation
  9. Experiment
  10. Evaluation of RepoLauch
  11. SWE Task Creation from GitHub
  12. Result
  13. Task Creation of SWE-bench-Live / MultiLang & Windows
  14. Evaluation Result of LLMs and Agents
  15. Comparison with repo2run & SWE-agent
  16. ...and 18 more sections

Figures (4)

  • Figure 1: RepoLaunch Pipeline.
  • Figure 2: RepoLauch-Automated SWE Dataset Creation. It should be noted that Cross-Platform Code Migration Bench is our future work, also an SWE benchmark with task instances created by RepoLaunch.
  • Figure 3: The proportion of different failure patterns in the Build and the Release stage .
  • Figure 4: The proportion of different failure patterns of coding agents on SWE-bench-Live.