SWE-agent

SWE-agent

by Princeton University NLP Group
SWE-agent is an open-source AI programmer and software engineer system developed by Princeton University's NLP group, capable of automatically solving issues in GitHub repositories using large language models like GPT-4.

What is SWE-agent?

SWE-agent is an open-source AI programmer and software engineer system developed by researchers at Princeton University's NLP group. It leverages the capabilities of large language models (e.g., GPT-4) to automatically resolve issues in GitHub repositories. SWE-agent interacts with codebases through an Agent-Computer Interface (ACI), enabling it to perform tasks such as browsing, editing, testing, and executing code. The system has demonstrated accuracy similar to the closed-source AI programmer Devin on the SWE-bench test set, solving issues in an average of 93 seconds and achieving state-of-the-art (SOTA) performance.

Official Website of SWE-agent

Main Features of SWE-agent

  • Pull Request Issue Resolution: SWE-agent can understand issues in GitHub repositories and attempt to fix them by creating pull requests.
  • Code Editing and Fixing: SWE-agent can browse and edit files in the codebase, automatically fixing errors and vulnerabilities in the code.
  • Automatic Syntax Checking: During code editing, SWE-agent can run linters to ensure the code adheres to syntax standards.
  • File Viewer: Provides a specially built file viewer that displays 100 lines of code per round, supporting scrolling and search functions for more efficient code viewing and editing.
  • Full Directory String Search: SWE-agent has a full directory string search function, listing all files and code snippets that match the search criteria.
  • Command and Feedback: Through the Agent-Computer Interface (ACI), SWE-agent can receive and execute commands given in natural language and provide corresponding feedback.
  • Test Writing and Execution: SWE-agent can write and execute test code to verify the effectiveness of fixes.

Workflow of SWE-agent

  1. Understanding the Issue: SWE-agent uses natural language processing (NLP) techniques to understand the issue descriptions in GitHub repositories.
  2. Agent-Computer Interface (ACI): SWE-agent uses ACI to interact with the codebase, enabling it to browse, search, view, edit, and execute code.
  3. Code Analysis and Fixing: After understanding the issue, SWE-agent analyzes the relevant code, locates potential errors or vulnerabilities, and generates fixes.
  4. Automated Testing: SWE-agent writes and executes test cases to verify the effectiveness of the fixes.
  5. Performance Feedback: Each step performed by SWE-agent generates feedback, which is used to evaluate the effectiveness of its work.
  6. Iteration and Optimization: The research team collects feedback and performance data to continuously improve SWE-agent's problem-solving ability and code-fixing accuracy.

Features & Capabilities

What You Can Do
Pull Request Issue Resolution Code Editing And Fixing Automatic Syntax Checking File Viewing Full Directory String Search Command Execution Test Writing And Execution
Categories
AI Programmer Software Engineering Open Source GitHub Code Fixing Automated Testing Natural Language Processing Large Language Models Developer Tools Code Analysis
Example Uses
  • Fixing issues in GitHub repositories
  • Automated code editing and fixing
  • Writing and executing test cases

Getting Started

Pricing
free

Screenshots & Images

Primary Screenshot
Additional Images
Try It Now View Demo

Stats

0 Views
0 Likes

Similar Tools

SadTalker by Xi'an Jiaotong University, Tencent AI Lab, Ant Group
0
CrayoAI
0
SciSpace
0
2025© BestAI