A continuously updated repository systematically organizing 106 research papers on Large Language Model-based agents for Software Engineering, categorized from both Software Engineering and agent architecture perspectives.
One-Minute Overview#
This is a continuously updated repository for papers on "Large Language Model-Based Agents for Software Engineering: A Survey". It systematically organizes 106 research papers from both Software Engineering tasks and agent architecture perspectives. Ideal for researchers, software developers, and AI engineers wanting to understand the latest developments in this field.
Core Value: Provides a systematically organized resource of LLM-SE research papers, helping researchers quickly grasp the development trajectory of this field.
Quick Start#
Installation Difficulty: Low - This repository contains only paper lists and categorization information, requiring no installation. It can be accessed directly.
# Clone the repository to get the latest paper list
git clone https://github.com/FudanSELab/Agent4SE-Paper-List.git
Is this suitable for me?
- ✅ Researchers: Need to systematically understand the latest developments of LLM applications in software engineering
- ✅ Software Engineers: Want to understand the academic foundations of AI-assisted programming tools and methods
- ✅ AI Engineers: Exploring application cases of large language models in specialized domains
- ❌ Developers seeking ready-to-use LLM tools: This is a paper organization repository, not a code repository
Core Capabilities#
1. Systematic Paper Classification - Categorized by SE tasks and agent architecture#
- Categorizes across the entire software engineering lifecycle, from requirements engineering to code generation, testing, and debugging
- Organized from both agent architecture perspectives including agent frameworks, multi-agent systems, and human-agent collaboration Actual Value: Helps researchers quickly locate research areas of interest, avoiding the randomness of literature searches
2. Continuous Update Mechanism - Keeping pace with field development#
- Regularly adds new research findings (e.g., newly released research on programming-driven workflows in November 2024)
- Clearly shows research progress in chronological order Actual Value: Ensures researchers can access the latest developments in the field, avoiding information lag
3. Structured Literature Organization - Two-dimensional classification system#
- Classifies from intersecting perspectives of Software Engineering (SE) and Agent
- Includes specific sub-directions under each category (e.g., multi-agent systems under code generation, test-driven development, etc.) Actual Value: Provides multi-angle literature organization, promoting cross-domain research and discovering potential connections
Technology Stack & Integration#
Development Language: Not explicitly specified (as a paper organization repository, it focuses on content rather than implementation technology) Main Dependencies: GitHub platform, Markdown format Integration Method: Knowledge base/references, with direct links to papers
Maintenance Status#
- Development Activity: High - As a survey project in a rapidly evolving research field, it maintains regular updates
- Recent Updates: November 2024, with continuous addition of new research findings
- Community Response: Good - Has gained academic attention (GitHub star system)
Documentation & Learning Resources#
- Documentation Quality: Comprehensive - Contains complete paper categorization and links, but no detailed tutorials
- Official Documentation: GitHub Repository
- Sample Code: Not applicable (paper organization project)