Awesome GitHub RepositoriesCode Refinement

OpenHands is an autonomous AI software engineer and coding assistant designed to execute software engineering tasks by interacting directly with codebases and development environments. It functions as a platform for running AI agents that can write code and manage files to automate complex development workflows. The system distinguishes itself through a container-based execution environment that isolates agent actions within a sandboxed Linux environment. It employs an autonomous agent loop of observation, planning, and action, supported by a standardized communication protocol that allows it

OpenUI is an AI design sandbox and natural language prototyping tool used to generate and render live user interface components from text descriptions. It functions as an LLM UI generator that translates natural language into executable HTML and CSS code. The system provides a pipeline for iterative refinement, allowing users to update existing interfaces by feeding previous code versions and new instructions back into the model. It also acts as a frontend framework converter, transforming HTML markup into different library formats to maintain styling consistency across various web frameworks

bolt.new is an AI-powered full-stack web builder and browser-based IDE that generates, edits, and deploys web applications using natural language prompts. It functions as an AI-driven application orchestrator, managing the entire development lifecycle through a chat interface. The platform distinguishes itself by integrating a WebAssembly-based runtime and virtual terminal emulation directly in the browser. This allows an AI agent to execute tool calls, manage a virtual filesystem, install packages, and run servers without requiring a local development environment. The system covers a compre

KeepChatGPT is a browser extension designed to enhance the ChatGPT web experience by acting as a session manager, UI optimizer, and privacy guard. It focuses on maintaining active connections to prevent session timeouts and improving the overall interface for better readability and organization. The project distinguishes itself through privacy and security features that block tracking telemetry and use regular expressions to mask sensitive data before it is sent. It also includes tools to mitigate conversation auditing and bypass bot verification challenges to reduce the risk of account restr

llm-universe is a structured learning resource and technical guide focused on the development of large language model applications. It serves as a curriculum for mastering model orchestration, the creation of autonomous conversational agents, and the implementation of retrieval-augmented generation systems. The project provides detailed instructions on connecting model APIs with memory and tools to create execution chains. It specifically covers the construction of retrieval pipelines, including the process of cleaning raw documents, generating embeddings, and integrating vector databases to

This project is an AI software engineering tool and framework for building autonomous coding agents. It provides a system for automating program synthesis and bug fixing by integrating large language models with codebase analysis and iterative refinement loops. The framework features an agentic development server that exposes task execution interfaces to remote agents through a structured protocol. This allows for the remote execution of development tasks and the embedding of autonomous program synthesis capabilities into external software projects. The toolset covers AI-driven project scaff

This project is an automated prompt engineering and optimization tool designed to iteratively create, test, and refine prompts using a language model to improve output quality. It functions as a framework for generating candidate prompts and ranking their performance through correctness matching and ELO-based ratings. The system includes capabilities for model distillation, generating high-quality example pairs from frontier models to create training data for smaller models. It also provides tools to condense prompts for smaller models and transform instruction-tuned prompts into completion-b

LLM4Decompile 是一个用于二进制到源代码翻译的工具集和框架。它利用大语言模型将机器代码转换为可读的源代码，并恢复编译后可执行文件的原始逻辑。 该项目包含一个专门的流水线，通过将源代码转换为汇编对来生成合成训练数据集。它提供了一个微调框架，用于在这些二进制到源代码数据集上优化深度学习模型，从而提高代码恢复的准确性。 该系统还具有细化反编译伪代码的功能。此过程侧重于恢复二进制文件的结构骨架和变量名，以提高反汇编逻辑的可读性。

ChatGPT-AutoExpert is an AI prompt engineering framework and persona management system designed to improve the technical accuracy and nuance of large language model outputs. It provides a collection of curated system prompts and custom instructions to refine user queries and remove conversational filler. The system employs a persona framework to assign specialized expert roles based on the request context. It utilizes a command shortcut system that maps short text sequences to complex instructional sets, enabling the rapid execution of repetitive tasks. For software development, the project

OpenEvolve 是一个进化算法框架，使用大语言模型来自主发现和优化编程算法。它作为一个算法发现引擎和代码搜索工具，进化候选程序群体以找到高效的实现和特定于硬件的加速。 该系统将代码和系统指令都视为可进化的实体，利用自动提示优化器迭代地改进模型性能。它通过基于生态位的群体管理来保持搜索稳定性以维持多样性，并采用闭环反馈机制，将运行时错误和日志注入回生成过程中以进行自主错误修复。 该框架还包括一个代理编排器，使用加权逻辑和回退策略聚合来自多个模型 API 的响应。为了支持科学计算，它通过管理所有随机组件的一致随机种子来实现确定性执行。 该项目提供了一个用于可视化进化进度和实时性能指标的交互式仪表板。

vibesdk 是一个代理式软件开发平台和框架，旨在协调能够从自然语言编写、调试和优化全栈应用程序的自主代理。它作为一个云原生应用程序编排器和 LLM 驱动的代码生成框架，通过迭代对话和多阶段代理行为将提示词转换为功能代码。 该项目的特色在于提供了一套完整的工具链，用于构建 AI 开发平台。这包括集成各种模型提供商、构建自定义 LLM 工具包以及通过无服务器部署工具链和程序化 TypeScript SDK 管理 AI 生成应用的整个生命周期。 该平台涵盖了广泛的功能，包括用于隔离执行和实时预览的 AI 沙箱编排、用于版本跟踪的 Git 后端虚拟文件系统，以及到无服务器工作者平台的自动化云部署。它还集成了用于数据库模式管理、分层密钥加密以及通过 WebSocket 进行实时状态同步的系统。 用户可以通过命令行界面或使用提供的 SDK 以编程方式管理项目工作流。

AlphaCodium 是一个 LLM 代码生成框架和自动化编程基准，旨在通过迭代生成和测试来解决编程问题。它作为一个迭代代码优化系统，通过将输出与预期结果进行比较并重新提示模型，来提高生成代码的精度。 该项目实现了一个流程工程管道，使用结构化的提示阶段序列，通过生成、评估和纠正的循环来优化代码。这种方法允许系统处理编程数据集，并根据测试用例衡量生成解决方案的准确性。 该框架涵盖了自动化代码生成的广泛功能，包括测试驱动的 AI 开发和编程数据集评估。它通过包含规划、起草和审查的多阶段合成管道来管理这些任务。

Darwin Gödel Machine: Open-Ended Evolution of Self-Improving Agents

With Self-Refine, LLMs can generate feedback on their work, use it to improve the output, and repeat this process.

AgentCoder is a novel multiagent-code generation framework that leverages the power of large language models (LLMs) to enhance the effectiveness of code generation. The framework consists of three specialized agents: the programmer agent, the test designer agent, and the test executor agent.…

Code and data for paper "Self-Evolving Multi-Agent Collaboration Networks for Software Development".

please visit: https://zenodo.org/records/11666403