[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"tool-openai--openai-agents-python":3,"similar-openai--openai-agents-python":174},{"id":4,"github_repo":5,"name":6,"description_en":7,"description_zh":8,"ai_summary_zh":8,"readme_en":9,"readme_zh":10,"quickstart_zh":11,"use_case_zh":12,"hero_image_url":13,"owner_login":14,"owner_name":15,"owner_avatar_url":16,"owner_bio":17,"owner_company":18,"owner_location":18,"owner_email":18,"owner_twitter":18,"owner_website":19,"owner_url":20,"languages":21,"stars":38,"forks":39,"last_commit_at":40,"license":41,"difficulty_score":42,"env_os":43,"env_gpu":44,"env_ram":44,"env_deps":45,"category_tags":58,"github_topics":63,"view_count":42,"oss_zip_url":18,"oss_zip_packed_at":18,"status":69,"created_at":70,"updated_at":71,"faqs":72,"releases":73},5984,"openai\u002Fopenai-agents-python","openai-agents-python","A lightweight, powerful framework for multi-agent workflows","openai-agents-python 是 OpenAI 推出的一款轻量级却功能强大的 Python 框架，专为构建多智能体（Multi-Agent）工作流而设计。它旨在解决开发者在协调多个 AI 智能体协作时的复杂性难题，让构建能够自主规划、调用工具并相互交接任务的智能系统变得简单高效。\n\n这款工具非常适合希望深入开发复杂 AI 应用的软件工程师、研究人员及技术团队使用。其核心亮点在于“提供商无关”的设计，不仅完美支持 OpenAI 自家的 API，还能无缝对接超过 100 种其他大语言模型。框架内置了丰富的核心概念：支持将智能体作为工具互相调用（Handoffs），提供可配置的安全护栏（Guardrails）以确保输入输出合规，原生集成“人机协同”机制以便在关键节点引入人工干预，并具备自动会话管理和全链路追踪功能，帮助开发者轻松调试和优化工作流。此外，它还特别支持基于 `gpt-realtime` 模型的实时语音智能体开发。无论是快速原型验证还是生产级部署，openai-agents-python 都能以简洁的代码结构，助你灵活搭建高度自治的 AI 协作网络。","# OpenAI Agents SDK [![PyPI](https:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fv\u002Fopenai-agents?label=pypi%20package)](https:\u002F\u002Fpypi.org\u002Fproject\u002Fopenai-agents\u002F)\n\nThe OpenAI Agents SDK is a lightweight yet powerful framework for building multi-agent workflows. It is provider-agnostic, supporting the OpenAI Responses and Chat Completions APIs, as well as 100+ other LLMs.\n\n\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fopenai_openai-agents-python_readme_8acfe7a1082a.png\" alt=\"Image of the Agents Tracing UI\" style=\"max-height: 803px;\">\n\n> [!NOTE]\n> Looking for the JavaScript\u002FTypeScript version? Check out [Agents SDK JS\u002FTS](https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-js).\n\n### Core concepts:\n\n1. [**Agents**](https:\u002F\u002Fopenai.github.io\u002Fopenai-agents-python\u002Fagents): LLMs configured with instructions, tools, guardrails, and handoffs\n1. **[Agents as tools](https:\u002F\u002Fopenai.github.io\u002Fopenai-agents-python\u002Ftools\u002F#agents-as-tools) \u002F [Handoffs](https:\u002F\u002Fopenai.github.io\u002Fopenai-agents-python\u002Fhandoffs\u002F)**: Delegating to other agents for specific tasks\n1. [**Tools**](https:\u002F\u002Fopenai.github.io\u002Fopenai-agents-python\u002Ftools\u002F): Various Tools let agents take actions (functions, MCP, hosted tools)\n1. [**Guardrails**](https:\u002F\u002Fopenai.github.io\u002Fopenai-agents-python\u002Fguardrails\u002F): Configurable safety checks for input and output validation\n1. [**Human in the loop**](https:\u002F\u002Fopenai.github.io\u002Fopenai-agents-python\u002Fhuman_in_the_loop\u002F): Built-in mechanisms for involving humans across agent runs\n1. [**Sessions**](https:\u002F\u002Fopenai.github.io\u002Fopenai-agents-python\u002Fsessions\u002F): Automatic conversation history management across agent runs\n1. [**Tracing**](https:\u002F\u002Fopenai.github.io\u002Fopenai-agents-python\u002Ftracing\u002F): Built-in tracking of agent runs, allowing you to view, debug and optimize your workflows\n1. [**Realtime Agents**](https:\u002F\u002Fopenai.github.io\u002Fopenai-agents-python\u002Frealtime\u002Fquickstart\u002F): Build powerful voice agents with `gpt-realtime-1.5` and full agent features\n\nExplore the [examples](https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Ftree\u002Fmain\u002Fexamples) directory to see the SDK in action, and read our [documentation](https:\u002F\u002Fopenai.github.io\u002Fopenai-agents-python\u002F) for more details.\n\n## Get started\n\nTo get started, set up your Python environment (Python 3.10 or newer required), and then install OpenAI Agents SDK package.\n\n### venv\n\n```bash\npython -m venv .venv\nsource .venv\u002Fbin\u002Factivate  # On Windows: .venv\\Scripts\\activate\npip install openai-agents\n```\n\nFor voice support, install with the optional `voice` group: `pip install 'openai-agents[voice]'`. For Redis session support, install with the optional `redis` group: `pip install 'openai-agents[redis]'`.\n\n### uv\n\nIf you're familiar with [uv](https:\u002F\u002Fdocs.astral.sh\u002Fuv\u002F), installing the package would be even easier:\n\n```bash\nuv init\nuv add openai-agents\n```\n\nFor voice support, install with the optional `voice` group: `uv add 'openai-agents[voice]'`. For Redis session support, install with the optional `redis` group: `uv add 'openai-agents[redis]'`.\n\n## Run your first agent\n\n```python\nfrom agents import Agent, Runner\n\nagent = Agent(name=\"Assistant\", instructions=\"You are a helpful assistant\")\n\nresult = Runner.run_sync(agent, \"Write a haiku about recursion in programming.\")\nprint(result.final_output)\n\n# Code within the code,\n# Functions calling themselves,\n# Infinite loop's dance.\n```\n\n(_If running this, ensure you set the `OPENAI_API_KEY` environment variable_)\n\n(_For Jupyter notebook users, see [hello_world_jupyter.ipynb](https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fblob\u002Fmain\u002Fexamples\u002Fbasic\u002Fhello_world_jupyter.ipynb)_)\n\nExplore the [examples](https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Ftree\u002Fmain\u002Fexamples) directory to see the SDK in action, and read our [documentation](https:\u002F\u002Fopenai.github.io\u002Fopenai-agents-python\u002F) for more details.\n\n## Acknowledgements\n\nWe'd like to acknowledge the excellent work of the open-source community, especially:\n\n- [Pydantic](https:\u002F\u002Fdocs.pydantic.dev\u002Flatest\u002F)\n- [Requests](https:\u002F\u002Fgithub.com\u002Fpsf\u002Frequests)\n- [MCP Python SDK](https:\u002F\u002Fgithub.com\u002Fmodelcontextprotocol\u002Fpython-sdk)\n- [Griffe](https:\u002F\u002Fgithub.com\u002Fmkdocstrings\u002Fgriffe)\n\nThis library has these optional dependencies:\n\n- [websockets](https:\u002F\u002Fgithub.com\u002Fpython-websockets\u002Fwebsockets)\n- [SQLAlchemy](https:\u002F\u002Fgithub.com\u002Fsqlalchemy\u002Fsqlalchemy)\n- [any-llm](https:\u002F\u002Fgithub.com\u002Fmozilla-ai\u002Fany-llm) and [LiteLLM](https:\u002F\u002Fgithub.com\u002FBerriAI\u002Flitellm)\n\nWe also rely on the following tools to manage the project:\n\n- [uv](https:\u002F\u002Fgithub.com\u002Fastral-sh\u002Fuv) and [ruff](https:\u002F\u002Fgithub.com\u002Fastral-sh\u002Fruff)\n- [mypy](https:\u002F\u002Fgithub.com\u002Fpython\u002Fmypy) and [Pyright](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002Fpyright)\n- [pytest](https:\u002F\u002Fgithub.com\u002Fpytest-dev\u002Fpytest) and [Coverage.py](https:\u002F\u002Fgithub.com\u002Fcoveragepy\u002Fcoveragepy)\n- [MkDocs](https:\u002F\u002Fgithub.com\u002Fsquidfunk\u002Fmkdocs-material)\n\nWe're committed to continuing to build the Agents SDK as an open source framework so others in the community can expand on our approach.","# OpenAI Agents SDK [![PyPI](https:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fv\u002Fopenai-agents?label=pypi%20package)](https:\u002F\u002Fpypi.org\u002Fproject\u002Fopenai-agents\u002F)\n\nOpenAI Agents SDK 是一个轻量级但功能强大的框架，用于构建多智能体工作流。它与提供商无关，支持 OpenAI 的 Responses 和 Chat Completions API，以及 100 多种其他大语言模型。\n\n\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fopenai_openai-agents-python_readme_8acfe7a1082a.png\" alt=\"Image of the Agents Tracing UI\" style=\"max-height: 803px;\">\n\n> [!NOTE]\n> 您在寻找 JavaScript\u002FTypeScript 版本吗？请查看 [Agents SDK JS\u002FTS](https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-js)。\n\n### 核心概念：\n\n1. [**智能体**](https:\u002F\u002Fopenai.github.io\u002Fopenai-agents-python\u002Fagents)：配置了指令、工具、安全约束和交接机制的大语言模型\n1. **[作为工具的智能体](https:\u002F\u002Fopenai.github.io\u002Fopenai-agents-python\u002Ftools\u002F#agents-as-tools) \u002F [交接](https:\u002F\u002Fopenai.github.io\u002Fopenai-agents-python\u002Fhandoffs\u002F)**：将特定任务委托给其他智能体\n1. [**工具**](https:\u002F\u002Fopenai.github.io\u002Fopenai-agents-python\u002Ftools\u002F)：各种工具使智能体能够执行操作（函数、MCP、托管工具）\n1. [**安全约束**](https:\u002F\u002Fopenai.github.io\u002Fopenai-agents-python\u002Fguardrails\u002F)：可配置的安全检查，用于输入和输出验证\n1. [**人类参与环节**](https:\u002F\u002Fopenai.github.io\u002Fopenai-agents-python\u002Fhuman_in_the_loop\u002F)：内置机制，可在智能体运行过程中引入人工干预\n1. [**会话**](https:\u002F\u002Fopenai.github.io\u002Fopenai-agents-python\u002Fsessions\u002F)：自动管理跨智能体运行的对话历史\n1. [**追踪**](https:\u002F\u002Fopenai.github.io\u002Fopenai-agents-python\u002Ftracing\u002F)：内置的智能体运行跟踪功能，允许您查看、调试和优化工作流\n1. [**实时智能体**](https:\u002F\u002Fopenai.github.io\u002Fopenai-agents-python\u002Frealtime\u002Fquickstart\u002F)：使用 `gpt-realtime-1.5` 和完整智能体功能构建强大的语音智能体\n\n浏览 [examples](https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Ftree\u002Fmain\u002Fexamples) 目录，了解 SDK 的实际应用；更多详细信息请参阅我们的 [文档](https:\u002F\u002Fopenai.github.io\u002Fopenai-agents-python\u002F)。\n\n## 开始使用\n\n要开始使用，请设置您的 Python 环境（需 Python 3.10 或更高版本），然后安装 OpenAI Agents SDK 包。\n\n### venv\n\n```bash\npython -m venv .venv\nsource .venv\u002Fbin\u002Factivate  # 在 Windows 上：.venv\\Scripts\\activate\npip install openai-agents\n```\n\n如需语音支持，请使用可选的 `voice` 组进行安装：`pip install 'openai-agents[voice]'`。如需 Redis 会话支持，请使用可选的 `redis` 组进行安装：`pip install 'openai-agents[redis]'`。\n\n### uv\n\n如果您熟悉 [uv](https:\u002F\u002Fdocs.astral.sh\u002Fuv\u002F)，安装该包将更加简单：\n\n```bash\nuv init\nuv add openai-agents\n```\n\n如需语音支持，请使用可选的 `voice` 组进行安装：`uv add 'openai-agents[voice]'`。如需 Redis 会话支持，请使用可选的 `redis` 组进行安装：`uv add 'openai-agents[redis]'`。\n\n## 运行您的第一个智能体\n\n```python\nfrom agents import Agent, Runner\n\nagent = Agent(name=\"Assistant\", instructions=\"您是一位乐于助人的助手\")\n\nresult = Runner.run_sync(agent, \"写一首关于编程中递归的俳句。\")\nprint(result.final_output)\n\n# 代码之中藏代码，\n# 函数自调自身时，\n# 无限循环舞翩跹。\n```\n\n（_运行此代码时，请确保已设置 `OPENAI_API_KEY` 环境变量_）\n\n（_对于 Jupyter Notebook 用户，请参阅 [hello_world_jupyter.ipynb](https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fblob\u002Fmain\u002Fexamples\u002Fbasic\u002Fhello_world_jupyter.ipynb)_）\n\n浏览 [examples](https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Ftree\u002Fmain\u002Fexamples) 目录，了解 SDK 的实际应用；更多详细信息请参阅我们的 [文档](https:\u002F\u002Fopenai.github.io\u002Fopenai-agents-python\u002F)。\n\n## 致谢\n\n我们谨向开源社区的杰出工作表示感谢，特别是：\n\n- [Pydantic](https:\u002F\u002Fdocs.pydantic.dev\u002Flatest\u002F)\n- [Requests](https:\u002F\u002Fgithub.com\u002Fpsf\u002Frequests)\n- [MCP Python SDK](https:\u002F\u002Fgithub.com\u002Fmodelcontextprotocol\u002Fpython-sdk)\n- [Griffe](https:\u002F\u002Fgithub.com\u002Fmkdocstrings\u002Fgriffe)\n\n本库具有以下可选依赖项：\n\n- [websockets](https:\u002F\u002Fgithub.com\u002Fpython-websockets\u002Fwebsockets)\n- [SQLAlchemy](https:\u002F\u002Fgithub.com\u002Fsqlalchemy\u002Fsqlalchemy)\n- [any-llm](https:\u002F\u002Fgithub.com\u002Fmozilla-ai\u002Fany-llm) 和 [LiteLLM](https:\u002F\u002Fgithub.com\u002FBerriAI\u002Flitellm)\n\n我们还依靠以下工具来管理项目：\n\n- [uv](https:\u002F\u002Fgithub.com\u002Fastral-sh\u002Fuv) 和 [ruff](https:\u002F\u002Fgithub.com\u002Fastral-sh\u002Fruff)\n- [mypy](https:\u002F\u002Fgithub.com\u002Fpython\u002Fmypy) 和 [Pyright](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002Fpyright)\n- [pytest](https:\u002F\u002Fgithub.com\u002Fpytest-dev\u002Fpytest) 和 [Coverage.py](https:\u002F\u002Fgithub.com\u002Fcoveragepy\u002Fcoveragepy)\n- [MkDocs](https:\u002F\u002Fgithub.com\u002Fsquidfunk\u002Fmkdocs-material)\n\n我们致力于继续将 Agents SDK 打造成一个开源框架，以便社区中的其他人能够在此基础上进一步发展。","# OpenAI Agents SDK 快速上手指南\n\nOpenAI Agents SDK 是一个轻量级但功能强大的框架，用于构建多智能体（Multi-Agent）工作流。它不依赖特定提供商，支持 OpenAI Responses 和 Chat Completions API，以及 100+ 其他大语言模型。\n\n## 环境准备\n\n- **操作系统**：Linux、macOS 或 Windows\n- **Python 版本**：Python 3.10 或更高版本\n- **API 密钥**：需准备 `OPENAI_API_KEY` 环境变量\n\n## 安装步骤\n\n你可以选择使用标准的 `venv` 或现代化的 `uv` 工具进行安装。\n\n### 方式一：使用 venv（标准方案）\n\n```bash\npython -m venv .venv\nsource .venv\u002Fbin\u002Factivate  # Windows 用户请使用：.venv\\Scripts\\activate\n\n# 基础安装\npip install openai-agents\n\n# 可选：安装语音支持\npip install 'openai-agents[voice]'\n\n# 可选：安装 Redis 会话支持\npip install 'openai-agents[redis]'\n```\n\n> **国内加速建议**：如果下载速度较慢，可使用清华或阿里镜像源：\n> `pip install -i https:\u002F\u002Fpypi.tuna.tsinghua.edu.cn\u002Fsimple openai-agents`\n\n### 方式二：使用 uv（推荐方案）\n\n如果你已安装 [uv](https:\u002F\u002Fdocs.astral.sh\u002Fuv\u002F)，安装过程将更加简洁高效：\n\n```bash\nuv init\nuv add openai-agents\n\n# 可选：安装语音支持\nuv add 'openai-agents[voice]'\n\n# 可选：安装 Redis 会话支持\nuv add 'openai-agents[redis]'\n```\n\n## 基本使用\n\n以下是最简单的单智能体运行示例。确保在运行前已设置环境变量 `OPENAI_API_KEY`。\n\n```python\nfrom agents import Agent, Runner\n\nagent = Agent(name=\"Assistant\", instructions=\"You are a helpful assistant\")\n\nresult = Runner.run_sync(agent, \"Write a haiku about recursion in programming.\")\nprint(result.final_output)\n\n# 输出示例：\n# Code within the code,\n# Functions calling themselves,\n# Infinite loop's dance.\n```\n\n**Jupyter 用户注意**：请参考官方仓库中的 [hello_world_jupyter.ipynb](https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fblob\u002Fmain\u002Fexamples\u002Fbasic\u002Fhello_world_jupyter.ipynb) 示例。\n\n更多高级功能（如智能体移交、工具调用、护栏机制等）请查阅 [官方文档](https:\u002F\u002Fopenai.github.io\u002Fopenai-agents-python\u002F) 或浏览 [examples](https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Ftree\u002Fmain\u002Fexamples) 目录。","某电商技术团队正在构建一个能自动处理复杂售后请求的智能客服系统，需协调退款审核、物流查询与情绪安抚等多个环节。\n\n### 没有 openai-agents-python 时\n- **流程割裂难协同**：开发者需手动编写大量胶水代码串联不同功能的 LLM，导致“退款专员”与“物流助手”之间无法自动传递上下文，任务经常中断。\n- **状态管理混乱**：多轮对话的历史记录和中间状态需自行设计数据库 schema 存储，容易出现数据不一致或会话丢失，调试时难以复现问题现场。\n- **安全风控缺失**：缺乏统一的输入输出过滤机制，模型可能输出不合规承诺或被恶意提示词注入，每次迭代都需重复编写校验逻辑。\n- **人工介入成本高**：遇到疑难杂症需要转人工时，没有标准化的“挂起 - 唤醒”机制，往往需要重新向人类客服陈述整个背景，效率极低。\n\n### 使用 openai-agents-python 后\n- **原生多代理编排**：利用 Handoffs 机制定义清晰的职责边界，主代理可一键将任务移交给专门的“退款代理”或“物流代理”，上下文自动无缝流转。\n- **内置会话与追踪**：Sessions 模块自动管理对话历史，配合 Tracing 可视化界面，开发者能像看调用栈一样清晰复盘每个代理的决策路径和耗时。\n- **标准化护栏体系**：通过 Guardrails 配置统一的安全检查策略，自动拦截敏感输入并修正违规输出，让业务逻辑更专注于核心功能而非防御性编程。\n- **流畅的人机协作**：借助 Human in the loop 特性，系统在不确定时自动暂停并请求人工确认，审批通过后自动恢复执行，实现了真正的混合智能工作流。\n\nopenai-agents-python 将原本碎片化的脚本拼凑升级为可观测、可管控的企业级多智能体协作网络，大幅降低了复杂自动化流程的开发与维护门槛。","https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fopenai_openai-agents-python_8acfe7a1.png","openai","OpenAI","https:\u002F\u002Foss.gittoolsai.com\u002Favatars\u002Fopenai_1960bbf4.png","",null,"https:\u002F\u002Fopenai.com\u002F","https:\u002F\u002Fgithub.com\u002Fopenai",[22,26,30,34],{"name":23,"color":24,"percentage":25},"Python","#3572A5",99.5,{"name":27,"color":28,"percentage":29},"Shell","#89e051",0.3,{"name":31,"color":32,"percentage":33},"PowerShell","#012456",0.1,{"name":35,"color":36,"percentage":37},"Makefile","#427819",0,20666,3389,"2026-04-09T14:08:37","MIT",2,"Linux, macOS, Windows","未说明",{"notes":46,"python":47,"dependencies":48},"该工具是一个轻量级框架，主要用于编排多 Agent 工作流，本身不运行本地大模型，因此无特定 GPU 或显存要求。需设置 OPENAI_API_KEY 环境变量。支持通过可选组件安装语音功能（voice）或 Redis 会话支持（redis）。推荐使用 venv 或 uv 管理环境。","3.10+",[49,50,51,52,53,54,55,56,57],"openai-agents","pydantic","requests","mcp-python-sdk","griffe","websockets (可选)","sqlalchemy (可选)","any-llm (可选)","litellm (可选)",[59,60,61,62],"开发框架","语言模型","Agent","图像",[64,65,66,67,68,14],"agents","ai","framework","llm","python","ready","2026-03-27T02:49:30.150509","2026-04-10T04:28:28.239578",[],[74,79,84,89,94,99,104,109,114,119,124,129,134,139,144,149,154,159,164,169],{"id":75,"version":76,"summary_zh":77,"released_at":78},180259,"v0.13.6","## 变更内容\n\n* 修复：#2863 由 @seratch 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2864 中实现的 SQLiteSession 导出延迟加载问题\n* 修复：#2856 由 @seratch 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2860 中修复的递归跟踪预览截断问题\n* 修复：#2854 由 @seratch 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2854 中增强 SQLAlchemySession 对临时 SQLite 锁的抵抗力\n\n### 文档及其他变更\n\n* 文档：针对 #2844 的更改更新，由 @seratch 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2845 中完成\n* 文档：由 @github-actions[bot] 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2847 中更新的翻译文档页面\n* 文档（追踪）：将 HoneyHive 添加到追踪集成列表中，由 @m1lestones 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2851 中完成\n* 文档：由 @github-actions[bot] 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2853 中更新的翻译文档页面\n* 版本 0.13.6，由 @github-actions[bot] 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2861 中发布\n\n## 新贡献者\n* @m1lestones 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2851 中完成了首次贡献\n\n**完整变更日志**：https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fcompare\u002Fv0.13.5...v0.13.6","2026-04-09T04:10:23",{"id":80,"version":81,"summary_zh":82,"released_at":83},180260,"v0.13.5","## 变更内容\n\n* 功能：#2807 支持本地 MCP 服务器的可调用审批策略，由 @seratch 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2818 中实现\n* 功能：#2135 添加公共 flush_traces API，由 @seratch 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2844 中实现\n* 修复：#2823 为可迭代的 vLLM\u002Fany-llm Reasoning 对象提取 AnyLLM 推理结果，由 @ziqi-cn 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2822 中实现\n* 修复：使用共享文件锁序列化 SQLite 会话写入，由 @seratch 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2843 中实现\n\n### 文档及其他变更\n\n* 杂项（依赖）：将 openai\u002Fcodex-action 从 1.4 升级至 1.6，由 @dependabot[bot] 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2819 中完成\n* 杂项（依赖）：将 astral-sh\u002Fsetup-uv 从 7.3.1 升级至 8.0.0，由 @dependabot[bot] 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2820 中完成\n* 文档：同步 examples.md 文件与当前的 examples 目录内容，由 @AbdullahBapra 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2827 中完成\n* 文档：更新已翻译的文档页面，由 @github-actions[bot] 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2828 中完成\n* 发布 0.13.5 版本，由 @github-actions[bot] 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2821 中完成\n\n## 新贡献者\n* @AbdullahBapra 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2827 中完成了首次贡献\n* @ziqi-cn 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2822 中完成了首次贡献\n\n**完整变更日志**：https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fcompare\u002Fv0.13.4...v0.13.5","2026-04-06T04:11:19",{"id":85,"version":86,"summary_zh":87,"released_at":88},180261,"v0.13.4","## 变更内容\n* 修复：#2806 在验证之前对 AnyLLM 响应进行输入转义，由 @seratch 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2813 中完成\n\n### 文档及其他变更\n\n* 修复：将扩展相关更改映射到 feature:extensions 标签，由 @seratch 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2814 中完成\n* 发布 0.13.4，由 @github-actions[bot] 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2815 中完成\n\n**完整变更日志**：https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fcompare\u002Fv0.13.3...v0.13.4","2026-04-01T02:37:47",{"id":90,"version":91,"summary_zh":92,"released_at":93},180262,"v0.13.3","## 变更内容\n\n* 修复：#2798 通过 @seratch 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2800 中的提交，避免在服务器会话跟踪器中出现过时的已挂载输入 ID。\n* 修复：#2797 通过 @seratch 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2799 中的提交，允许在聊天补全输入中接受原始 image_url 内容部分。\n\n### 文档及其他变更\n\n* 文档：@Muttaqi110 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2792 中添加了非 OpenAI 提供商的代码示例。\n* 文档：@github-actions[bot] 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2793 中更新了已翻译的文档页面。\n* 杂项：@seratch 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2795 中添加了针对 Ruff 整理的 Codex Stop 钩子。\n* 版本 0.13.3：由 @github-actions[bot] 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2801 中发布。\n\n## 新贡献者\n* @Muttaqi110 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2792 中完成了他们的首次贡献。\n\n**完整变更日志**：https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fcompare\u002Fv0.13.2...v0.13.3","2026-03-31T04:33:12",{"id":95,"version":96,"summary_zh":97,"released_at":98},180263,"v0.13.2","## 变更内容\n\n* 修复：#2783 直接依赖 griffelib 进行文档字符串解析，由 @seratch 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2791 中完成\n* 修复：#2776 将私有工具元数据排除在持久化会话项之外，由 @seratch 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2781 中完成\n* 修复：#2778 使 LiteLLM 的 reasoning_effort 在不同提供商之间保持可移植性，由 @seratch 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2782 中完成\n* 修复：更新较新模型的默认推理力度，由 @seratch 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2773 中完成\n* 功能：向 WebSearchTool 添加 external_web_access，由 @ymuichiro 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2786 中完成\n* 修复：将 LiteLLM 的上限版本锁定为 1.82.6，以缓解供应链攻击，由 @gn00295120 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2772 中完成\n\n### 文档及其他变更\n\n* 文档：更新页面以添加 any-llm 适配器，由 @seratch 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2715 中完成\n* 文档：更新已翻译的文档页面，由 @github-actions[bot] 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2771 中完成\n* 修复：改进实时模块中的一个不稳定测试，由 @seratch 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2787 中完成\n* 发布 0.13.2，由 @github-actions[bot] 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2774 中完成\n\n## 新贡献者\n* @ymuichiro 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2786 中完成了首次贡献\n\n**完整变更日志**：https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fcompare\u002Fv0.13.1...v0.13.2","2026-03-26T23:56:53",{"id":100,"version":101,"summary_zh":102,"released_at":103},180264,"v0.13.1","## 主要变更\n\n### Any-LLM 扩展\n\n自本版本起，扩展模块新增了 [any-llm](https:\u002F\u002Fgithub.com\u002Fmozilla-ai\u002Fany-llm) 适配器。请参阅更新后的文档和示例：https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Ftree\u002Fmain\u002Fexamples\u002Fmodel_providers\n\n## 变更内容\n\n* 功能：添加 any-llm 模型支持，并实现响应兼容的路由功能，由 @seratch 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2706 中完成。\n* 修复：在转换函数工具时保留静态 MCP 元数据，由 @seratch 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2769 中完成。\n* 修复：#2760 等待实时响应完成后再发起后续响应创建，由 @seratch 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2763 中完成。\n* 修复：将已取消的单个函数工具视为工具调用失败处理，由 @elainegan-openai 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2762 中完成。\n* 修复：使初始化通知的容忍度可配置，由 @elainegan-openai 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2765 中完成。\n* 修复：从 RunItem 联合类型中移除重复的 CompactionItem，由 @KanchiShimono 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2761 中完成。\n\n### 文档及其他变更\n\n* 文档：添加 0.13 版本的变更日志，由 @seratch 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2744 中完成。\n* 文档：更新翻译后的文档页面，由 @github-actions[bot] 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2759 中完成。\n* 修复：增强示例自动运行对 PATH 和端口冲突的健壮性，由 @seratch 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2770 中完成。\n* 发布 0.13.1 版本，由 @github-actions[bot] 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2768 中完成。\n\n\n**完整变更日志**：https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fcompare\u002Fv0.13.0...v0.13.1","2026-03-25T07:44:57",{"id":105,"version":106,"summary_zh":107,"released_at":108},180265,"v0.13.0","## 主要变更\n\n此小版本更新**不**引入破坏性变更，但包含一项重要的实时默认模型更新，以及新的 MCP 功能和运行时稳定性修复。亮点如下：\n\n- 默认的 WebSocket 实时模型现已更新为 `gpt-realtime-1.5`，因此新创建的实时代理无需额外配置即可使用较新的模型。\n- `MCPServer` 现在公开了 `list_resources()`、`list_resource_templates()` 和 `read_resource()` 方法；同时，`MCPServerStreamableHttp` 现在公开了 `session_id`，以便流式 HTTP 会话可以在重新连接或无状态工作进程中恢复。\n- 聊天补全集成现在可以通过 `should_replay_reasoning_content` 选项启用推理内容重播功能，从而提升 LiteLLM\u002FDeepSeek 等适配器在特定提供商上的推理与工具调用连续性。\n- 修复了多个运行时和会话边缘场景中的问题，包括 `SQLAlchemySession` 中并发首次写入的问题、推理内容剥离后出现孤立助手消息 ID 的压缩请求问题、`remove_all_tools()` 方法未能移除 MCP 和推理相关条目，以及函数工具批量执行器中的竞态条件。\n\n## 变更内容\n\n* 修复：将默认实时 WebSocket 模型设置为 gpt-realtime-1.5，由 @seratch 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2737 中完成。\n* 修复：#2729 避免函数工具批量执行器中的急切任务竞态问题，由 @seratch 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2731 中完成。\n* 修复：在运行循环中保留流式输出的护栏触发机制，由 @seratch 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2758 中完成。\n* 修复（压缩）：在移除推理内容后清理孤立的助手消息 ID，由 @Ratnaditya-J 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2728 中完成。\n* 功能：#2669 为聊天补全模型添加推理内容重播的可选功能，由 @seratch 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2670 中完成。\n* 修复：`remove_all_tools()` 方法未移除 MCP 和推理相关条目，由 @joaquinhuigomez 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2700 中完成。\n* 功能（MCP）：在 MCPServer 上公开 `list_resources`、`list_resource_templates` 和 `read_resource` 方法，由 @adityasingh2400 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2721 中完成。\n* 功能（MCP）：在 MCPServerStreamableHttp 上公开 `session_id`，由 @adityasingh2400 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2708 中完成。\n* 修复（内存）：使 SQLAlchemySession 的首次写入操作具备线程安全性，由 @FelmonFekadu 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2725 中完成。\n\n### 文档及其他变更\n\n* 文档：在追踪文档中添加 PromptLayer 集成链接，由 @adagradschool 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2718 中完成。\n* 文档：更新已翻译的文档页面，由 @github-actions[bot] 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2719 中完成。\n* 文档：在公开文档中突出显示 gpt-realtime-1.5，由 @teri-oai 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2724 中完成。\n* 文档：更新已翻译的文档页面，由 @github-actions[bot] 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2726 中完成。\n* ","2026-03-23T06:19:55",{"id":110,"version":111,"summary_zh":112,"released_at":113},180266,"v0.12.5","## 变更内容\n\n* 功能(mcp)：在 SSE\u002F可流式 HTTP 参数中公开 auth 和 httpx_client_factory，由 @adityasingh2400 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2713 中实现\n* 修复：在取消回退之前恢复流式嵌套代理输出，由 @elainegan-openai 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2714 中实现\n* 修复(mcp)：在隔离会话中重试 ClosedResourceError 错误，由 @elainegan-openai 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2711 中实现\n* 修复(mcp)：在隔离会话中重试 MCP 408 错误，由 @elainegan-openai 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2709 中实现\n\n### 文档及其他变更\n\n* 发布 0.12.5，由 @github-actions[bot] 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2716 中完成\n\n**完整变更日志**：https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fcompare\u002Fv0.12.4...v0.12.5","2026-03-19T07:18:48",{"id":115,"version":116,"summary_zh":117,"released_at":118},180267,"v0.12.4","## 变更内容\n\n* 修复：将已取消的 MCP 调用规范化为工具错误，由 @elainegan-openai 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2704 中完成\n* 修复：在隔离会话中重试瞬态可流式传输 HTTP 的 MCP 工具失败情况，由 @elainegan-openai 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2703 中完成\n* 修复（内存）：在 AdvancedSQLiteSession 中尊重自定义表名，由 @seratch 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2694 中完成\n* 修复（重试）：将抖动延迟限制为最大延迟，由 @Dhakshin2007 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2676 中完成\n\n### 文档及其他变更\n\n* ci：加强 PR 自动打标签功能，防止 diff 和输出漂移，由 @seratch 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2691 中完成\n* 文档：简化模型页面并进行 UI 微调，由 @seratch 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2696 中完成\n* 文档：更新翻译文档页面，由 @github-actions[bot] 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2697 中完成\n* 发布 0.12.4 版本，由 @github-actions[bot] 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2705 中完成\n\n## 新贡献者\n* @Dhakshin2007 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2676 中完成了首次贡献\n\n**完整变更日志**：https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fcompare\u002Fv0.12.3...v0.12.4","2026-03-18T02:10:27",{"id":120,"version":121,"summary_zh":122,"released_at":123},180268,"v0.12.3","## 变更内容\n\n* 修复：由 @elainegan-openai 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2681 中实现，将内部 MCP 工具的取消操作视为工具错误处理。\n* 修复：由 @elainegan-openai 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2682 中实现，按会话序列化可流式传输的 HTTP MCP 请求。\n\n### 文档及其他变更\n\n* 文档：由 @github-actions[bot] 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2668 中更新了翻译文档页面。\n* 文档：由 @luojiyin1987 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2674 中修复了提供程序注释和测试名称中的小拼写错误。\n* 版本 0.12.3：由 @github-actions[bot] 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2684 中发布。\n\n## 新贡献者\n* @luojiyin1987 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2674 中完成了首次贡献。\n* @elainegan-openai 在 https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2681 中完成了首次贡献。\n\n**完整变更日志**：https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fcompare\u002Fv0.12.2...v0.12.3","2026-03-16T04:04:07",{"id":125,"version":126,"summary_zh":127,"released_at":128},180269,"v0.12.2","## What's Changed\r\n\r\n* fix: #2664 drop orphan hosted shell calls before multi-turn replay by @seratch in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2665\r\n* fix: #2258 add normalized to_input_list mode for filtered handoff follow-ups by @seratch in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2667\r\n\r\n### Documentation & Other Changes\r\n\r\n* docs: update translated document pages by @github-actions[bot] in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2663\r\n* Release 0.12.2 by @github-actions[bot] in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2666\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fcompare\u002Fv0.12.1...v0.12.2","2026-03-14T01:18:44",{"id":130,"version":131,"summary_zh":132,"released_at":133},180270,"v0.12.1","## What's Changed\r\n\r\n* feat: #2658 preserve explicit approval rejection messages across resume flows by @seratch in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2660\r\n\r\n### Documentation & Other Changes\r\n\r\n* docs: add retry settings by @seratch in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2655\r\n* docs: update translated document pages by @github-actions[bot] in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2656\r\n* Release 0.12.1 by @github-actions[bot] in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2662\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fcompare\u002Fv0.12.0...v0.12.1","2026-03-13T06:45:49",{"id":135,"version":136,"summary_zh":137,"released_at":138},180271,"v0.12.0","## Key Changes\r\n\r\n### Opt-in retry settings\r\n\r\nNew opt-in settings for retries of model API calls is available in this version. The new configuration is available as part of `ModelSettings` and it can be passed either as run config or model settings per an agent. \r\n\r\nRefer to the following code examples for details:\r\n\r\n- https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fblob\u002Fmain\u002Fexamples\u002Fbasic\u002Fretry.py\r\n- https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fblob\u002Fmain\u002Fexamples\u002Fbasic\u002Fretry_litellm.py\r\n\r\n## What's Changed\r\n\r\n* feat: add opt-in model retry policies by @seratch in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2651\r\n\r\n### Documentation & Other Changes\r\n\r\n* Release 0.12.0 by @github-actions[bot] in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2653\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fcompare\u002Fv0.11.1...v0.12.0","2026-03-12T08:52:13",{"id":140,"version":141,"summary_zh":142,"released_at":143},180272,"v0.11.1","## What's Changed\r\n\r\n* fix: show the GA computer alias in tracing by @seratch in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2641\r\n\r\n### Documentation & Other Changes\r\n\r\n* docs: add tool search coverage across Python guides by @seratch in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2622\r\n* docs: cover GA computer tool migration across Python guides by @seratch in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2629\r\n* docs: update translated document pages by @github-actions[bot] in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2640\r\n* Release 0.11.1 by @github-actions[bot] in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2642\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fcompare\u002Fv0.11.0...v0.11.1","2026-03-09T06:33:38",{"id":145,"version":146,"summary_zh":147,"released_at":148},180273,"v0.11.0","## Key Changes\r\n\r\n### Tool search support\r\n\r\nAs of this version, the Agents SDK supports the tool search tool, including namespaces: https:\u002F\u002Fdevelopers.openai.com\u002Fapi\u002Fdocs\u002Fguides\u002Ftools-tool-search\u002F\r\n\r\nPlease refer to [this concrete example](https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fblob\u002Fmain\u002Fexamples\u002Ftools\u002Ftool_search.py), as well as the updated documentation.\r\n\r\n### Computer use tool GA\r\n\r\nAs of this version, you can use the gpt-5.4 model for the computer use tool, in addition to the existing computer-use-preview model: https:\u002F\u002Fdevelopers.openai.com\u002Fapi\u002Fdocs\u002Fguides\u002Ftools-computer-use\u002F\r\n\r\nPlease refer to [this concrete example](https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fblob\u002Fmain\u002Fexamples\u002Ftools\u002Fcomputer_use.py), as well as the updated documentation.\r\n\r\n## What's Changed\r\n\r\n* feat: add Responses API tool search support by @seratch in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2610\r\n* fix: #2624 migrate ComputerTool to the GA computer tool by @seratch in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2626\r\n* fix: #2636 make Computer preview metadata optional on GA requests by @seratch in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2639\r\n* fix: Serialize structured realtime tool outputs as JSON by @lavish0000 in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2608\r\n* fix: Finish reasoning summaries before text deltas by @lavish0000 in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2609\r\n* fix: #2617 lower minimum pydantic version to 2.12.2 by @seratch in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2620\r\n* fix(memory): make SQLiteSession satisfy Session protocol by @lavish0000 in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2619\r\n* fix: align RunState schema policy with release boundaries by @seratch in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2632\r\n* fix: preserve plain run context for run-context tool wrappers by @seratch in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2634\r\n* fix: #2603 preserve MCP tool title metadata across tool call items by @seratch in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2621\r\n* fix: enable handling video_url in litellm and chat completions models by @seokhyunan in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2614\r\n\r\n### Documentation & Other Changes\r\n\r\n* docs: updates for #2593 by @seratch in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2594\r\n* docs: refresh docs and examples to gpt-5.4 by @seratch in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2615\r\n* docs: Add local shell skill example by @ihower in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2612\r\n* docs: update translated document pages by @github-actions[bot] in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2606\r\n* docs: update translated document pages by @github-actions[bot] in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2613\r\n* docs: update translated document pages by @github-actions[bot] in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2616\r\n* docs: update translated document pages by @github-actions[bot] in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2627\r\n* chore: run pyright across src and tests in typecheck by @seratch in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2623\r\n* test: improve coverage for tracing and runtime helpers by @seratch in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2635\r\n* Release 0.11.0 by @github-actions[bot] in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2611\r\n\r\n## New Contributors\r\n* @lavish0000 made their first contribution in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2608\r\n* @seokhyunan made their first contribution in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2614\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fcompare\u002Fv0.10.5...v0.11.0\r\n","2026-03-09T04:10:12",{"id":150,"version":151,"summary_zh":152,"released_at":153},180274,"v0.10.5","## What's Changed\r\n\r\n* feat: #2492 add explicit MultiProvider prefix modes by @seratch in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2593\r\n* fix: #879 return McpError as a structured error result instead of crashing the agent run by @adityasingh2400 in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2598\r\n\r\n### Documents & Other Changes\r\n\r\n* docs: improve doc coverage for Dapr sessions and advanced model settings by @seratch in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2587\r\n* docs: update translated document pages by @github-actions[bot] in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2589\r\n* docs: clarify realtime transport boundaries and SIP attach flows by @seratch in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2591\r\n* docs: update translated document pages by @github-actions[bot] in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2592\r\n* docs: clarify agent runtime, handoff, HITL, and results flows by @seratch in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2596\r\n* docs: update translated document pages by @github-actions[bot] in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2597\r\n* chore: add implementation strategy guidance for compatibility decisions by @seratch in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2600\r\n* docs: fix deprecated capwords import for Python 3.11+ compatibility by @Jah-yee in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2605\r\n* Release 0.10.5 by @github-actions[bot] in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2595\r\n\r\n## New Contributors\r\n* @adityasingh2400 made their first contribution in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2598\r\n* @Jah-yee made their first contribution in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2605\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fcompare\u002Fv0.10.4...v0.10.5","2026-03-05T20:43:30",{"id":155,"version":156,"summary_zh":157,"released_at":158},180275,"v0.10.4","## What's Changed\r\n\r\n* fix: isolate parallel function tool failures by @seratch in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2584\r\n\r\n### Documents & Other Changes\r\n\r\n* docs: improve documentation entry points and navigation by @seratch in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2578\r\n* docs: update translated document pages by @github-actions[bot] in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2579\r\n* docs: update translated document pages by @github-actions[bot] in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2581\r\n* ci: add repeated async teardown stability checks by @seratch in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2585\r\n* Release 0.10.4 by @github-actions[bot] in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2582\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fcompare\u002Fv0.10.3...v0.10.4","2026-03-03T21:19:50",{"id":160,"version":161,"summary_zh":162,"released_at":163},180276,"v0.10.3","## What's Changed\r\n\r\n* feat: expose immutable agent tool invocation metadata on run results (ref #2575) by @seratch in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2576\r\n* feat: #2247 add RunResult tool_context accessor for agent tools by @seratch in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2575\r\n* perf: use deque for BFS queue and voice task ordering by @giulio-leone in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2559\r\n\r\n### Documents & Other Changes\r\n\r\n* docs: update translated document pages by @github-actions[bot] in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2555\r\n* docs: Add `nest_handoff_history` to `Handoffs` docs by @Hammad-Noman in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2560\r\n* docs: update translated document pages by @github-actions[bot] in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2561\r\n* docs: update a few docs and code comments by @seratch in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2564\r\n* docs: update translated document pages by @github-actions[bot] in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2565\r\n* docs: reorganize navigation and clarify runtime guides by @seratch in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2568\r\n* docs: update translated document pages by @github-actions[bot] in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2570\r\n* docs: update translated document pages by @github-actions[bot] in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2571\r\n* docs: change http to https in tracing links by @04cb in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2572\r\n* test: speed up make tests with xdist and deterministic waits by @seratch in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2563\r\n* test(mcp): assert runner emits tool_call_output_item on MCP failures by @OiPunk in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2556\r\n* chore(deps): bump astral-sh\u002Fsetup-uv from 7.2.1 to 7.3.1 by @dependabot[bot] in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2566\r\n* chore(deps): bump actions\u002Fstale from 10.1.1 to 10.2.0 by @dependabot[bot] in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2567\r\n* Release 0.10.3 by @github-actions[bot] in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2569\r\n\r\n## New Contributors\r\n* @giulio-leone made their first contribution in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2559\r\n* @Hammad-Noman made their first contribution in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2560\r\n* @04cb made their first contribution in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2572\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fcompare\u002Fv0.10.2...v0.10.3","2026-03-02T05:13:46",{"id":165,"version":166,"summary_zh":167,"released_at":168},180277,"v0.10.2","## What's Changed\r\n\r\n* fix: #2540 reattach resumed traces without duplicate trace starts by @seratch in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2547\r\n* fix: #260 sanitize oversized tracing span payloads by @seratch in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2549\r\n* fix: Require approval for callable MCP policies when agent is omitted by @tiffanycitra in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2553\r\n* fix: #1121 expose model request IDs on raw responses by @seratch in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2552\r\n\r\n### Documents & Other Changes\r\n\r\n* docs: structure improvements by @seratch in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2538\r\n* docs: update translated document pages by @github-actions[bot] in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2539\r\n* Release 0.10.2 by @github-actions[bot] in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2548\r\n\r\n## New Contributors\r\n* @tiffanycitra made their first contribution in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2553\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fcompare\u002Fv0.10.1...v0.10.2","2026-02-26T08:06:10",{"id":170,"version":171,"summary_zh":172,"released_at":173},180278,"v0.10.1","## What's Changed\r\n\r\n* fix: sync realtime model literals in realtime config by @seratch in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2535\r\n\r\n### Documents & Other Changes\r\n\r\n* docs: add responses websocket support by @seratch in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2533\r\n* docs: update translated document pages by @github-actions[bot] in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2534\r\n* Release 0.10.1 by @github-actions[bot] in https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fpull\u002F2536\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-agents-python\u002Fcompare\u002Fv0.10.0...v0.10.1","2026-02-24T01:48:51",[175,185,193,201,209,218],{"id":176,"name":177,"github_repo":178,"description_zh":179,"stars":180,"difficulty_score":181,"last_commit_at":182,"category_tags":183,"status":69},4358,"openclaw","openclaw\u002Fopenclaw","OpenClaw 是一款专为个人打造的本地化 AI 助手，旨在让你在自己的设备上拥有完全可控的智能伙伴。它打破了传统 AI 助手局限于特定网页或应用的束缚，能够直接接入你日常使用的各类通讯渠道，包括微信、WhatsApp、Telegram、Discord、iMessage 等数十种平台。无论你在哪个聊天软件中发送消息，OpenClaw 都能即时响应，甚至支持在 macOS、iOS 和 Android 设备上进行语音交互，并提供实时的画布渲染功能供你操控。\n\n这款工具主要解决了用户对数据隐私、响应速度以及“始终在线”体验的需求。通过将 AI 部署在本地，用户无需依赖云端服务即可享受快速、私密的智能辅助，真正实现了“你的数据，你做主”。其独特的技术亮点在于强大的网关架构，将控制平面与核心助手分离，确保跨平台通信的流畅性与扩展性。\n\nOpenClaw 非常适合希望构建个性化工作流的技术爱好者、开发者，以及注重隐私保护且不愿被单一生态绑定的普通用户。只要具备基础的终端操作能力（支持 macOS、Linux 及 Windows WSL2），即可通过简单的命令行引导完成部署。如果你渴望拥有一个懂你",349277,3,"2026-04-06T06:32:30",[61,59,62,184],"数据工具",{"id":186,"name":187,"github_repo":188,"description_zh":189,"stars":190,"difficulty_score":181,"last_commit_at":191,"category_tags":192,"status":69},3808,"stable-diffusion-webui","AUTOMATIC1111\u002Fstable-diffusion-webui","stable-diffusion-webui 是一个基于 Gradio 构建的网页版操作界面，旨在让用户能够轻松地在本地运行和使用强大的 Stable Diffusion 图像生成模型。它解决了原始模型依赖命令行、操作门槛高且功能分散的痛点，将复杂的 AI 绘图流程整合进一个直观易用的图形化平台。\n\n无论是希望快速上手的普通创作者、需要精细控制画面细节的设计师，还是想要深入探索模型潜力的开发者与研究人员，都能从中获益。其核心亮点在于极高的功能丰富度：不仅支持文生图、图生图、局部重绘（Inpainting）和外绘（Outpainting）等基础模式，还独创了注意力机制调整、提示词矩阵、负向提示词以及“高清修复”等高级功能。此外，它内置了 GFPGAN 和 CodeFormer 等人脸修复工具，支持多种神经网络放大算法，并允许用户通过插件系统无限扩展能力。即使是显存有限的设备，stable-diffusion-webui 也提供了相应的优化选项，让高质量的 AI 艺术创作变得触手可及。",162132,"2026-04-05T11:01:52",[59,62,61],{"id":194,"name":195,"github_repo":196,"description_zh":197,"stars":198,"difficulty_score":42,"last_commit_at":199,"category_tags":200,"status":69},1381,"everything-claude-code","affaan-m\u002Feverything-claude-code","everything-claude-code 是一套专为 AI 编程助手（如 Claude Code、Codex、Cursor 等）打造的高性能优化系统。它不仅仅是一组配置文件，而是一个经过长期实战打磨的完整框架，旨在解决 AI 代理在实际开发中面临的效率低下、记忆丢失、安全隐患及缺乏持续学习能力等核心痛点。\n\n通过引入技能模块化、直觉增强、记忆持久化机制以及内置的安全扫描功能，everything-claude-code 能显著提升 AI 在复杂任务中的表现，帮助开发者构建更稳定、更智能的生产级 AI 代理。其独特的“研究优先”开发理念和针对 Token 消耗的优化策略，使得模型响应更快、成本更低，同时有效防御潜在的攻击向量。\n\n这套工具特别适合软件开发者、AI 研究人员以及希望深度定制 AI 工作流的技术团队使用。无论您是在构建大型代码库，还是需要 AI 协助进行安全审计与自动化测试，everything-claude-code 都能提供强大的底层支持。作为一个曾荣获 Anthropic 黑客大奖的开源项目，它融合了多语言支持与丰富的实战钩子（hooks），让 AI 真正成长为懂上",147882,"2026-04-09T11:32:47",[59,61,60],{"id":202,"name":203,"github_repo":204,"description_zh":205,"stars":206,"difficulty_score":42,"last_commit_at":207,"category_tags":208,"status":69},2271,"ComfyUI","Comfy-Org\u002FComfyUI","ComfyUI 是一款功能强大且高度模块化的视觉 AI 引擎，专为设计和执行复杂的 Stable Diffusion 图像生成流程而打造。它摒弃了传统的代码编写模式，采用直观的节点式流程图界面，让用户通过连接不同的功能模块即可构建个性化的生成管线。\n\n这一设计巧妙解决了高级 AI 绘图工作流配置复杂、灵活性不足的痛点。用户无需具备编程背景，也能自由组合模型、调整参数并实时预览效果，轻松实现从基础文生图到多步骤高清修复等各类复杂任务。ComfyUI 拥有极佳的兼容性，不仅支持 Windows、macOS 和 Linux 全平台，还广泛适配 NVIDIA、AMD、Intel 及苹果 Silicon 等多种硬件架构，并率先支持 SDXL、Flux、SD3 等前沿模型。\n\n无论是希望深入探索算法潜力的研究人员和开发者，还是追求极致创作自由度的设计师与资深 AI 绘画爱好者，ComfyUI 都能提供强大的支持。其独特的模块化架构允许社区不断扩展新功能，使其成为当前最灵活、生态最丰富的开源扩散模型工具之一，帮助用户将创意高效转化为现实。",108111,"2026-04-08T11:23:26",[59,62,61],{"id":210,"name":211,"github_repo":212,"description_zh":213,"stars":214,"difficulty_score":42,"last_commit_at":215,"category_tags":216,"status":69},4721,"markitdown","microsoft\u002Fmarkitdown","MarkItDown 是一款由微软 AutoGen 团队打造的轻量级 Python 工具，专为将各类文件高效转换为 Markdown 格式而设计。它支持 PDF、Word、Excel、PPT、图片（含 OCR）、音频（含语音转录）、HTML 乃至 YouTube 链接等多种格式的解析，能够精准提取文档中的标题、列表、表格和链接等关键结构信息。\n\n在人工智能应用日益普及的今天，大语言模型（LLM）虽擅长处理文本，却难以直接读取复杂的二进制办公文档。MarkItDown 恰好解决了这一痛点，它将非结构化或半结构化的文件转化为模型“原生理解”且 Token 效率极高的 Markdown 格式，成为连接本地文件与 AI 分析 pipeline 的理想桥梁。此外，它还提供了 MCP（模型上下文协议）服务器，可无缝集成到 Claude Desktop 等 LLM 应用中。\n\n这款工具特别适合开发者、数据科学家及 AI 研究人员使用，尤其是那些需要构建文档检索增强生成（RAG）系统、进行批量文本分析或希望让 AI 助手直接“阅读”本地文件的用户。虽然生成的内容也具备一定可读性，但其核心优势在于为机器",93400,"2026-04-06T19:52:38",[217,59],"插件",{"id":219,"name":220,"github_repo":221,"description_zh":222,"stars":223,"difficulty_score":181,"last_commit_at":224,"category_tags":225,"status":69},4487,"LLMs-from-scratch","rasbt\u002FLLMs-from-scratch","LLMs-from-scratch 是一个基于 PyTorch 的开源教育项目，旨在引导用户从零开始一步步构建一个类似 ChatGPT 的大型语言模型（LLM）。它不仅是同名技术著作的官方代码库，更提供了一套完整的实践方案，涵盖模型开发、预训练及微调的全过程。\n\n该项目主要解决了大模型领域“黑盒化”的学习痛点。许多开发者虽能调用现成模型，却难以深入理解其内部架构与训练机制。通过亲手编写每一行核心代码，用户能够透彻掌握 Transformer 架构、注意力机制等关键原理，从而真正理解大模型是如何“思考”的。此外，项目还包含了加载大型预训练权重进行微调的代码，帮助用户将理论知识延伸至实际应用。\n\nLLMs-from-scratch 特别适合希望深入底层原理的 AI 开发者、研究人员以及计算机专业的学生。对于不满足于仅使用 API，而是渴望探究模型构建细节的技术人员而言，这是极佳的学习资源。其独特的技术亮点在于“循序渐进”的教学设计：将复杂的系统工程拆解为清晰的步骤，配合详细的图表与示例，让构建一个虽小但功能完备的大模型变得触手可及。无论你是想夯实理论基础，还是为未来研发更大规模的模型做准备",90106,"2026-04-06T11:19:32",[60,62,61,59]]