[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"similar-assafelovic--gpt-researcher":3,"tool-assafelovic--gpt-researcher":64},[4,17,27,35,43,56],{"id":5,"name":6,"github_repo":7,"description_zh":8,"stars":9,"difficulty_score":10,"last_commit_at":11,"category_tags":12,"status":16},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。",162132,3,"2026-04-05T11:01:52",[13,14,15],"开发框架","图像","Agent","ready",{"id":18,"name":19,"github_repo":20,"description_zh":21,"stars":22,"difficulty_score":23,"last_commit_at":24,"category_tags":25,"status":16},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",138956,2,"2026-04-05T11:33:21",[13,15,26],"语言模型",{"id":28,"name":29,"github_repo":30,"description_zh":31,"stars":32,"difficulty_score":23,"last_commit_at":33,"category_tags":34,"status":16},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。",107662,"2026-04-03T11:11:01",[13,14,15],{"id":36,"name":37,"github_repo":38,"description_zh":39,"stars":40,"difficulty_score":23,"last_commit_at":41,"category_tags":42,"status":16},3704,"NextChat","ChatGPTNextWeb\u002FNextChat","NextChat 是一款轻量且极速的 AI 助手，旨在为用户提供流畅、跨平台的大模型交互体验。它完美解决了用户在多设备间切换时难以保持对话连续性，以及面对众多 AI 模型不知如何统一管理的痛点。无论是日常办公、学习辅助还是创意激发，NextChat 都能让用户随时随地通过网页、iOS、Android、Windows、MacOS 或 Linux 端无缝接入智能服务。\n\n这款工具非常适合普通用户、学生、职场人士以及需要私有化部署的企业团队使用。对于开发者而言，它也提供了便捷的自托管方案，支持一键部署到 Vercel 或 Zeabur 等平台。\n\nNextChat 的核心亮点在于其广泛的模型兼容性，原生支持 Claude、DeepSeek、GPT-4 及 Gemini Pro 等主流大模型，让用户在一个界面即可自由切换不同 AI 能力。此外，它还率先支持 MCP（Model Context Protocol）协议，增强了上下文处理能力。针对企业用户，NextChat 提供专业版解决方案，具备品牌定制、细粒度权限控制、内部知识库整合及安全审计等功能，满足公司对数据隐私和个性化管理的高标准要求。",87618,"2026-04-05T07:20:52",[13,26],{"id":44,"name":45,"github_repo":46,"description_zh":47,"stars":48,"difficulty_score":23,"last_commit_at":49,"category_tags":50,"status":16},2268,"ML-For-Beginners","microsoft\u002FML-For-Beginners","ML-For-Beginners 是由微软推出的一套系统化机器学习入门课程，旨在帮助零基础用户轻松掌握经典机器学习知识。这套课程将学习路径规划为 12 周，包含 26 节精炼课程和 52 道配套测验，内容涵盖从基础概念到实际应用的完整流程，有效解决了初学者面对庞大知识体系时无从下手、缺乏结构化指导的痛点。\n\n无论是希望转型的开发者、需要补充算法背景的研究人员，还是对人工智能充满好奇的普通爱好者，都能从中受益。课程不仅提供了清晰的理论讲解，还强调动手实践，让用户在循序渐进中建立扎实的技能基础。其独特的亮点在于强大的多语言支持，通过自动化机制提供了包括简体中文在内的 50 多种语言版本，极大地降低了全球不同背景用户的学习门槛。此外，项目采用开源协作模式，社区活跃且内容持续更新，确保学习者能获取前沿且准确的技术资讯。如果你正寻找一条清晰、友好且专业的机器学习入门之路，ML-For-Beginners 将是理想的起点。",84991,"2026-04-05T10:45:23",[14,51,52,53,15,54,26,13,55],"数据工具","视频","插件","其他","音频",{"id":57,"name":58,"github_repo":59,"description_zh":60,"stars":61,"difficulty_score":10,"last_commit_at":62,"category_tags":63,"status":16},3128,"ragflow","infiniflow\u002Fragflow","RAGFlow 是一款领先的开源检索增强生成（RAG）引擎，旨在为大语言模型构建更精准、可靠的上下文层。它巧妙地将前沿的 RAG 技术与智能体（Agent）能力相结合，不仅支持从各类文档中高效提取知识，还能让模型基于这些知识进行逻辑推理和任务执行。\n\n在大模型应用中，幻觉问题和知识滞后是常见痛点。RAGFlow 通过深度解析复杂文档结构（如表格、图表及混合排版），显著提升了信息检索的准确度，从而有效减少模型“胡编乱造”的现象，确保回答既有据可依又具备时效性。其内置的智能体机制更进一步，使系统不仅能回答问题，还能自主规划步骤解决复杂问题。\n\n这款工具特别适合开发者、企业技术团队以及 AI 研究人员使用。无论是希望快速搭建私有知识库问答系统，还是致力于探索大模型在垂直领域落地的创新者，都能从中受益。RAGFlow 提供了可视化的工作流编排界面和灵活的 API 接口，既降低了非算法背景用户的上手门槛，也满足了专业开发者对系统深度定制的需求。作为基于 Apache 2.0 协议开源的项目，它正成为连接通用大模型与行业专有知识之间的重要桥梁。",77062,"2026-04-04T04:44:48",[15,14,13,26,54],{"id":65,"github_repo":66,"name":67,"description_en":68,"description_zh":69,"ai_summary_zh":69,"readme_en":70,"readme_zh":71,"quickstart_zh":72,"use_case_zh":73,"hero_image_url":74,"owner_login":75,"owner_name":76,"owner_avatar_url":77,"owner_bio":78,"owner_company":79,"owner_location":80,"owner_email":80,"owner_twitter":81,"owner_website":82,"owner_url":83,"languages":84,"stars":117,"forks":118,"last_commit_at":119,"license":120,"difficulty_score":23,"env_os":121,"env_gpu":122,"env_ram":122,"env_deps":123,"category_tags":134,"github_topics":135,"view_count":23,"oss_zip_url":80,"oss_zip_packed_at":80,"status":16,"created_at":147,"updated_at":148,"faqs":149,"releases":179},3578,"assafelovic\u002Fgpt-researcher","gpt-researcher","An autonomous agent that conducts deep research on any data using any LLM providers","GPT Researcher 是一款开源的自主智能体，专为执行深度网络和本地研究任务而设计。它能够利用任意大语言模型提供商的能力，针对任何主题自动开展调研，并生成包含引用来源的详细、客观且事实准确的研究报告。\n\n传统人工研究往往耗时数周且资源消耗巨大，而现有大模型常因训练数据过时产生“幻觉”，或受限于上下文长度无法生成长篇报告，加之网络来源单一导致信息片面甚至误导。GPT Researcher 正是为解决这些痛点而生。它借鉴了“规划与求解”及检索增强生成（RAG）的前沿理念，通过独特的“规划者”与“执行者”多智能体协作架构，将研究问题拆解并行处理。这种机制不仅显著提升了检索速度和稳定性，还有效消除了信息来源的偏差，确保结论的可靠性。\n\n这款工具非常适合需要高效获取深度行业洞察的研究人员、分析师，以及希望构建定制化领域研究助手的开发者。无论是撰写学术综述、市场分析还是技术调研，GPT Researcher 都能帮助用户在极短时间内获得高质量、可验证的专业信息，让准确的事实触手可及。","\u003Cdiv align=\"center\" id=\"top\">\n\n\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fassafelovic_gpt-researcher_readme_cbf8f73ce226.png\" alt=\"Logo\" width=\"80\">\n\n####\n\n[![Website](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FOfficial%20Website-gptr.dev-teal?style=for-the-badge&logo=world&logoColor=white&color=0891b2)](https:\u002F\u002Fgptr.dev)\n[![Documentation](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FDocumentation-DOCS-f472b6?logo=googledocs&logoColor=white&style=for-the-badge)](https:\u002F\u002Fdocs.gptr.dev)\n[![Discord](https:\u002F\u002Fimg.shields.io\u002Fdiscord\u002F1127851779011391548?logo=discord&logoColor=white&label=Discord&color=34b76a&style=for-the-badge)](https:\u002F\u002Fdiscord.gg\u002FQgZXvJAccX)\n\n\n[![PyPI version](https:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fv\u002Fgpt-researcher?logo=pypi&logoColor=white&style=flat)](https:\u002F\u002Fbadge.fury.io\u002Fpy\u002Fgpt-researcher)\n![GitHub Release](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fv\u002Frelease\u002Fassafelovic\u002Fgpt-researcher?style=flat&logo=github)\n[![Open In Colab](https:\u002F\u002Fimg.shields.io\u002Fstatic\u002Fv1?message=Open%20in%20Colab&logo=googlecolab&labelColor=grey&color=yellow&label=%20&style=flat&logoSize=40)](https:\u002F\u002Fcolab.research.google.com\u002Fgithub\u002Fassafelovic\u002Fgpt-researcher\u002Fblob\u002Fmaster\u002Fdocs\u002Fdocs\u002Fexamples\u002Fpip-run.ipynb)\n[![Docker Image Version](https:\u002F\u002Fimg.shields.io\u002Fdocker\u002Fv\u002Felestio\u002Fgpt-researcher\u002Flatest?arch=amd64&style=flat&logo=docker&logoColor=white&color=1D63ED)](https:\u002F\u002Fhub.docker.com\u002Fr\u002Fgptresearcher\u002Fgpt-researcher)\n[![Skill](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FClaude%20Skill-skills.sh-blueviolet?style=flat&logo=anthropic&logoColor=white)](https:\u002F\u002Fskills.sh\u002Fassafelovic\u002Fgpt-researcher\u002Fgpt-researcher)\n[![Twitter Follow](https:\u002F\u002Fimg.shields.io\u002Ftwitter\u002Ffollow\u002Fassaf_elovic?style=social)](https:\u002F\u002Ftwitter.com\u002Fassaf_elovic)\n\n[English](README.md) | [中文](README-zh_CN.md) | [日本語](README-ja_JP.md) | [한국어](README-ko_KR.md)\n\n\u003C\u002Fdiv>\n\n# 🔎 GPT Researcher\n\n**GPT Researcher is an open deep research agent designed for both web and local research on any given task.** \n\nThe agent produces detailed, factual, and unbiased research reports with citations. GPT Researcher provides a full suite of customization options to create tailor made and domain specific research agents. Inspired by the recent [Plan-and-Solve](https:\u002F\u002Farxiv.org\u002Fabs\u002F2305.04091) and [RAG](https:\u002F\u002Farxiv.org\u002Fabs\u002F2005.11401) papers, GPT Researcher addresses misinformation, speed, determinism, and reliability by offering stable performance and increased speed through parallelized agent work.\n\n**Our mission is to empower individuals and organizations with accurate, unbiased, and factual information through AI.**\n\n## Why GPT Researcher?\n\n- Objective conclusions for manual research can take weeks, requiring vast resources and time.\n- LLMs trained on outdated information can hallucinate, becoming irrelevant for current research tasks.\n- Current LLMs have token limitations, insufficient for generating long research reports.\n- Limited web sources in existing services lead to misinformation and shallow results.\n- Selective web sources can introduce bias into research tasks.\n\n## Demo\n\u003Ca href=\"https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=f60rlc_QCxE\" target=\"_blank\" rel=\"noopener\">\n  \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fassafelovic_gpt-researcher_readme_f5625647c7b8.png\" alt=\"Demo video\" width=\"800\" target=\"_blank\" \u002F>\n\u003C\u002Fa>\n\n## Install as Claude Skill\n\nExtend Claude's deep research capabilities by installing GPT Researcher as a [Claude Skill](https:\u002F\u002Fskills.sh\u002Fassafelovic\u002Fgpt-researcher\u002Fgpt-researcher):\n\n```bash\nnpx skills add assafelovic\u002Fgpt-researcher\n```\n\nOnce installed, Claude can leverage GPT Researcher's deep research capabilities directly within your conversations.\n\n## Architecture\n\nThe core idea is to utilize 'planner' and 'execution' agents. The planner generates research questions, while the execution agents gather relevant information. The publisher then aggregates all findings into a comprehensive report.\n\n\u003Cdiv align=\"center\">\n\u003Cimg align=\"center\" height=\"600\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fassafelovic_gpt-researcher_readme_679ff9ca3bff.png\">\n\u003C\u002Fdiv>\n\nSteps:\n* Create a task-specific agent based on a research query.\n* Generate questions that collectively form an objective opinion on the task.\n* Use a crawler agent for gathering information for each question.\n* Summarize and source-track each resource.\n* Filter and aggregate summaries into a final research report.\n\n## Tutorials\n - [How it Works](https:\u002F\u002Fdocs.gptr.dev\u002Fblog\u002Fbuilding-gpt-researcher)\n - [How to Install](https:\u002F\u002Fwww.loom.com\u002Fshare\u002F04ebffb6ed2a4520a27c3e3addcdde20?sid=da1848e8-b1f1-42d1-93c3-5b0b9c3b24ea)\n - [Live Demo](https:\u002F\u002Fwww.loom.com\u002Fshare\u002F6a3385db4e8747a1913dd85a7834846f?sid=a740fd5b-2aa3-457e-8fb7-86976f59f9b8)\n\n## Features\n\n- 📝 Generate detailed research reports using web and local documents.\n- 🖼️ Smart image scraping and filtering for reports.\n- 🍌 **AI-generated inline images** using Google Gemini (Nano Banana) for visual illustrations.\n- 📜 Generate detailed reports exceeding 2,000 words.\n- 🌐 Aggregate over 20 sources for objective conclusions.\n- 🖥️ Frontend available in lightweight (HTML\u002FCSS\u002FJS) and production-ready (NextJS + Tailwind) versions.\n- 🔍 JavaScript-enabled web scraping.\n- 📂 Maintains memory and context throughout research.\n- 📄 Export reports to PDF, Word, and other formats.\n\n## 📖 Documentation\n\nSee the [Documentation](https:\u002F\u002Fdocs.gptr.dev\u002Fdocs\u002Fgpt-researcher\u002Fgetting-started) for:\n- Installation and setup guides\n- Configuration and customization options\n- How-To examples\n- Full API references\n\n## ⚙️ Getting Started\n\n### Installation\n\n1. Install Python 3.11 or later. [Guide](https:\u002F\u002Fwww.tutorialsteacher.com\u002Fpython\u002Finstall-python).\n2. Clone the project and navigate to the directory:\n\n    ```bash\n    git clone https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher.git\n    cd gpt-researcher\n    ```\n\n3. Set up API keys by exporting them or storing them in a `.env` file.\n\n    ```bash\n    export OPENAI_API_KEY={Your OpenAI API Key here}\n    export TAVILY_API_KEY={Your Tavily API Key here}\n    ```\n\n    (Optional) For enhanced tracing and observability, you can also set:\n    \n    ```bash\n    # export LANGCHAIN_TRACING_V2=true\n    # export LANGCHAIN_API_KEY={Your LangChain API Key here}\n    ```\n\n    For custom OpenAI-compatible APIs (e.g., local models, other providers), you can also set:\n    \n    ```bash\n    export OPENAI_BASE_URL={Your custom API base URL here}\n    ```\n\n4. Install dependencies and start the server:\n\n    ```bash\n    pip install -r requirements.txt\n    python -m uvicorn main:app --reload\n    ```\n\nVisit [http:\u002F\u002Flocalhost:8000](http:\u002F\u002Flocalhost:8000) to start.\n\nFor other setups (e.g., Poetry or virtual environments), check the [Getting Started page](https:\u002F\u002Fdocs.gptr.dev\u002Fdocs\u002Fgpt-researcher\u002Fgetting-started).\n\n## Run as PIP package\n```bash\npip install gpt-researcher\n\n```\n### Example Usage:\n```python\n...\nfrom gpt_researcher import GPTResearcher\n\nquery = \"why is Nvidia stock going up?\"\nresearcher = GPTResearcher(query=query)\n# Conduct research on the given query\nresearch_result = await researcher.conduct_research()\n# Write the report\nreport = await researcher.write_report()\n...\n```\n\n**For more examples and configurations, please refer to the [PIP documentation](https:\u002F\u002Fdocs.gptr.dev\u002Fdocs\u002Fgpt-researcher\u002Fgptr\u002Fpip-package) page.**\n\n### 🔧 MCP Client\nGPT Researcher supports MCP integration to connect with specialized data sources like GitHub repositories, databases, and custom APIs. This enables research from data sources alongside web search.\n\n```bash\nexport RETRIEVER=tavily,mcp  # Enable hybrid web + MCP research\n```\n\n```python\nfrom gpt_researcher import GPTResearcher\nimport asyncio\nimport os\n\nasync def mcp_research_example():\n    # Enable MCP with web search\n    os.environ[\"RETRIEVER\"] = \"tavily,mcp\"\n    \n    researcher = GPTResearcher(\n        query=\"What are the top open source web research agents?\",\n        mcp_configs=[\n            {\n                \"name\": \"github\",\n                \"command\": \"npx\",\n                \"args\": [\"-y\", \"@modelcontextprotocol\u002Fserver-github\"],\n                \"env\": {\"GITHUB_TOKEN\": os.getenv(\"GITHUB_TOKEN\")}\n            }\n        ]\n    )\n    \n    research_result = await researcher.conduct_research()\n    report = await researcher.write_report()\n    return report\n```\n\n> For comprehensive MCP documentation and advanced examples, visit the [MCP Integration Guide](https:\u002F\u002Fdocs.gptr.dev\u002Fdocs\u002Fgpt-researcher\u002Fretrievers\u002Fmcp-configs).\n\n## 🍌 Inline Image Generation\n\nGPT Researcher can automatically generate and embed AI-created illustrations in your research reports using Google's Gemini models (Nano Banana).\n\n```bash\n# Enable in your .env file\nIMAGE_GENERATION_ENABLED=true\nGOOGLE_API_KEY=your_google_api_key\nIMAGE_GENERATION_MODEL=models\u002Fgemini-2.5-flash-image\n```\n\nWhen enabled, the system will:\n1. Analyze your research context to identify visualization opportunities\n2. Pre-generate 2-3 relevant images during the research phase\n3. Embed them inline as the report is written\n\nImages are generated with dark-mode styling that matches the GPT Researcher UI, featuring professional infographic aesthetics with teal accents.\n\n[Learn more about Image Generation](https:\u002F\u002Fdocs.gptr.dev\u002Fdocs\u002Fgpt-researcher\u002Fgptr\u002Fimage_generation) in our documentation.\n\n## ✨ Deep Research\n\nGPT Researcher now includes Deep Research - an advanced recursive research workflow that explores topics with agentic depth and breadth. This feature employs a tree-like exploration pattern, diving deeper into subtopics while maintaining a comprehensive view of the research subject.\n\n- 🌳 Tree-like exploration with configurable depth and breadth\n- ⚡️ Concurrent processing for faster results\n- 🤝 Smart context management across research branches\n- ⏱️ Takes ~5 minutes per deep research\n- 💰 Costs ~$0.4 per research (using `o3-mini` on \"high\" reasoning effort)\n\n[Learn more about Deep Research](https:\u002F\u002Fdocs.gptr.dev\u002Fdocs\u002Fgpt-researcher\u002Fgptr\u002Fdeep_research) in our documentation.\n\n## Run with Docker\n\n> **Step 1** - [Install Docker](https:\u002F\u002Fdocs.gptr.dev\u002Fdocs\u002Fgpt-researcher\u002Fgetting-started\u002Fgetting-started-with-docker)\n\n> **Step 2** - Clone the '.env.example' file, add your API Keys to the cloned file and save the file as '.env'\n\n> **Step 3** - Within the docker-compose file comment out services that you don't want to run with Docker.\n\n```bash\ndocker-compose up --build\n```\n\nIf that doesn't work, try running it without the dash:\n```bash\ndocker compose up --build\n```\n\n> **Step 4** - By default, if you haven't uncommented anything in your docker-compose file, this flow will start 2 processes:\n - the Python server running on localhost:8000\u003Cbr>\n - the React app running on localhost:3000\u003Cbr>\n\nVisit localhost:3000 on any browser and enjoy researching!\n\n\n## 📄 Research on Local Documents\n\nYou can instruct the GPT Researcher to run research tasks based on your local documents. Currently supported file formats are: PDF, plain text, CSV, Excel, Markdown, PowerPoint, and Word documents.\n\nStep 1: Add the env variable `DOC_PATH` pointing to the folder where your documents are located.\n\n```bash\nexport DOC_PATH=\".\u002Fmy-docs\"\n```\n\nStep 2: \n - If you're running the frontend app on localhost:8000, simply select \"My Documents\" from the \"Report Source\" Dropdown Options.\n - If you're running GPT Researcher with the [PIP package](https:\u002F\u002Fdocs.tavily.com\u002Fguides\u002Fgpt-researcher\u002Fgpt-researcher#pip-package), pass the `report_source` argument as \"local\" when you instantiate the `GPTResearcher` class [code sample here](https:\u002F\u002Fdocs.gptr.dev\u002Fdocs\u002Fgpt-researcher\u002Fcontext\u002Ftailored-research).\n\n\n## 🤖 MCP Server\n\nWe've moved our MCP server to a dedicated repository: [gptr-mcp](https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgptr-mcp).\n\nThe GPT Researcher MCP Server enables AI applications like Claude to conduct deep research. While LLM apps can access web search tools with MCP, GPT Researcher MCP delivers deeper, more reliable research results.\n\nFeatures:\n- Deep research capabilities for AI assistants\n- Higher quality information with optimized context usage\n- Comprehensive results with better reasoning for LLMs\n- Claude Desktop integration\n\nFor detailed installation and usage instructions, please visit the [official repository](https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgptr-mcp).\n\n\n## 👪 Multi-Agent Assistant\nAs AI evolves from prompt engineering and RAG to multi-agent systems, we're excited to introduce multi-agent assistants built with [LangGraph](https:\u002F\u002Fpython.langchain.com\u002Fv0.1\u002Fdocs\u002Flanggraph\u002F) and [AG2](https:\u002F\u002Fgithub.com\u002Fag2ai\u002Fag2).\n\nBy using multi-agent frameworks, the research process can be significantly improved in depth and quality by leveraging multiple agents with specialized skills. Inspired by the recent [STORM](https:\u002F\u002Farxiv.org\u002Fabs\u002F2402.14207) paper, this project showcases how a team of AI agents can work together to conduct research on a given topic, from planning to publication.\n\nAn average run generates a 5-6 page research report in multiple formats such as PDF, Docx and Markdown.\n\nCheck it out [here](https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Ftree\u002Fmaster\u002Fmulti_agents) or head over to our documentation for [LangGraph](https:\u002F\u002Fdocs.gptr.dev\u002Fdocs\u002Fgpt-researcher\u002Fmulti_agents\u002Flanggraph) and [AG2](https:\u002F\u002Fdocs.gptr.dev\u002Fdocs\u002Fgpt-researcher\u002Fmulti_agents\u002Fag2) for more information.\n\n## 🔍 Observability\n\nGPT Researcher supports **LangSmith** for enhanced tracing and observability, making it easier to debug and optimize complex multi-agent workflows.\n\nTo enable tracing:\n1. Set the following environment variables:\n   ```bash\n   export LANGCHAIN_TRACING_V2=true\n   export LANGCHAIN_API_KEY=your_api_key\n   export LANGCHAIN_PROJECT=\"gpt-researcher\"\n   ```\n2. Run your research tasks as usual. All LangGraph-based agent interactions will be automatically traced and visualized in your LangSmith dashboard.\n\n## 🖥️ Frontend Applications\n\nGPT-Researcher now features an enhanced frontend to improve the user experience and streamline the research process. The frontend offers:\n\n- An intuitive interface for inputting research queries\n- Real-time progress tracking of research tasks\n- Interactive display of research findings\n- Customizable settings for tailored research experiences\n\nTwo deployment options are available:\n1. A lightweight static frontend served by FastAPI\n2. A feature-rich NextJS application for advanced functionality\n\nFor detailed setup instructions and more information about the frontend features, please visit our [documentation page](https:\u002F\u002Fdocs.gptr.dev\u002Fdocs\u002Fgpt-researcher\u002Ffrontend\u002Fintroduction).\n\n## 🚀 Contributing\nWe highly welcome contributions! Please check out [contributing](https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fblob\u002Fmaster\u002FCONTRIBUTING.md) if you're interested.\n\nPlease check out our [roadmap](https:\u002F\u002Ftrello.com\u002Fb\u002F3O7KBePw\u002Fgpt-researcher-roadmap) page and reach out to us via our [Discord community](https:\u002F\u002Fdiscord.gg\u002FQgZXvJAccX) if you're interested in joining our mission.\n\u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fgraphs\u002Fcontributors\">\n  \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fassafelovic_gpt-researcher_readme_0f0231909789.png\" \u002F>\n\u003C\u002Fa>\n## ✉️ Support \u002F Contact us\n- [Community Discord](https:\u002F\u002Fdiscord.gg\u002FspBgZmm3Xe)\n- Author Email: assaf.elovic@gmail.com\n\n## 🛡 Disclaimer\n\nThis project, GPT Researcher, is an experimental application and is provided \"as-is\" without any warranty, express or implied. We are sharing codes for academic purposes under the Apache 2 license. Nothing herein is academic advice, and NOT a recommendation to use in academic or research papers.\n\nOur view on unbiased research claims:\n1. The main goal of GPT Researcher is to reduce incorrect and biased facts. How? We assume that the more sites we scrape the less chances of incorrect data. By scraping multiple sites per research, and choosing the most frequent information, the chances that they are all wrong is extremely low.\n2. We do not aim to eliminate biases; we aim to reduce it as much as possible. **We are here as a community to figure out the most effective human\u002Fllm interactions.**\n3. In research, people also tend towards biases as most have already opinions on the topics they research about. This tool scrapes many opinions and will evenly explain diverse views that a biased person would never have read.\n\n---\n\n\u003Cp align=\"center\">\n\u003Ca href=\"https:\u002F\u002Fstar-history.com\u002F#assafelovic\u002Fgpt-researcher\">\n  \u003Cpicture>\n    \u003Csource media=\"(prefers-color-scheme: dark)\" srcset=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fassafelovic_gpt-researcher_readme_3a005b208f06.png&theme=dark\" \u002F>\n    \u003Csource media=\"(prefers-color-scheme: light)\" srcset=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fassafelovic_gpt-researcher_readme_3a005b208f06.png\" \u002F>\n    \u003Cimg alt=\"Star History Chart\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fassafelovic_gpt-researcher_readme_3a005b208f06.png\" \u002F>\n  \u003C\u002Fpicture>\n\u003C\u002Fa>\n\u003C\u002Fp>\n\n\n\u003Cp align=\"right\">\n  \u003Ca href=\"#top\">⬆️ Back to Top\u003C\u002Fa>\n\u003C\u002Fp>\n","\u003Cdiv align=\"center\" id=\"top\">\n\n\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fassafelovic_gpt-researcher_readme_cbf8f73ce226.png\" alt=\"Logo\" width=\"80\">\n\n####\n\n[![官网](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FOfficial%20Website-gptr.dev-teal?style=for-the-badge&logo=world&logoColor=white&color=0891b2)](https:\u002F\u002Fgptr.dev)\n[![文档](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FDocumentation-DOCS-f472b6?logo=googledocs&logoColor=white&style=for-the-badge)](https:\u002F\u002Fdocs.gptr.dev)\n[![Discord](https:\u002F\u002Fimg.shields.io\u002Fdiscord\u002F1127851779011391548?logo=discord&logoColor=white&label=Discord&color=34b76a&style=for-the-badge)](https:\u002F\u002Fdiscord.gg\u002FQgZXvJAccX)\n\n\n[![PyPI版本](https:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fv\u002Fgpt-researcher?logo=pypi&logoColor=white&style=flat)](https:\u002F\u002Fbadge.fury.io\u002Fpy\u002Fgpt-researcher)\n![GitHub发布](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fv\u002Frelease\u002Fassafelovic\u002Fgpt-researcher?style=flat&logo=github)\n[![在Colab中打开](https:\u002F\u002Fimg.shields.io\u002Fstatic\u002Fv1?message=Open%20in%20Colab&logo=googlecolab&labelColor=grey&color=yellow&label=%20&style=flat&logoSize=40)](https:\u002F\u002Fcolab.research.google.com\u002Fgithub\u002Fassafelovic\u002Fgpt-researcher\u002Fblob\u002Fmaster\u002Fdocs\u002Fdocs\u002Fexamples\u002Fpip-run.ipynb)\n[![Docker镜像版本](https:\u002F\u002Fimg.shields.io\u002Fdocker\u002Fv\u002Felestio\u002Fgpt-researcher\u002Flatest?arch=amd64&style=flat&logo=docker&logoColor=white&color=1D63ED)](https:\u002F\u002Fhub.docker.com\u002Fr\u002Fgptresearcher\u002Fgpt-researcher)\n[![技能](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FClaude%20Skill-skills.sh-blueviolet?style=flat&logo=anthropic&logoColor=white)](https:\u002F\u002Fskills.sh\u002Fassafelovic\u002Fgpt-researcher\u002Fgpt-researcher)\n[![Twitter关注](https:\u002F\u002Fimg.shields.io\u002Ftwitter\u002Ffollow\u002Fassaf_elovic?style=social)](https:\u002F\u002Ftwitter.com\u002Fassaf_elovic)\n\n[English](README.md) | [中文](README-zh_CN.md) | [日本語](README-ja_JP.md) | [한국어](README-ko_KR.md)\n\n\u003C\u002Fdiv>\n\n# 🔎 GPT Researcher\n\n**GPT Researcher 是一款开源的深度研究代理，专为任何任务的网络和本地研究而设计。**\n\n该代理能够生成详细、基于事实且无偏见的研究报告，并附有引用来源。GPT Researcher 提供了全套的自定义选项，以创建量身定制的领域特定研究代理。受近期 [Plan-and-Solve](https:\u002F\u002Farxiv.org\u002Fabs\u002F2305.04091) 和 [RAG](https:\u002F\u002Farxiv.org\u002Fabs\u002F2005.11401) 论文启发，GPT Researcher 通过并行化代理工作，提供稳定的性能和更高的速度，从而解决虚假信息、效率、确定性和可靠性等问题。\n\n**我们的使命是通过人工智能，为个人和组织提供准确、公正、基于事实的信息。**\n\n## 为什么选择 GPT Researcher？\n\n- 对于人工研究而言，得出客观结论可能需要数周时间，耗费大量资源和精力。\n- 基于过时信息训练的语言模型可能会产生幻觉，使其无法满足当前的研究需求。\n- 现有的语言模型存在 token 数量限制，难以生成较长的研究报告。\n- 现有服务中的网络资源有限，容易导致错误信息和浅层结果。\n- 选择性地使用网络资源可能会给研究任务带来偏差。\n\n## 演示\n\u003Ca href=\"https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=f60rlc_QCxE\" target=\"_blank\" rel=\"noopener\">\n  \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fassafelovic_gpt-researcher_readme_f5625647c7b8.png\" alt=\"演示视频\" width=\"800\" target=\"_blank\" \u002F>\n\u003C\u002Fa>\n\n## 作为 Claude Skill 安装\n通过将 GPT Researcher 作为 [Claude Skill](https:\u002F\u002Fskills.sh\u002Fassafelovic\u002Fgpt-researcher\u002Fgpt-researcher) 安装，扩展 Claude 的深度研究能力：\n\n```bash\nnpx skills add assafelovic\u002Fgpt-researcher\n```\n\n安装完成后，Claude 就可以在对话中直接利用 GPT Researcher 的深度研究功能。\n\n## 架构\n核心思想是利用“规划者”和“执行者”两个代理。规划者生成研究问题，而执行者则收集相关信息。随后，发布者会将所有发现汇总成一份全面的研究报告。\n\n\u003Cdiv align=\"center\">\n\u003Cimg align=\"center\" height=\"600\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fassafelovic_gpt-researcher_readme_679ff9ca3bff.png\">\n\u003C\u002Fdiv>\n\n步骤：\n* 根据研究查询创建特定任务的代理。\n* 生成一系列问题，这些问题共同构成对任务的客观看法。\n* 使用爬虫代理为每个问题收集信息。\n* 对每条资源进行总结并追踪来源。\n* 过滤并汇总这些摘要，形成最终的研究报告。\n\n## 教程\n - [工作原理](https:\u002F\u002Fdocs.gptr.dev\u002Fblog\u002Fbuilding-gpt-researcher)\n - [安装方法](https:\u002F\u002Fwww.loom.com\u002Fshare\u002F04ebffb6ed2a4520a27c3e3addcdde20?sid=da1848e8-b1f1-42d1-93c3-5b0b9c3b24ea)\n - [现场演示](https:\u002F\u002Fwww.loom.com\u002Fshare\u002F6a3385db4e8747a1913dd85a7834846f?sid=a740fd5b-2aa3-457e-8fb7-86976f59f9b8)\n\n## 特性\n\n- 📝 利用网络和本地文档生成详细的研究报告。\n- 🖼️ 智能图像抓取与过滤功能，用于研究报告。\n- 🍌 **AI生成的内嵌图片**，使用 Google Gemini（Nano Banana）进行视觉说明。\n- 📜 生成超过2,000字的详细报告。\n- 🌐 聚合超过20个来源，以得出客观结论。\n- 🖥️ 提供轻量级（HTML\u002FCSS\u002FJS）和生产就绪版（NextJS + Tailwind）前端。\n- 🔍 支持JavaScript的网页抓取。\n- 📂 在整个研究过程中保持记忆和上下文。\n- 📄 可将报告导出为PDF、Word等格式。\n\n## 📖 文档\n请参阅 [文档](https:\u002F\u002Fdocs.gptr.dev\u002Fdocs\u002Fgpt-researcher\u002Fgetting-started) 了解：\n- 安装与设置指南\n- 配置与自定义选项\n- 操作示例\n- 完整的API参考\n\n## ⚙️ 开始使用\n\n### 安装\n1. 安装 Python 3.11 或更高版本。[教程](https:\u002F\u002Fwww.tutorialsteacher.com\u002Fpython\u002Finstall-python)。\n2. 克隆项目并进入目录：\n\n    ```bash\n    git clone https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher.git\n    cd gpt-researcher\n    ```\n\n3. 设置 API 密钥，可以通过导出或将其存储在 `.env` 文件中来完成。\n\n    ```bash\n    export OPENAI_API_KEY={您的 OpenAI API 密钥}\n    export TAVILY_API_KEY={您的 Tavily API 密钥}\n    ```\n\n    （可选）为了增强跟踪和可观ability，您还可以设置：\n    \n    ```bash\n    # export LANGCHAIN_TRACING_V2=true\n    # export LANGCHAIN_API_KEY={您的 LangChain API 密钥}\n    ```\n\n    对于自定义的 OpenAI 兼容 API（例如本地模型或其他提供商），您也可以设置：\n    \n    ```bash\n    export OPENAI_BASE_URL={您的自定义 API 基础 URL}\n    ```\n\n4. 安装依赖并启动服务器：\n\n    ```bash\n    pip install -r requirements.txt\n    python -m uvicorn main:app --reload\n    ```\n\n访问 [http:\u002F\u002Flocalhost:8000](http:\u002F\u002Flocalhost:8000) 即可开始使用。\n\n对于其他设置方式（例如 Poetry 或虚拟环境），请查看 [入门页面](https:\u002F\u002Fdocs.gptr.dev\u002Fdocs\u002Fgpt-researcher\u002Fgetting-started)。\n\n## 以 PIP 包运行\n```bash\npip install gpt-researcher\n\n```\n### 示例用法：\n```python\n...\nfrom gpt_researcher import GPTResearcher\n\nquery = \"为什么英伟达股价上涨？\"\nresearcher = GPTResearcher(query=query)\n\n# 对给定的查询进行研究\nresearch_result = await researcher.conduct_research()\n# 撰写报告\nreport = await researcher.write_report()\n...\n```\n\n**更多示例和配置，请参阅 [PIP 文档](https:\u002F\u002Fdocs.gptr.dev\u002Fdocs\u002Fgpt-researcher\u002Fgptr\u002Fpip-package) 页面。**\n\n### 🔧 MCP 客户端\nGPT Researcher 支持 MCP 集成，可连接到 GitHub 仓库、数据库和自定义 API 等专业数据源。这使得研究不仅可以通过网页搜索进行，还可以从这些数据源中获取信息。\n\n```bash\nexport RETRIEVER=tavily,mcp  # 启用混合式网页 + MCP 研究\n```\n\n```python\nfrom gpt_researcher import GPTResearcher\nimport asyncio\nimport os\n\nasync def mcp_research_example():\n    # 启用 MCP 与网页搜索\n    os.environ[\"RETRIEVER\"] = \"tavily,mcp\"\n    \n    researcher = GPTResearcher(\n        query=\"什么是顶级的开源网络研究代理？\",\n        mcp_configs=[\n            {\n                \"name\": \"github\",\n                \"command\": \"npx\",\n                \"args\": [\"-y\", \"@modelcontextprotocol\u002Fserver-github\"],\n                \"env\": {\"GITHUB_TOKEN\": os.getenv(\"GITHUB_TOKEN\")}\n            }\n        ]\n    )\n    \n    research_result = await researcher.conduct_research()\n    report = await researcher.write_report()\n    return report\n```\n\n> 如需全面的 MCP 文档及高级示例，请访问 [MCP 集成指南](https:\u002F\u002Fdocs.gptr.dev\u002Fdocs\u002Fgpt-researcher\u002Fretrievers\u002Fmcp-configs)。\n\n## 🍌 内联图像生成\n\nGPT Researcher 可以使用 Google 的 Gemini 模型（Nano Banana）自动为您的研究报告生成并嵌入 AI 创作的插图。\n\n```bash\n# 在 .env 文件中启用\nIMAGE_GENERATION_ENABLED=true\nGOOGLE_API_KEY=your_google_api_key\nIMAGE_GENERATION_MODEL=models\u002Fgemini-2.5-flash-image\n```\n\n启用后，系统将：\n1. 分析您的研究背景以识别可视化机会\n2. 在研究阶段预先生成 2–3 张相关图像\n3. 在撰写报告时将其内联嵌入\n\n生成的图像采用与 GPT Researcher UI 匹配的暗色模式设计，具有青蓝色点缀的专业信息图美学风格。\n\n[在我们的文档中了解更多关于图像生成的信息](https:\u002F\u002Fdocs.gptr.dev\u002Fdocs\u002Fgpt-researcher\u002Fgptr\u002Fimage_generation)。\n\n## ✨ 深度研究\n\nGPT Researcher 现已包含深度研究功能——一种先进的递归研究工作流，能够以智能体式的深度和广度探索主题。此功能采用树状探索模式，在深入子主题的同时保持对研究主题的全面视角。\n\n- 🌳 可配置深度和广度的树状探索\n- ⚡️ 并发处理以加快结果生成\n- 🤝 跨研究分支的智能上下文管理\n- ⏱️ 每次深度研究约需 5 分钟\n- 💰 每次研究成本约 0.4 美元（使用 `o3-mini` 模型，设置为“高”推理强度）\n\n[在我们的文档中了解更多关于深度研究的信息](https:\u002F\u002Fdocs.gptr.dev\u002Fdocs\u002Fgpt-researcher\u002Fgptr\u002Fdeep_research)。\n\n## 使用 Docker 运行\n\n> **步骤 1** - [安装 Docker](https:\u002F\u002Fdocs.gptr.dev\u002Fdocs\u002Fgpt-researcher\u002Fgetting-started\u002Fgetting-started-with-docker)\n\n> **步骤 2** - 克隆 `.env.example` 文件，将您的 API 密钥添加到克隆后的文件中，并将其保存为 `.env` 文件。\n\n> **步骤 3** - 在 `docker-compose` 文件中，注释掉您不想通过 Docker 运行的服务。\n\n```bash\ndocker-compose up --build\n```\n\n如果上述命令不起作用，请尝试不带破折号运行：\n```bash\ndocker compose up --build\n```\n\n> **步骤 4** - 默认情况下，如果您未取消注释 `docker-compose` 文件中的任何内容，该流程将启动两个进程：\n - 在 localhost:8000 上运行的 Python 服务器\u003Cbr>\n - 在 localhost:3000 上运行的 React 应用程序\u003Cbr>\n\n在任意浏览器中访问 localhost:3000，尽情享受研究的乐趣吧！\n\n## 📄 基于本地文档的研究\n您可以指示 GPT Researcher 根据您的本地文档执行研究任务。目前支持的文件格式包括：PDF、纯文本、CSV、Excel、Markdown、PowerPoint 和 Word 文档。\n\n步骤 1：添加环境变量 `DOC_PATH`，指向存放您文档的文件夹。\n\n```bash\nexport DOC_PATH=\".\u002Fmy-docs\"\n```\n\n步骤 2：\n - 如果您在 localhost:8000 上运行前端应用，只需从“报告来源”下拉选项中选择“我的文档”。\n - 如果您使用 [PIP 包](https:\u002F\u002Fdocs.tavily.com\u002Fguides\u002Fgpt-researcher\u002Fgpt-researcher#pip-package) 运行 GPT Researcher，则在实例化 `GPTResearcher` 类时，将 `report_source` 参数设置为“local” [代码示例在此](https:\u002F\u002Fdocs.gptr.dev\u002Fdocs\u002Fgpt-researcher\u002Fcontext\u002Ftailored-research)。\n\n## 🤖 MCP 服务器\n我们已将 MCP 服务器迁移到专用仓库：[gptr-mcp](https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgptr-mcp)。\n\nGPT Researcher 的 MCP 服务器使像 Claude 这样的 AI 应用程序能够进行深度研究。虽然大型语言模型应用程序可以通过 MCP 访问网页搜索工具，但 GPT Researcher 的 MCP 服务器能够提供更深入、更可靠的研究结果。\n\n特点：\n- 为 AI 助手提供深度研究能力\n- 优化上下文使用，提升信息质量\n- 为大型语言模型提供更全面、更具推理能力的结果\n- 支持 Claude Desktop 集成\n\n有关详细的安装和使用说明，请访问 [官方仓库](https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgptr-mcp)。\n\n## 👪 多智能体助手\n随着 AI 从提示工程和 RAG 发展到多智能体系统，我们很高兴推出基于 [LangGraph](https:\u002F\u002Fpython.langchain.com\u002Fv0.1\u002Fdocs\u002Flanggraph\u002F) 和 [AG2](https:\u002F\u002Fgithub.com\u002Fag2ai\u002Fag2) 构建的多智能体助手。\n\n通过使用多智能体框架，可以利用多个具备专业技能的智能体显著提升研究的深度和质量。受近期 [STORM](https:\u002F\u002Farxiv.org\u002Fabs\u002F2402.14207) 论文启发，该项目展示了由一组 AI 智能体如何协同工作，从规划到发表，共同完成一项研究任务。\n\n一次平均运行可生成 5–6 页的研究报告，并支持 PDF、Docx 和 Markdown 等多种格式。\n\n请在此处查看：[https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Ftree\u002Fmaster\u002Fmulti_agents](https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Ftree\u002Fmaster\u002Fmulti_agents)，或前往我们的文档，了解关于 [LangGraph](https:\u002F\u002Fdocs.gptr.dev\u002Fdocs\u002Fgpt-researcher\u002Fmulti_agents\u002Flanggraph) 和 [AG2](https:\u002F\u002Fdocs.gptr.dev\u002Fdocs\u002Fgpt-researcher\u002Fmulti_agents\u002Fag2) 的更多信息。\n\n## 🔍 可观测性\nGPT Researcher 支持 **LangSmith**，以增强跟踪和可观ability，从而更轻松地调试和优化复杂的多智能体工作流。\n\n要启用跟踪：\n1. 设置以下环境变量：\n   ```bash\n   export LANGCHAIN_TRACING_V2=true\n   export LANGCHAIN_API_KEY=your_api_key\n   export LANGCHAIN_PROJECT=\"gpt-researcher\"\n   ```\n2. 正常运行您的研究任务。所有基于 LangGraph 的智能体交互都将自动被跟踪，并在您的 LangSmith 仪表板上可视化。\n\n## 🖥️ 前端应用\n\nGPT-Researcher 现在配备了增强的前端界面，以提升用户体验并简化研究流程。前端提供了：\n\n- 直观的研究查询输入界面\n- 研究任务的实时进度跟踪\n- 研究成果的交互式展示\n- 可自定义的设置，以满足个性化的研究需求\n\n目前提供两种部署方式：\n1. 由 FastAPI 提供服务的轻量级静态前端\n2. 具备丰富功能的 Next.js 应用程序\n\n如需详细的安装说明及更多关于前端功能的信息，请访问我们的[文档页面](https:\u002F\u002Fdocs.gptr.dev\u002Fdocs\u002Fgpt-researcher\u002Ffrontend\u002Fintroduction)。\n\n## 🚀 贡献\n我们非常欢迎各位的贡献！如果您感兴趣，请查看[贡献指南](https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fblob\u002Fmaster\u002FCONTRIBUTING.md)。\n\n请浏览我们的[路线图](https:\u002F\u002Ftrello.com\u002Fb\u002F3O7KBePw\u002Fgpt-researcher-roadmap)页面，并通过我们的[Discord 社区](https:\u002F\u002Fdiscord.gg\u002FQgZXvJAccX)与我们联系，加入我们的使命吧。\n\u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fgraphs\u002Fcontributors\">\n  \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fassafelovic_gpt-researcher_readme_0f0231909789.png\" \u002F>\n\u003C\u002Fa>\n## ✉️ 支持 \u002F 联系我们\n- [社区 Discord](https:\u002F\u002Fdiscord.gg\u002FspBgZmm3Xe)\n- 作者邮箱：assaf.elovic@gmail.com\n\n## 🛡 免责声明\n\n本项目 GPT Researcher 属于实验性应用，按“原样”提供，不提供任何明示或暗示的担保。我们依据 Apache 2 许可证将代码分享用于学术目的。本文档内容不构成学术建议，亦不建议将其用于学术或研究论文中。\n\n我们对无偏见研究的立场如下：\n1. GPT Researcher 的主要目标是减少错误和有偏见的事实。如何实现？我们认为，抓取的网站越多，出现错误数据的可能性就越小。通过为每次研究抓取多个网站，并选择出现频率最高的信息，这些信息全部出错的概率极低。\n2. 我们并不旨在完全消除偏见，而是力求尽可能地降低其影响。**我们作为一个社区，共同探索最有效的人机协作方式。**\n3. 在研究过程中，人们往往也会带有自身的偏见，因为大多数人在研究特定主题时早已形成了自己的观点。而这款工具会抓取大量不同的观点，并均衡地呈现多元化的看法，这正是那些带有偏见的研究者通常不会接触到的内容。\n\n---\n\n\u003Cp align=\"center\">\n\u003Ca href=\"https:\u002F\u002Fstar-history.com\u002F#assafelovic\u002Fgpt-researcher\">\n  \u003Cpicture>\n    \u003Csource media=\"(prefers-color-scheme: dark)\" srcset=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fassafelovic_gpt-researcher_readme_3a005b208f06.png&theme=dark\" \u002F>\n    \u003Csource media=\"(prefers-color-scheme: light)\" srcset=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fassafelovic_gpt-researcher_readme_3a005b208f06.png\" \u002F>\n    \u003Cimg alt=\"星标历史图表\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fassafelovic_gpt-researcher_readme_3a005b208f06.png\" \u002F>\n  \u003C\u002Fpicture>\n\u003C\u002Fa>\n\u003C\u002Fp>\n\n\n\u003Cp align=\"right\">\n  \u003Ca href=\"#top\">⬆️ 返回顶部\u003C\u002Fa>\n\u003C\u002Fp>","# GPT Researcher 快速上手指南\n\nGPT Researcher 是一款开源的深度研究智能体，专为网络和本地文档研究设计。它能生成详细、客观且带有引用的研究报告，通过并行化代理工作解决信息幻觉、速度慢和可靠性低等问题。\n\n## 环境准备\n\n在开始之前，请确保满足以下系统要求和前置依赖：\n\n*   **操作系统**：Linux, macOS 或 Windows (WSL 推荐)\n*   **Python 版本**：Python 3.11 或更高版本\n*   **API 密钥**：\n    *   **OpenAI API Key** (或其他兼容 OpenAI 格式的模型密钥)\n    *   **Tavily API Key** (用于网络搜索，注册可得免费额度)\n    *   *(可选)* **Google API Key** (如需启用 AI 内联图片生成功能)\n    *   *(可选)* **LangChain API Key** (如需启用追踪和可观测性)\n\n## 安装步骤\n\n### 1. 克隆项目并进入目录\n\n```bash\ngit clone https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher.git\ncd gpt-researcher\n```\n\n### 2. 配置环境变量\n\n你可以选择导出变量或创建 `.env` 文件。推荐创建 `.env` 文件以持久化配置：\n\n```bash\ncp .env.example .env\n```\n\n编辑 `.env` 文件，填入你的密钥：\n\n```bash\nOPENAI_API_KEY=你的_OpenAI_API_Key\nTAVILY_API_KEY=你的_Tavily_API_Key\n\n# (可选) 自定义 API 基础地址 (适用于本地模型或其他提供商)\n# OPENAI_BASE_URL=你的自定义_API_地址\n\n# (可选) 启用图片生成\n# IMAGE_GENERATION_ENABLED=true\n# GOOGLE_API_KEY=你的_Google_API_Key\n\n# (可选) 启用 LangChain 追踪\n# LANGCHAIN_TRACING_V2=true\n# LANGCHAIN_API_KEY=你的_LangChain_API_Key\n```\n\n或者直接导出环境变量：\n\n```bash\nexport OPENAI_API_KEY=你的_OpenAI_API_Key\nexport TAVILY_API_KEY=你的_Tavily_API_Key\n```\n\n### 3. 安装依赖并启动服务\n\n使用 pip 安装依赖：\n\n```bash\npip install -r requirements.txt\n```\n\n启动后端服务器：\n\n```bash\npython -m uvicorn main:app --reload\n```\n\n启动成功后，访问 [http:\u002F\u002Flocalhost:8000](http:\u002F\u002Flocalhost:8000) 即可在浏览器中使用图形界面进行研究。\n\n> **提示**：如果你更倾向于使用 Docker，只需确保已安装 Docker，配置好 `.env` 文件后运行 `docker-compose up --build` 即可同时启动后端和前端服务。\n\n## 基本使用\n\n除了网页版，你也可以将 `gpt-researcher` 作为 Python 包直接在代码中调用。\n\n### 1. 安装 PIP 包\n\n```bash\npip install gpt-researcher\n```\n\n### 2. 代码示例\n\n以下是最简单的使用示例，演示如何发起研究并生成报告：\n\n```python\nfrom gpt_researcher import GPTResearcher\nimport asyncio\n\nasync def get_report(query: str) -> str:\n    # 初始化研究者\n    researcher = GPTResearcher(query=query)\n    \n    # 执行研究过程\n    research_result = await researcher.conduct_research()\n    \n    # 撰写报告\n    report = await researcher.write_report()\n    \n    return report\n\nif __name__ == \"__main__\":\n    query = \"why is Nvidia stock going up?\"\n    report = asyncio.run(get_report(query))\n    print(report)\n```\n\n### 3. 进阶功能简述\n\n*   **本地文档研究**：设置 `DOC_PATH` 环境变量指向你的文档文件夹，并在初始化时指定 `report_source=\"local\"`，即可基于 PDF、Word、Markdown 等本地文件进行研究。\n*   **深度研究 (Deep Research)**：启用树状探索模式，可深入挖掘子主题，适合复杂课题（需在配置中开启相关选项）。\n*   **MCP 集成**：支持连接 GitHub、数据库等专用数据源，实现混合网络与本地数据的深度研究。\n\n更多详细配置和高级用法，请参考官方文档：https:\u002F\u002Fdocs.gptr.dev","某金融科技公司的行业分析师需要在 24 小时内完成一份关于“全球固态电池产业链最新突破与主要玩家”的深度调研报告，以支持投资决策。\n\n### 没有 gpt-researcher 时\n- **信息滞后与幻觉风险**：依赖通用大模型直接生成内容，常因训练数据截止而遗漏近半年的关键技术突破，甚至编造不存在的企业合作案例。\n- **人工检索效率低下**：分析师需手动在数十个新闻源、学术论文库和财报中交叉验证信息，耗时数天仍难以覆盖全面。\n- **观点主观且来源模糊**：人工整理的报告容易受个人搜索习惯影响产生偏差，且难以在文中精确标注每一条数据的原始出处，合规审查困难。\n- **长文逻辑断裂**：受限于 LLM 的上下文窗口，一次性生成的长篇报告往往结构松散，缺乏深度的因果推导和多视角对比。\n\n### 使用 gpt-researcher 后\n- **实时深度挖掘**：gpt-researcher 自主规划搜索路径，并行抓取最新的科研论文、新闻资讯及财报数据，确保内容涵盖截至当下的技术进展，杜绝幻觉。\n- **自动化全流程研究**：工具自动拆解任务为多个子问题，由执行代理并行收集信息并汇总，将原本数天的工作压缩至几分钟内完成初稿。\n- **客观引用与去偏**：生成的报告自带详细的脚注和超链接引用，基于多源交叉验证消除单一信源偏差，满足严格的金融合规要求。\n- **结构化专业输出**：内置的“规划 - 执行 - 发布”架构确保报告逻辑严密，自动生成包含市场格局、技术路线对比及风险提示的专业级长文档。\n\ngpt-researcher 将原本耗时数周的人工深度调研转化为分钟级的自动化流程，为决策者提供即时、准确且可追溯的事实依据。","https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fassafelovic_gpt-researcher_cbf8f73c.png","assafelovic","Assaf Elovic","https:\u002F\u002Foss.gittoolsai.com\u002Favatars\u002Fassafelovic_f3f34492.jpg","Building Tavily and GPT Researcher","Tavily.com",null,"assaf_elovic","www.assafelovic.com","https:\u002F\u002Fgithub.com\u002Fassafelovic",[85,89,93,97,101,105,109,113],{"name":86,"color":87,"percentage":88},"Python","#3572A5",54.9,{"name":90,"color":91,"percentage":92},"TypeScript","#3178c6",27.8,{"name":94,"color":95,"percentage":96},"JavaScript","#f1e05a",7.1,{"name":98,"color":99,"percentage":100},"CSS","#663399",6.3,{"name":102,"color":103,"percentage":104},"HCL","#844FBA",1.9,{"name":106,"color":107,"percentage":108},"HTML","#e34c26",1.8,{"name":110,"color":111,"percentage":112},"Dockerfile","#384d54",0.3,{"name":114,"color":115,"percentage":116},"Procfile","#3B2F63",0,26232,3501,"2026-04-05T00:13:24","Apache-2.0","Linux, macOS, Windows","未说明",{"notes":124,"python":125,"dependencies":126},"该工具主要依赖 API 调用（如 OpenAI, Tavily），而非本地运行大型模型，因此对 GPU 无特殊要求。需配置 OPENAI_API_KEY 和 TAVILY_API_KEY 环境变量。支持 Docker 部署和 PIP 包安装。若启用本地文档研究功能，需设置 DOC_PATH 环境变量。可选配置 LangChain 用于追踪。支持通过 MCP 协议连接 GitHub 等数据源。","3.11+",[127,128,129,130,131,132,133],"uvicorn","langchain","tavily-python","beautifulsoup4","playwright","openai","google-generativeai",[15,54,53,51,13,26,14],[136,137,138,139,140,141,142,143,144,145,146],"ai","python","agent","automation","research","search","webscraping","llms","deepresearch","mcp","mcp-server","2026-03-27T02:49:30.150509","2026-04-06T05:44:13.995537",[150,155,160,165,170,175],{"id":151,"question_zh":152,"answer_zh":153,"source_url":154},16384,"如何在 GPT Researcher 中正确配置 Azure OpenAI 的环境变量？","配置 Azure OpenAI 时，必须同时设置 `OPENAI_API_VERSION` 和 `AZURE_OPENAI_API_VERSION`，否则可能会报错。此外，模型名称需遵循 `provider:model_name` 格式，且部署名称必须与模型名称一致。推荐的环境变量配置如下：\n\n```\nEMBEDDING=\"azure_openai:text-embedding-3-small\"\nAZURE_OPENAI_API_KEY=\u003C你的密钥>\nAZURE_OPENAI_ENDPOINT=\u003C你的端点>\nOPENAI_API_VERSION=2024-02-15-preview\nAZURE_OPENAI_API_VERSION=2024-02-15-preview\nFAST_LLM=azure_openai:gpt-4o-mini\nSMART_LLM=azure_openai:gpt-4o\n```\n\n注意：API 版本号可能需要根据你的 Azure 区域进行调整。","https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fissues\u002F897",{"id":156,"question_zh":157,"answer_zh":158,"source_url":159},16385,"多代理（Multi-Agents）模式下生成的报告内容重复冗余怎么办？","官方已发布针对此问题的改进方案，建议更新到最新版本尝试。如果问题仍然存在，可以通过在提示词中添加更长的描述和明确的章节逻辑来缓解。例如，设置 `follow_guidelines` 并明确每个子章节的范围，虽然审查器默认只针对单个子章节工作，但详细的初始指令有助于减少跨章节的内容重叠。","https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fissues\u002F548",{"id":161,"question_zh":162,"answer_zh":163,"source_url":164},16386,"是否有预构建的 Docker 镜像可以直接拉取使用，而无需本地编译？","是的，现在可以通过 GitHub Container Registry (ghcr.io) 直接拉取预构建的镜像。该功能已通过 PR #1411 合并。用户不再需要手动执行 `docker build`，只需配置好环境变量并运行 `docker compose up` 即可。如果遇到浏览器相关错误，请确保设置 `BROWSER_HEADLESS=true`，并且当 `SCRAPER=browser` 时 Selenium 会自动安装。","https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fissues\u002F1226",{"id":166,"question_zh":167,"answer_zh":168,"source_url":169},16387,"遇到 DuckDuckGo 或 Tavily API 返回 502 Bad Gateway 错误或速率限制异常如何处理？","这通常是由于搜索服务提供商（如 Tavily 或 DuckDuckGo）的临时服务器错误或触发了速率限制。解决方法包括：\n1. 稍后重试，避开高峰时段。\n2. 检查是否使用了免费的 API 额度，考虑升级到付费计划以获得更高的速率限制。\n3. 如果是 Windows 用户遇到 `Proactor event loop` 警告，建议在代码启动前添加以下配置以优化异步处理：\n```python\nimport asyncio\nasyncio.set_event_loop_policy(asyncio.WindowsSelectorEventLoopPolicy())\n```\n4. 确保网络连接稳定，必要时切换网络环境。","https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fissues\u002F478",{"id":171,"question_zh":172,"answer_zh":173,"source_url":174},16388,"程序运行时无响应或卡在\"Thinking about research questions\"步骤怎么办？","如果程序长时间无响应，可能是网络连接问题或 SSL 证书验证失败。建议尝试以下步骤：\n1. 更新 `certifi` 库到最新版本：`pip install --upgrade certifi`。\n2. 如果正在使用 VPN，尝试关闭 VPN 后重新运行。\n3. 检查日志文件（如 `gpt-researcher-logs.txt`），确认是否有具体的连接超时或 SSL 错误信息，以便进一步排查防火墙或代理设置。","https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fissues\u002F90",{"id":176,"question_zh":177,"answer_zh":178,"source_url":154},16389,"为什么设置了 `.env` 文件中的 `AZURE_OPENAI_API_VERSION` 仍然报错找不到 API Key 或版本？","这是因为 GPT Researcher 内部代码与 LangChain 的变量命名习惯存在差异。代码中显式使用了 `AZURE_OPENAI_API_VERSION`，而部分底层依赖可能读取 `OPENAI_API_VERSION`。解决方案是同时在 `.env` 文件中定义这两个变量，并将它们设置为相同的值（例如 `2024-02-15-preview`），以确保所有组件都能正确读取到 API 版本信息。",[180,185,190,195,200,205,210,215,220,225,230,235,240,245,250,255,260,265,270,275],{"id":181,"version":182,"summary_zh":183,"released_at":184},98711,"v3.4.3","重磅发布，包含重大修复和性能优化。欢迎所有新加入的优秀贡献者，并特别感谢 @Br1an67 的重要贡献！\n\n## 变更内容\n* 修复：在 PyMuPDFScraper 中读取所有页面，而不仅仅是第一页，由 @MattBenesch 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1623 中完成。\n* 文档：修复 LLM 文档中无效的 Discord 邀请链接，由 @Br1an67 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1649 中完成。\n* 修复：在运行时安装软件包后失效导入缓存，由 @Br1an67 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1650 中完成。\n* 修复：对于 Azure 嵌入，回退到 OPENAI_API_VERSION，由 @Br1an67 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1651 中完成。\n* 修复：增加 PyMuPDF 下载超时时间，并添加 SSL 回退机制，由 @Br1an67 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1652 中完成。\n* 修复：在深度研究规划中使用所有已配置的检索器，由 @Br1an67 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1653 中完成。\n* 修复：为 FastAPI 前端报告容器添加表格样式，由 @Br1an67 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1654 中完成。\n* 功能：在研究过程中跟踪每一步的成本，由 @Br1an67 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1655 中完成。\n* 功能：为 FastAPI 前端添加抓取次数选项，由 @Br1an67 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1656 中完成。\n* 功能：为 FastAPI 前端添加思考问题部分，由 @Br1an67 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1657 中完成。\n* 修复：如果 ReportGenerator 中缺少代理角色，则动态获取，由 @kiwabb 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1658 中完成。\n* 修复文档链接，并添加 ag2 流程图，由 @qingyun-wu 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1662 中完成。\n* 修复：改进 create_chat_completion 中的重试处理，由 @jhyz 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1664 中完成。\n* 在 MCP 客户端\u002F服务器配置中添加对连接头的支持，由 @GeorgelPreput 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1665 中完成。\n\n## 新贡献者\n* @MattBenesch 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1623 中完成了首次贡献。\n* @Br1an67 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1649 中完成了首次贡献。\n* @kiwabb 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1658 中完成了首次贡献。\n* @qingyun-wu 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1662 中完成了首次贡献。\n* @jhyz 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1664 中完成了首次贡献。\n\n**完整变更日志**：https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fcompare\u002Fv3.4.2...v3.4.3","2026-03-13T14:06:38",{"id":186,"version":187,"summary_zh":188,"released_at":189},98712,"v3.4.2","本次发布带来了大量的性能和质量提升！同时，我们也欢迎几位新加入的优秀贡献者，一如既往地令人欣喜！\n\n## 变更内容\n* 更新自托管实例的 FireCrawl 设置说明，由 @technot80 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1639 中完成。\n* 修复（服务器）：解决多智能体运行 `run_research_task` 时出现的 `NameError` 错误，由 @AlexanderCGO2 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1640 中完成。\n* 性能优化：针对小文档采用智能快速路径优化上下文压缩，由 @maanavagrawal 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1641 中完成。\n* 性能优化：添加 URL 去重功能，以避免重复抓取，由 @maanavagrawal 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1642 中完成。\n* 修复：将 9 处裸 `except` 子句替换为 `except Exception`，由 @haosenwang1018 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1643 中完成。\n* 修复：解决无浏览器模式爬虫的 Docker 兼容性问题及浏览器池死锁问题，由 @mareurs 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1645 中完成。\n* 新特性：新增 Avian 作为 LLM 提供商，由 @avianion 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1646 中完成。\n\n## 新贡献者\n* @AlexanderCGO2 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1640 中完成了首次贡献。\n* @maanavagrawal 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1641 中完成了首次贡献。\n* @haosenwang1018 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1643 中完成了首次贡献。\n* @mareurs 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1645 中完成了首次贡献。\n* @avianion 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1646 中完成了首次贡献。\n\n**完整变更日志**：https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fcompare\u002Fv3.4.1...v3.4.2","2026-03-01T14:20:43",{"id":191,"version":192,"summary_zh":193,"released_at":194},98713,"v3.4.1","重磅发布，包含诸多令人振奋的改进！也欢迎所有新贡献者加入！详情如下：\n\n## 变更内容\n* 更新 server_utils.py - 通过哈希值修复文件名，由 @MaxAkaAltmer 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1611 中完成\n* 新增 bocha 搜索工具，由 @weijintaocode 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1618 中完成\n* 添加 OpenRouter 嵌入支持，由 @Joly0 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1620 中完成\n* Claude\u002F分析项目结构 md q kw，由 @aiiscoding 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1621 中完成\n* 功能：添加 Forge LLM 提供商支持，由 @Yiiii0 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1626 中完成\n* 新增基于 AG2 的示例，由 @VasiliyRad 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1627 中完成\n* 如果运行时未提供配置路径，则尝试从环境变量中加载配置路径，由 @GeorgelPreput 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1630 中完成\n* 修复：在深度研究中处理列表上下文，以防止 AttributeError，由 @technot80 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1637 中完成\n\n## 新贡献者\n* @MaxAkaAltmer 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1611 中完成了首次贡献\n* @weijintaocode 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1618 中完成了首次贡献\n* @Joly0 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1620 中完成了首次贡献\n* @aiiscoding 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1621 中完成了首次贡献\n* @Yiiii0 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1626 中完成了首次贡献\n* @VasiliyRad 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1627 中完成了首次贡献\n* @GeorgelPreput 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1630 中完成了首次贡献\n* @technot80 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1637 中完成了首次贡献\n\n**完整变更日志**：https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fcompare\u002Fv3.4.0...v3.4.1","2026-02-21T07:19:02",{"id":196,"version":197,"summary_zh":198,"released_at":199},98714,"v3.4.0","如此令人兴奋的版本发布，带来了强大的新功能。  \n\n## 🖼️ 内联图像生成  \n\nGPT Researcher 现在可以使用 Google 的 Gemini 模型，自动为您的研究报告生成并嵌入由 AI 创作的插图。  \n\n**核心特性：**  \n- **预生成流程**：图像在研究过程中生成，而非完成后，实现无缝的用户体验  \n- **上下文感知**：LLM 会分析您的研究内容，以识别最佳的可视化时机  \n- **深色模式样式**：生成的图像默认与应用的整体风格保持一致  \n- **完全可配置**：您可以控制最大图像数量、样式（深色\u002F浅色\u002F自动）以及使用的模型  \n\n**快速设置：**  \n```bash  \nIMAGE_GENERATION_ENABLED=true  \nGOOGLE_API_KEY=your_key  \nIMAGE_GENERATION_MODEL=models\u002Fgemini-2.5-flash-image  \nIMAGE_GENERATION_STYLE=dark  \n```  \n\n[📖 图像生成文档](https:\u002F\u002Fdocs.gptr.dev\u002Fdocs\u002Fgpt-researcher\u002Fgptr\u002Fimage_generation)  \n\n---  \n\n## 🔍 LangSmith 集成  \n\n新增对 LangSmith 跟踪与可观测性的原生支持。您可以通过全面了解研究流程，监控研究工作流、调试 LLM 调用，并分析性能。  \n\n**快速设置：**  \n```bash  \nLANGCHAIN_TRACING_V2=true  \nLANGCHAIN_API_KEY=your_langsmith_key  \nLANGCHAIN_PROJECT=gpt-researcher  \n```  \n\n**可追踪的内容：**  \n- LLM 调用及 token 使用情况  \n- 研究计划与子查询生成  \n- 报告生成步骤  \n- 每项研究任务的成本分析  \n\n[📖 LangSmith 日志文档](https:\u002F\u002Fdocs.gptr.dev\u002Fdocs\u002Fgpt-researcher\u002Fhandling-logs\u002Flangsmith-logs)  \n\n---  \n\n## 🤖 Claude Code Skills 集成  \n\n新增了完整的 `.claude\u002Fskills\u002F` 目录，使 Claude Code 能够有效理解、使用并扩展 GPT Researcher。  \n\n**包含内容：**  \n- `SKILL.md`（1,500+ 行）：完整的架构、工作流程、方法签名、数据流图，以及用于添加新功能的 8 步特征模式  \n- `REFERENCE.md`：所有配置变量、API 端点和 WebSocket 事件的快速参考  \n- 包含图像生成实现的真实案例作为参考  \n\n**优势：**  \n- 在 AI 辅助下加速贡献者入职  \n- 各项贡献之间代码模式的一致性  \n- 按照既定模式进行端到端的功能开发  \n\n[📖 AI 辅助开发文档](https:\u002F\u002Fdocs.gptr.dev\u002Fdocs\u002Fgpt-researcher\u002Fgptr\u002Fai-development)  \n\n---  \n\n## 📚 文档更新  \n\n- ✨ 新增：图像生成指南  \n- ✨ 新增：面向 Claude Code 用户的 AI 辅助开发指南  \n- 📝 更新 README，新增功能章节  \n\n---  \n\n## 安装  \n\n```bash  \npip install --upgrade gpt-researcher  \n```  \n\n或使用 Docker：  \n```bash  \ndocker pull gptresearcher\u002Fgpt-researcher:latest  \n```  \n\n---  \n\n**完整变更日志**：[v0.14.5...v0.14.6](https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fcompare\u002Fv0.14.5...v0.14.6)  \n\n## 变更内容  \n* 为 quick_search 添加汇总摘要标志，由 @TheSpaceGod 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1604 中完成  \n* 功能：启用 LangSmith 跟踪","2026-01-29T11:56:28",{"id":201,"version":202,"summary_zh":203,"released_at":204},98715,"v3.3.9","## 变更内容\n* 修复 CORS 默认配置：当 allow_credentials=true 时移除通配符来源，由 @PriscaAmajuoyi 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1595 中完成\n* 文档：为内存和上下文压缩模块添加文档字符串，由 @kga245 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1589 中完成\n* 重构：实现 ECR 和 GitHub Actions 的 Terraform 部署，并添加必要的配置和输出，由 @huytran-gg 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1590 中完成\n* 新特性：实现 WebSocket 聊天命令处理，并使 Tavily 成为可选配置，由 @tiandee 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1593 中完成\n* 将 \u002Fapi\u002Freports 历史记录持久化到本地 JSON 存储（支持 Next.js 历史同步），由 @PriscaAmajuoyi 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1594 中完成\n* 移除硬编码的身份验证令牌，由 @E-G-C 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1596 中完成\n\n## 新贡献者\n* @huytran-gg 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1590 中完成了首次贡献\n* @tiandee 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1593 中完成了首次贡献\n* @E-G-C 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1596 中完成了首次贡献\n\n**完整变更日志**：https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fcompare\u002Fv.3.3.8...v3.3.9","2026-01-25T06:41:51",{"id":206,"version":207,"summary_zh":208,"released_at":209},98716,"v.3.3.8","## 变更内容\n* 添加所有爬虫后端的可配置速率限制，由 @0xsalt 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1550 中实现\n* 修复：Firecrawl 爬虫在调用 'scrape_url' 时出现 AttributeError 错误 - 问题 #1551，由 @0xsalt 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1552 中修复\n* 为 CLI 添加输出格式控制标志，以实现报告的按需生成，由 @0xsalt 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1553 中实现\n* 修复 Railway 部署问题：添加 LangChain 迁移中缺失的依赖项，由 @kga245 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1555 中修复\n* 新特性：改进错误处理，提供更清晰的错误信息和日志记录，由 @Bolajiadesina 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1562 中实现\n* 修复：将域名过滤器 UI 提取为 DomainFilter 组件，由 @PriscaAmajuoyi 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1564 中修复\n* 修复：为 OpenRouter 提供商添加 request_timeout 参数，以防止无限期挂起，由 @sondrealf 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1565 中修复\n* 修复：避免在 multi_agents write_md_to_word 函数中出现重复的 .docx 扩展名，由 @Bolajiadesina 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1567 中修复\n* 杂项：使 multi_agents write_to_file 与后端的 UTF-8 编码处理保持一致，由 @PriscaAmajuoyi 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1568 中完成\n* 修复：从后端模块解析 markdown 转 PDF 的 CSS 路径，由 @Bolajiadesina 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1573 中修复\n\n## 新贡献者\n* @0xsalt 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1550 中完成了首次贡献\n* @Bolajiadesina 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1562 中完成了首次贡献\n* @PriscaAmajuoyi 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1564 中完成了首次贡献\n* @sondrealf 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1565 中完成了首次贡献\n\n**完整变更日志**：https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fcompare\u002Fv.3.3.7...v.3.3.8","2025-12-14T08:08:22",{"id":211,"version":212,"summary_zh":213,"released_at":214},98717,"v.3.3.7","# 发布说明 - LangChain v1 迁移\n\n## 破坏性变更\n\n**现需 Python 3.10+** - LangChain v1 已不再支持 Python 3.9。请升级至 Python 3.10 或更高版本（推荐使用 Python 3.12）。\n\n## 更新内容\n\n### LangChain v1 兼容性\n已升级至 LangChain v1.0+，并更新了整个代码库中的导入路径：\n- `langchain.prompts` → `langchain_core.prompts`\n- `langchain.text_splitter` → `langchain_text_splitters`\n- `langchain.retrievers` → `langchain_classic.retrievers`\n\n### 包版本\n- `langchain>=1.0.0`\n- `langchain-core>=1.0.0`\n- `langchain-community>=0.4.0`\n- `langchain-classic>=1.0.0`（新增）\n\n### 依赖项简化\n通过移除传递性依赖，将 `requirements.txt` 从 134 行减少至 63 行，现在仅列出项目直接使用的包。\n\n## 📝 迁移指南\n\n```bash\n# 使用 Python 3.10+ 安装\npython -m pip install -r requirements.txt\npip install gpt-researcher --upgrade\n```\n\n更多详细信息，请参阅 [LangChain v1 迁移指南](https:\u002F\u002Fdocs.langchain.com\u002Foss\u002Fpython\u002Fmigrate\u002Flangchain-v1)。\n\n## 修复\n- 更新所有文档示例，使用 LangChain v1 的导入语句\n- 清理生产部署的后端依赖项\n\n---\n\n## 本次发布变更内容\n* 修复：将 MCP 配置传播到后续研究人员；传递研究人员 i… 由 @sriramsowmithri9807 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1529 中提出\n* 功能：添加完善的多平台 Docker 构建支持 由 @TheSpaceGod 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1530 中提出\n* 修复：更新 CLI 文档中的仓库 URL 由 @Hnatekmar 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1533 中提出\n* 修复 PubMed 检索器，使其能够正确从 PMC 获取全文。由 @pgrosjean 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1545 中提出\n* LangChain 1.0 迁移 由 @assafelovic 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1547 中提出\n\n## 新贡献者\n* @Hnatekmar 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1533 中完成了首次贡献\n* @pgrosjean 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1545 中完成了首次贡献\n\n**完整变更日志**：https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fcompare\u002Fv.3.3.6...v.3.3.7","2025-11-07T14:10:38",{"id":216,"version":217,"summary_zh":218,"released_at":219},98718,"v.3.3.6","一些紧急修复，以确保 GPTR 能够在提供者无关的情况下顺利运行。如果您发现有任何遗漏，请提交 issue！\n\n## 变更内容\n* 1514：@thepawarrahul 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1516 中添加了 Tavily 的安装要求说明文件。\n* @CezaryKlus 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1522 中修复了前端 Docker 构建问题。\n* @sriramsowmithri9807 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1518 中修复了 MCP 检索器忽略 structured_content 的问题：优先使用结构化数据，不足时再回退到 content #1480（已解决）。\n* 功能：@TheSpaceGod 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1511 中为前端添加了 PWA 功能。\n\n## 新贡献者\n* @thepawarrahul 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1516 中完成了首次贡献。\n* @CezaryKlus 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1522 中完成了首次贡献。\n* @TheSpaceGod 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1511 中完成了首次贡献。\n\n**完整变更日志**：https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fcompare\u002Fv.3.3.5...v.3.3.6","2025-10-07T14:44:05",{"id":221,"version":222,"summary_zh":223,"released_at":224},98719,"v.3.3.5","隆重推出我们最新的 Next.js 应用，为您带来全新的科研体验。现在，您可以与科研成果进行对话、深入研究，直接通过 UI 连接 MCP 等更多功能！\r\n\r\n\u003Ca href=\"https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=f60rlc_QCxE\" target=\"_blank\" rel=\"noopener\">\r\n  \u003Cimg src=\"https:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002Fac2ec55f-b487-4b3f-ae6f-b8743ad296e4\" alt=\"演示视频\" width=\"800\" target=\"_blank\" \u002F>\r\n\u003C\u002Fa>","2025-09-23T07:52:30",{"id":226,"version":227,"summary_zh":228,"released_at":229},98720,"v.3.3.4","## 变更内容\n* 功能 (LLM)：停止向不支持的模型发送温度参数，由 @expphoto 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1485 中实现。\n* 修复 (前端)：使用现有的 gptr-logo.png 作为背景，以避免 404 错误，由 @expphoto 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1484 中完成。\n* 更新 duckduckgo.py 文件，由 @adityasingla71 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1491 中完成。\n* 更新了 max_tokens 的验证逻辑，允许的最大值为 32,000（此前上限为 16,000），由 @hc-tec 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1488 中实现。\n* 功能：为兼容 OpenAI 的 API 添加自定义 OpenAI 基础 URL 的支持，由 @borderlessboy 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1499 中实现。\n* 摘要：修复缺失的 WebSocketDisconnect 导入问题，由 @hasbegun 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1497 中完成。\n* Bug 修复：从 kwargs 中移除 encoding 参数，以避免在 LLM 调用中传递不支持的参数。由 @whaler404 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1494 中完成。\n* 修复文件操作中的关键路径遍历漏洞，由 @saharmor 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1438 中完成。\n* 功能：添加 NetMind 提供商支持，由 @anhuaxiang 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1459 中实现。\n\n## 新贡献者\n* @expphoto 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1485 中完成了首次贡献。\n* @adityasingla71 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1491 中完成了首次贡献。\n* @hc-tec 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1488 中完成了首次贡献。\n* @borderlessboy 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1499 中完成了首次贡献。\n* @hasbegun 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1497 中完成了首次贡献。\n* @whaler404 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1494 中完成了首次贡献。\n* @saharmor 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1438 中完成了首次贡献。\n* @anhuaxiang 在 https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1459 中完成了首次贡献。\n\n**完整变更日志**：https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fcompare\u002Fv.3.3.3...v.3.3.4","2025-09-17T11:12:43",{"id":231,"version":232,"summary_zh":233,"released_at":234},98721,"v.3.3.3","## What's Changed\r\n* GPT-5 Support! Less costs, better performance and larger input context windows\r\n* feat: Add Hallucination Evaluation for GPTResearcher + `judges` library by @julianeagu in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1439\r\n* fix: make exa retriever respect query domains by @mys721tx in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1464\r\n* bug_fix: vectorstore_filter not passed by @viboognesh in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1461\r\n* Update config.md by @sdisaacson in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1465\r\n* feat: Add enhanced Serper API support with country, language, date filters and configurable site exclusion by @hurxxxx in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1472\r\n* Fix scraper FireCrawl markdown error by @hung96ad in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1473\r\n\r\n## New Contributors\r\n* @julianeagu made their first contribution in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1439\r\n* @mys721tx made their first contribution in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1464\r\n* @viboognesh made their first contribution in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1461\r\n* @hung96ad made their first contribution in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1473\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fcompare\u002Fv.3.3.1...v.3.3.3","2025-08-08T05:03:19",{"id":236,"version":237,"summary_zh":238,"released_at":239},98722,"v.3.3.2","## What's Changed\r\n* feat: Add Hallucination Evaluation for GPTResearcher + `judges` library by @julianeagu in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1439\r\n* fix: make exa retriever respect query domains by @mys721tx in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1464\r\n* bug_fix: vectorstore_filter not passed by @viboognesh in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1461\r\n* Update config.md by @sdisaacson in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1465\r\n\r\n## New Contributors\r\n* @julianeagu made their first contribution in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1439\r\n* @mys721tx made their first contribution in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1464\r\n* @viboognesh made their first contribution in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1461\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fcompare\u002Fv.3.3.1...v.3.3.2","2025-07-19T07:20:52",{"id":241,"version":242,"summary_zh":243,"released_at":244},98723,"v.3.3.1","This version focuses on bug fixes and performance improvements. Thank you and welcome to all the new contributors to the community!\r\n\r\n## What's Changed\r\n* fix: Resolve 'ScrapeResponse' object subscriptable error in FireCrawl scraper by @sriramsowmithri9807 in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1432\r\n* remove spurious file (submodule ref) by @grota in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1440\r\n* fix: linux\u002Farm64 machine cannot pull arm64 image from ghcr by @arielweinberger in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1436\r\n* Add `Dockerfile.fullstack` - single container for nextjs\u002Fbackend by @th3w1zard1 in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1433\r\n* fix: langchain dashscope provider out-of-dated by @TRSWNCA in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1455\r\n* Fix zero subtopic error by @Yuxuan1998 in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1454\r\n\r\n## New Contributors\r\n* @sriramsowmithri9807 made their first contribution in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1432\r\n* @grota made their first contribution in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1440\r\n* @arielweinberger made their first contribution in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1436\r\n* @TRSWNCA made their first contribution in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1455\r\n* @Yuxuan1998 made their first contribution in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1454\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fcompare\u002Fv3.3.0...v.3.3.1","2025-07-04T08:51:49",{"id":246,"version":247,"summary_zh":248,"released_at":249},98724,"v3.3.0","# 🔗 MCP Integration Now Live 🥳\r\n\r\nGPT Researcher now supports **Model Context Protocol (MCP)** - connect to specialized data sources alongside web search for comprehensive research.\r\n\r\n## Key Features\r\n- **Two-stage intelligent approach**: Auto-selects relevant tools and generates contextual research\r\n- **Hybrid strategies**: Combine web search with MCP servers (`RETRIEVER=tavily,mcp`)\r\n- **Multi-server support**: GitHub, financial APIs, academic databases, custom tools\r\n- **Zero-config optimization**: Works out-of-the-box with sensible defaults\r\n\r\n## Quick Start\r\n```python\r\nfrom gpt_researcher import GPTResearcher\r\nimport os\r\n\r\n# Enable hybrid research\r\nos.environ[\"RETRIEVER\"] = \"tavily,mcp\"\r\n\r\nresearcher = GPTResearcher(\r\n    query=\"What are the latest React patterns?\",\r\n    mcp_configs=[{\r\n        \"name\": \"github\",\r\n        \"command\": \"npx\",\r\n        \"args\": [\"-y\", \"@modelcontextprotocol\u002Fserver-github\"],\r\n        \"env\": {\"GITHUB_TOKEN\": os.getenv(\"GITHUB_TOKEN\")}\r\n    }]\r\n)\r\n\r\ncontext = await researcher.conduct_research()\r\nreport = await researcher.write_report()\r\n```\r\n\r\n## Use Cases\r\n- **Code research**: GitHub repos, documentation, technical analysis\r\n- **Financial analysis**: Market data, stock trends, business intelligence  \r\n- **Academic research**: ArXiv papers, research databases\r\n- **Enterprise integration**: Internal systems, compliance workflows\r\n\r\n[Full MCP Documentation](https:\u002F\u002Fdocs.gptr.dev\u002Fdocs\u002Fgpt-researcher\u002Fretrievers\u002Fmcp-configs) | [Examples](https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Ftree\u002Fmaster\u002Fexamples)\r\n\r\n## What's Changed\r\n* MCP client-server Integration for GPTR by @assafelovic in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1424\r\n* Fix #1418: OSError: [Errno 36] File name too long with long Chinese prompt by @qylf0000 in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1428\r\n\r\n## New Contributors\r\n* @qylf0000 made their first contribution in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1428\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fcompare\u002Fv3.2.9...v3.3.0","2025-06-08T17:45:48",{"id":251,"version":252,"summary_zh":253,"released_at":254},98725,"v3.2.9","Welcome new contributors to the community! Another massive performance boost to GPT researcher with better reasoning model control, improved static UI, additional integrations and bug fixes!\r\n\r\n## What's Changed\r\n* Add \"Deep Research\" option to selection and fix formatting in base.py by @hide-lgtm in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1394\r\n* fix: improve error handling in SerperSearch results processing by @grapestore in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1388\r\n* Update firecrawl-py requirement from ^1.12.0 to ^2.5.3 by @dependabot in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1382\r\n* Add WebSocket support to FastAPI server by @dreroc in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1400\r\n* Fix return statements in PyMuPDFScraper by @kongzii in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1407\r\n* [Integration] New provider AI\u002FML API Added by @D1m7asis in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1399\r\n* add vllm chat client by @teenaxta in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1409\r\n* Expose **kwargs throughout core modules to unlock LangChain integrations by @teenaxta in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1372\r\n* Massively improve static frontend. by @th3w1zard1 in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1401\r\n* Add `REASONING_EFFORT` to configuration by @KennyDizi in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1416\r\n* ✨ Containerized Multi-Arch Workflows: Production-Ready Docker, and Robust Dependency Management by @th3w1zard1 in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1411\r\n\r\n## New Contributors\r\n* @hide-lgtm made their first contribution in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1394\r\n* @dreroc made their first contribution in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1400\r\n* @kongzii made their first contribution in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1407\r\n* @D1m7asis made their first contribution in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1399\r\n* @teenaxta made their first contribution in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1409\r\n* @th3w1zard1 made their first contribution in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1401\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fcompare\u002Fv3.2.8...v3.2.9","2025-06-01T15:33:49",{"id":256,"version":257,"summary_zh":258,"released_at":259},98726,"v3.2.8","Welcome to all the new contributors who have helped fixing so many issues, adding new amazing features and improving performance!\r\n\r\n## What's Changed\r\n* Fix broken link in llms.md by @robsonos in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1333\r\n* Fix #1323: Ollama model parsing issue in multi-agent mode by @mratsim in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1324\r\n* Add support for o4-mini model in temperature & reasoning-effort config by @arminfabritzek in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1346\r\n* EMBEDDING = '\u003Chugginface>:\u003Csentence-transformers\u002Fall-MiniLM-L6-v2>' by @N4SIRODDIN3 in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1313\r\n* feat: Adds support for gigachat embeddings to the gigachat LLM add-on. by @Rai220 in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1280\r\n* fix: nodriver scraper wait timeout behavior by @ewgdg in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1351\r\n* Strip spaces from \"self.output_dir\" to prevent exception when writing… by @ivan-latunov in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1350\r\n* IBM Granite by @gabe-l-hart in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1278\r\n* Update filtering-by-domain.md by @kicksent in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1370\r\n* Update getting started link to the doc by @irak99 in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1364\r\n* fix firecrawl error: 'ScrapeResponse' object is not subscriptable by @tvoilex in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1363\r\n* Respect my verbosity!  by @tClelford in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1362\r\n* fix python3.11 f-string cannot have backslash by @anexplore in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1352\r\n* bugfix. third argument of ResearchAgent has to be tone by @Masao-Taketani in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1376\r\n* Fix a typo in curator.py by @yuefengz in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1371\r\n\r\n## New Contributors\r\n* @robsonos made their first contribution in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1333\r\n* @mratsim made their first contribution in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1324\r\n* @arminfabritzek made their first contribution in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1346\r\n* @N4SIRODDIN3 made their first contribution in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1313\r\n* @ivan-latunov made their first contribution in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1350\r\n* @gabe-l-hart made their first contribution in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1278\r\n* @kicksent made their first contribution in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1370\r\n* @irak99 made their first contribution in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1364\r\n* @tvoilex made their first contribution in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1363\r\n* @tClelford made their first contribution in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1362\r\n* @anexplore made their first contribution in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1352\r\n* @Masao-Taketani made their first contribution in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1376\r\n* @yuefengz made their first contribution in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1371\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fcompare\u002Fv3.2.7...v3.2.8","2025-05-04T05:27:32",{"id":261,"version":262,"summary_zh":263,"released_at":264},98727,"v3.2.7","We've put A LOT of effort on revamping the entire user experience around research workflows, making it feel like... magic. In addition, we've fixed all performance issues and stabilized the front end package all together. Check out a demo below!\r\n\r\nhttps:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002Fbb6d9bb4-d94b-4f25-bf4e-7edcdd5168bf\r\n\r\n\r\n## What's Changed\r\n* Nextjs fixes by @ElishaKay in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1316\r\n* Feauture\u002Ffrontend improvements by @assafelovic in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1322\r\n\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fcompare\u002Fv3.2.6...v3.2.7","2025-04-07T09:49:44",{"id":266,"version":267,"summary_zh":268,"released_at":269},98728,"v3.2.6","Super exciting release that includes 3 big improvements:\r\n- New MCP server for GPT Researcher! Check out the new repo [here](https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgptr-mcp).\r\n**What makes the MCP server special:**\r\n- 🚀 Seamless integration with Claude Desktop via MCP\r\n- 🔎 Transform basic searches into comprehensive research with a single command\r\n- ✨ Get high-quality, validated information from multiple reliable sources\r\n- 📊 Optimize your context window with only the most relevant content\r\n- 🧠 Enable your AI assistant to reason better with well-structured research\r\n\r\nMCP Server Docs: https:\u002F\u002Fdocs.gptr.dev\u002Fdocs\u002Fgpt-researcher\u002Fmcp-server\u002Fgetting-started\r\n\r\nClaude Demo:\r\nhttps:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002Fef97eea5-a409-42b9-8f6d-b82ab16c52a8\r\n\r\nIn addition:\r\n- You can now add a custom prompt to the write_report method. For example: `report = researcher.write_report(custom_prompt=\"write a short answer based on the research findings\")`\r\n- You can now also run quick web search with your dedicated search engine: `search_results = researcher.quick_search(query=query)`\r\n\r\n\r\n\r\n## What's Changed\r\n* Update to detailed_report.py to include complement_source_urls as a p… by @Laurie2905-JOHN in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1308\r\n* Feature\u002Fcustom report prompt by @assafelovic in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1309\r\n* feat: Add language support to reports and frontend input by @hurxxxx in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1213\r\n* MCP Server by @assafelovic in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1310\r\n* Frontend goodies by @ElishaKay in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1235\r\n\r\n## New Contributors\r\n* @Laurie2905-JOHN made their first contribution in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1308\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fcompare\u002Fv3.2.5...v3.2.6","2025-03-31T07:52:26",{"id":271,"version":272,"summary_zh":273,"released_at":274},98729,"v3.2.5","A lot of improved performance and reliability fixes. Thank you again to the amazing community and welcome to the new contributors!\r\n\r\n## What's Changed\r\n* fix: replace gpt-4o hardcode model in multi-agent with env var. by @galenyip in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1258\r\n* fix: truncating task file names for long queries by @pavlus007 in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1257\r\n* Blog post by @ElishaKay in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1250\r\n* Add missing imports by @rba100 in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1269\r\n* Fixed issue of my-docs folder being created uncessarily #1268 by @gaurav3247 in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1272\r\n* Streamline reasoning_effort use guide across flows by @KennyDizi in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1270\r\n* Add published date and author to Arxiv context; adjust PDF scraper return; adjust detail report prompt for consistent formating style by @akhyarthoriq in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1265\r\n* Added post and get http endpoints to server and updated npm package to allow sending http requests by @gaurav3247 in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1283\r\n* New provider openrouter.ai added by @bjoernh in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1302\r\n* Add User-Agent: Mozilla when downloading a file by @dcieslak19973 in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1297\r\n* Fix AttributeError in PublisherAgent when processing string values by @LavX in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1285\r\n\r\n## New Contributors\r\n* @galenyip made their first contribution in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1258\r\n* @pavlus007 made their first contribution in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1257\r\n* @rba100 made their first contribution in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1269\r\n* @KennyDizi made their first contribution in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1270\r\n* @akhyarthoriq made their first contribution in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1265\r\n* @bjoernh made their first contribution in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1302\r\n* @dcieslak19973 made their first contribution in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1297\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fcompare\u002Fv3.2.4...v3.2.5","2025-03-27T12:47:58",{"id":276,"version":277,"summary_zh":278,"released_at":279},98730,"v3.2.4","Many bug fixes and performance improvements. We've finally added a research history sidebar to the research tasks! Thank you and welcome to all new contributors!\r\n\r\n![Sidebar Demo](https:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002F6a205fa7-02cc-4d22-8066-af5a8d7aaff8)\r\n\r\n## What's Changed\r\n* fix: unblock web socket message queue by @ewgdg in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1220\r\n* perf: load balance nodriver browser scraper tabs by @ewgdg in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1219\r\n* Fix: image relevance filtering by @czakop in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1215\r\n* fix: cancel long running task gracefully by @ewgdg in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1221\r\n* polish npm run build while retaining domain filter setting by @ElishaKay in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1224\r\n* fix: add langchain-openai to project dependencies (to be safe on edge cases) by @milosz-l in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1227\r\n* feature: improve JSON parsing in multi agent setting (#1137) by @czakop in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1230\r\n* fix: resolve failing test in tests\u002Ftest_logging_output.py by @grapestore in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1238\r\n* fix: restrict Python version range in pyproject.toml by @grapestore in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1236\r\n* New Sidebar feature of recent Research History by @assafelovic in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1240\r\n* Fix KeyError: 'deep' in cli.py by @hd365247 in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1245\r\n\r\n## New Contributors\r\n* @milosz-l made their first contribution in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1227\r\n* @grapestore made their first contribution in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1238\r\n* @hd365247 made their first contribution in https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fpull\u002F1245\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fassafelovic\u002Fgpt-researcher\u002Fcompare\u002Fv3.2.3...v3.2.4","2025-03-10T06:48:32"]