--- type: 资源清单 status: 已发布 level: 通用 topic: - Agent - 框架工具 --- # AI Agent 开发者工具箱 > **工欲善其事,必先利其器!精选 Agent 开发必备工具** --- ## 🤖 AI 工具配置(核心) ### 💻 代码开发 - **Cursor** ⭐⭐⭐⭐⭐ 最推荐 - 链接:https://cursor.sh/ - 用途:AI 辅助编程,写 Agent 代码必备 - 特点:Claude 集成,代码补全,AI 问答 - **Claude Code** - 链接:https://claude.ai/ - 用途:复杂代码逻辑设计,架构咨询 - 特点:长上下文,代码理解能力强 ### 📚 论文阅读 - **腾讯元宝** ⭐⭐⭐⭐⭐ 最推荐 - 用途:读论文、总结论文、理解算法原理 - 特点:中文友好,免费 - **豆包** - 用途:快速理解论文核心思想 - 特点:响应快,免费 - **论文翻译工具** - 链接:https://hjfy.top/ - 用途:整篇论文翻译,保留格式 ### 📖 开源项目阅读 - **DeepWiki** ⭐⭐⭐⭐⭐ - 链接:https://opendeep.wiki/ - 用途:理解复杂开源项目(如 LangChain 源码) - 特点:AI 辅助代码理解 - **ZRead.ai** - 链接:https://zread.ai/ - 用途:快速理解 GitHub 项目 ### 💡 Idea 生成与验证 - **GPT-4** / **GPT-o3-mini** - 用途:头脑风暴,技术方案设计 - 配合高质量 Prompt 使用 ### 🎨 可视化工具 - **Excalidraw** ⭐⭐⭐⭐⭐ - 链接:https://excalidraw.com/ - 用途:绘制系统架构图、Agent 工作流图 - 特点:免费,简洁,适合技术图 - **Draw.io** - 链接:https://draw.io/ - 用途:复杂系统架构图 ### ✍️ 文档润色 - **Grok** (X.AI) - 用途:论文润色、降重 - 特点:生成质量高 - **Claude 4.5** - 用途:Paper Review,技术文档审查 - 特点:逻辑严谨 --- ## 📊 数据资源 ### 🔍 找数据集 - **GitHub**:搜索 `awesome-{topic}` 或 `{topic}-dataset` - **Hugging Face Datasets**:https://huggingface.co/datasets - **ModelScope**:https://modelscope.cn/datasets - **OpenLabData**:开放科研数据 - **OpenKG**:中文知识图谱数据 ### 📄 找论文 - **Hugging Face Papers(趋势榜)**:https://huggingface.co/papers/trending - **ArXiv**:https://arxiv.org/ - **OpenReview**:https://openreview.net/ ### 📖 找综述 - **GitHub搜索技巧**:`Awesome + 关键词` - 示例:`Awesome RAG`, `Awesome Agent`, `Awesome LLM` ### 🌍 找行业趋势 - **国外AI趋势**:https://www.theaivalley.com/ - **Bing搜索技巧**:关键词 + `pdf` 找研究报告 --- ## 🛠️ Agent 开发技术栈 ### 基础框架 - **Transformers**:https://github.com/huggingface/transformers - **vLLM**:https://github.com/vllm-project/vllm - **TGI (Text Generation Inference)**:https://github.com/huggingface/text-generation-inference - **LMDeploy**:https://github.com/InternLM/lmdeploy - **SGLang**:https://github.com/sgl-project/sglang ### 训练框架 - **DeepSpeed**:https://github.com/microsoft/DeepSpeed - **FairScale**:https://github.com/facebookresearch/fairscale - **Megatron-LM**:https://github.com/NVIDIA/Megatron-LM - **ColossalAI**:https://github.com/hpcaitech/ColossalAI ### 微调工具 - **Llama-Factory** ⭐⭐⭐⭐⭐:https://github.com/hiyouga/LLaMA-Factory - **Swift**:https://github.com/modelscope/swift - **Axolotl**:https://github.com/OpenAccess-AI-Collective/axolotl - **Unsloth**:https://github.com/unslothai/unsloth - **PEFT**:https://github.com/huggingface/peft ### RAG 技术栈 **RAG 框架**: - **LangChain** ⭐:https://github.com/langchain-ai/langchain - **LlamaIndex** ⭐:https://github.com/run-llama/llama_index - **Haystack**:https://github.com/deepset-ai/haystack - **RAGFlow**:https://github.com/infiniflow/ragflow - **Dify**:https://github.com/langgenius/dify **向量数据库**: - **Milvus** ⭐:https://github.com/milvus-io/milvus - **Weaviate**:https://github.com/weaviate/weaviate - **Pinecone**:https://www.pinecone.io/ - **Chroma**:https://github.com/chroma-core/chroma - **Qdrant**:https://github.com/qdrant/qdrant - **FAISS**:https://github.com/facebookresearch/faiss **Embedding 模型**: - **BGE** ⭐:https://github.com/FlagOpen/FlagEmbedding - **E5**:https://github.com/microsoft/unilm/tree/master/e5 - **Sentence-Transformers**:https://github.com/UKPLab/sentence-transformers - **Instructor**:https://github.com/xlang-ai/instructor-embedding **文档解析**: - **MinerU** ⭐:https://github.com/opendatalab/MinerU - **Unstructured**:https://github.com/Unstructured-IO/unstructured - **LlamaParse**:https://github.com/run-llama/llama_parse - **Docling**:https://github.com/DS4SD/docling - **PyPDF2**:https://github.com/py-pdf/pypdf2 ### Agent 技术栈 **Agent 框架**: - **AutoGPT**:https://github.com/Significant-Gravitas/AutoGPT - **LangGraph** ⭐:https://github.com/langchain-ai/langgraph - **CrewAI**:https://github.com/joaomdmoura/crewAI - **MetaGPT**:https://github.com/geekan/MetaGPT - **AutoGen** ⭐:https://github.com/microsoft/autogen - **Swarm**:https://github.com/openai/swarm **工具调用**: - **ToolBench**:https://github.com/OpenBMB/ToolBench - **Gorilla**:https://github.com/ShishirPatil/gorilla - **ToolLLM**:https://github.com/OpenBMB/ToolLLM **记忆模块**: - **Mem0** ⭐:https://github.com/mem0ai/mem0 - **MemGPT**:https://github.com/cpacker/MemGPT - **LangChain Memory**:内置在 LangChain 中 - **Zep**:https://github.com/getzep/zep **GUI Agent**: - **AppAgent**:https://github.com/mnotgod96/AppAgent - **SeeAct**:https://github.com/OSU-NLP-Group/SeeAct - **WebShop**:https://github.com/princeton-nlp/WebShop - **Mind2Web**:https://github.com/OSU-NLP-Group/Mind2Web ### 多模态技术 **视觉理解**: - **CLIP**:https://github.com/openai/CLIP - **BLIP**:https://github.com/salesforce/BLIP - **LLaVA**:https://github.com/haotian-liu/LLaVA - **Qwen-VL**:https://github.com/QwenLM/Qwen-VL - **InternVL**:https://github.com/OpenGVLab/InternVL **OCR 技术**: - **PaddleOCR** ⭐:https://github.com/PaddlePaddle/PaddleOCR - **EasyOCR**:https://github.com/JaidedAI/EasyOCR - **TrOCR**:https://github.com/microsoft/unilm/tree/master/trocr - **GOT-OCR**:https://github.com/Ucas-HaoranWei/GOT-OCR2.0 - **Rapid_OCR**:https://github.com/RapidAI/RapidOCR ### 部署与监控 **模型服务**: - **FastAPI** ⭐:https://github.com/tiangolo/fastapi - **Gradio**:https://github.com/gradio-app/gradio - **Streamlit**:https://github.com/streamlit/streamlit - **BentoML**:https://github.com/bentoml/BentoML **监控工具**: - **LangSmith** ⭐:https://smith.langchain.com/ - **Weights & Biases**:https://wandb.ai/ - **MLflow**:https://github.com/mlflow/mlflow - **TensorBoard**:https://github.com/tensorflow/tensorboard --- ## 📚 学习资源 ### 🎓 入门教程 - **阿东的大模型实验室**:https://ccn7vpu5l5y8.feishu.cn/wiki/Nh8NwmGpsi0cAykX6OWcHmS3nyc - Agent 学习资料汇总 - 实战项目案例 - 面试题库 - **WayToAGI 小白入门**:https://waytoagi.feishu.cn/wiki/QPe5w5g7UisbEkkow8XcDmOpn8e - 零基础AI入门 - 系统化学习路径 - **吴恩达 LLM 课程(中文版)**:https://github.com/datawhalechina/llm-cookbook - 大模型基础入门 - Datawhale 出品 ### 🎨 宝藏资源库 - **宝藏AI产品知识库**:https://zw73xyquvv.feishu.cn/wiki/V7JxwkK1ti8hjwkqxwAcf0cEngR - **AI开源产品解决方案库**:https://d.aigclink.ai/ ### 📰 技术公众号推荐 - 腾讯技术工程 - 阿里云开发者 - 阿里技术 - 大淘宝技术 --- ## 💡 高效学习工作流 ### 📖 学习 Agent 的最佳实践 ``` Step 1: 读论文理解原理 工具:腾讯元宝 / 豆包 目标:理解 ReAct、Reflexion 等架构 Step 2: 阅读开源项目源码 工具:DeepWiki / ZRead.ai 目标:理解 LangChain、AutoGen 的实现 Step 3: 动手实现项目 工具:Cursor + Claude Code 目标:完成 3 个简历级项目 Step 4: 绘制架构图 工具:Excalidraw / Draw.io 目标:能清晰讲解系统设计 Step 5: 准备面试 工具:AgentGuide 面试题库 目标:掌握高频面试题 ``` --- ## 🎯 不同场景的工具组合 ### 场景1:开发 RAG Agent ``` 数据准备:MinerU(文档解析) 向量化:BGE-Embedding 存储:Milvus 框架:LangChain 监控:LangSmith 部署:FastAPI ``` ### 场景2:开发 Multi-Agent 系统 ``` 框架:AutoGen / CrewAI 编排:LangGraph 记忆:Mem0 工具:各类API 监控:LangSmith 部署:FastAPI + Redis ``` ### 场景3:学习与研究 ``` 读论文:腾讯元宝 读源码:DeepWiki 写代码:Cursor 画图:Excalidraw 做笔记:Feishu / Notion ``` --- ## 📝 数据获取技巧 ### GitHub 搜索技巧 1. **找数据集**:`{topic} dataset` 2. **找综述**:`Awesome {topic}` 3. **找实现**:`{论文名} implementation` 4. **找教程**:`{topic} tutorial Chinese` ### Hugging Face 技巧 1. **找热门论文**:https://huggingface.co/papers/trending 2. **找数据集**:https://huggingface.co/datasets 3. **找模型**:https://huggingface.co/models --- ## 🚀 效率提升建议 ### 养成的好习惯 1. ✅ **做笔记**:每学一个知识点,立即记录 2. ✅ **写博客**:学完就写,加深理解 3. ✅ **写代码**:理论学完立即实践 4. ✅ **问AI**:遇到问题先问元宝/ChatGPT ### 知识平权时代的三大能力 1. **搜索能力**:快速找到高质量资源 2. **动手能力**:快速实现和验证想法 3. **思考能力**:不盲从,有自己的判断 > **核心理念**:选择大于努力!选对工具,效率翻倍! --- ## 🔗 相关资源 - [Agent 开发框架推荐](./agent/frameworks.md) - [必读论文清单](./31篇LLM的经典论文速读版清单.md) - [优质博客与教程](./learning-resources.md) - [开源项目推荐](../projects/README.md)