# AI 工具与应用 面向内容、搜索、知识、生产力、多模态和垂直场景的 AI 应用与工具。 数量:82 | 名称 | 概括 | 标签 | 链接 | | --- | --- | --- | --- | | A2UI - Google 出品的 Agent-to-UI 框架,将 AI Agent 输出转化为交互式用户界面 | Google 出品的 Agent-to-UI 框架,将 AI Agent 输出转化为交互式用户界面 | multi-agent, agent, framework, 构建Agent, 需要开发 | [打开](https://github.com/google/A2UI) | | ace-step-ui - 开源Suno替代品,ACE-Step AI音乐生成专业UI | 开源Suno替代品,ACE-Step AI音乐生成专业UI | open-source, 拿来就用 | [打开](https://github.com/fspecii/ace-step-ui) | | agent-lens - AI 编程 Agent 使用情况可视化仪表盘,追踪 Token 成本和工具调用 | AI 编程 Agent 使用情况可视化仪表盘,追踪 Token 成本和工具调用 | coding, multi-agent, tool, agent, 构建Agent, 部署运维 | [打开](https://github.com/naimjeem/agent-lens) | | agent-skills - 生产级 AI 编程 Agent 技能/提示词模式集合,30K+ Star | 生产级 AI 编程 Agent 技能/提示词模式集合,30K+ Star | coding, multi-agent, agent, 构建Agent, 学习参考 | [打开](https://github.com/addyosmani/agent-skills) | | agent-skills-eval - AI Agent 技能测试运行器 | AI Agent 技能测试运行器 | multi-agent, agent, 构建Agent | [打开](https://github.com/darkrishabh/agent-skills-eval) | | agentic-inbox - Cloudflare 出品的自托管邮件客户端,内置 AI Agent,运行在 Workers 上 | Cloudflare 出品的自托管邮件客户端,内置 AI Agent,运行在 Workers 上 | multi-agent, agent, 构建Agent, 自动化 | [打开](https://github.com/cloudflare/agentic-inbox) | | agentmemory - 基于真实基准测试的 AI 编程 Agent 持久化记忆方案,排名第一 | 基于真实基准测试的 AI 编程 Agent 持久化记忆方案,排名第一 | coding, multi-agent, agent, 构建Agent | [打开](https://github.com/rohitg00/agentmemory) | | ai-cli - Vercel 出品终端 AI 生成工具 | Vercel 出品终端 AI 生成工具 | tool, 拿来就用 | [打开](https://github.com/vercel-labs/ai-cli) | | AI-Novel-Writing-Assistant - 面向长篇小说的 AI Native 开源系统,集成 Agent、世界观引擎、写作引擎、RAG | 面向长篇小说的 AI Native 开源系统,集成 Agent、世界观引擎、写作引擎、RAG。 | open-source, rag, multi-agent, agent, 构建Agent, 需要开发 | [打开](https://github.com/ExplosiveCoderflome/AI-Novel-Writing-Assistant) | | airllm - 单张 4GB GPU 推理 70B 大模型,极低显存推理方案 | 单张 4GB GPU 推理 70B 大模型,极低显存推理方案 | llm, 需要开发, 部署运维 | [打开](https://github.com/lyogavin/airllm) | | AiToEarn - 用 AI 赚钱!AI 变现方法与实践合集 | 用 AI 赚钱!AI 变现方法与实践合集 | ai, 直接变现, 学习参考 | [打开](https://github.com/yikart/AiToEarn) | | android-reverse-engineering-skill - Android 逆向工程 AI 技能 | Android 逆向工程 AI 技能。 | ai, 构建Agent, 学习参考 | [打开](https://github.com/SimoneAvogadro/android-reverse-engineering-skill) | | AngelSlim - 腾讯的 AI 模型压缩/加速框架 | 腾讯的 AI 模型压缩/加速框架。 | framework, 需要开发, 部署运维 | [打开](https://github.com/Tencent/AngelSlim) | | autoresearch - Karpathy 的 AI 研究自动化 Agent,单 GPU 运行 | Karpathy 的 AI 研究自动化 Agent,单 GPU 运行 | agent, multi-agent, 构建Agent, 自动化 | [打开](https://github.com/karpathy/autoresearch) | | BettaFish - 多智能体舆情分析助手(微舆),预测趋势 | 多智能体舆情分析助手(微舆),预测趋势 | ai, 构建Agent | [打开](https://github.com/666ghj/BettaFish) | | byterover-cli - AI 编程 Agent 便携记忆层 | AI 编程 Agent 便携记忆层 | coding, multi-agent, agent, 构建Agent | [打开](https://github.com/campfirein/byterover-cli) | | cherry-studio - AI生产力工作室,集成智能对话、自主Agent和300+助手 | AI生产力工作室,集成智能对话、自主Agent和300+助手 | agent, multi-agent, 拿来就用 | [打开](https://github.com/CherryHQ/cherry-studio) | | cocoindex - 面向长时间运行智能体的增量数据引擎 | 面向长时间运行智能体的增量数据引擎 | ai, 构建Agent, 需要开发 | [打开](https://github.com/cocoindex-io/cocoindex) | | codex-seo - Codex-first SEO Agent 技能套件,26 Agent + 24 TOML 工作流 | Codex-first SEO Agent 技能套件,26 Agent + 24 TOML 工作流 | coding, multi-agent, agent, 构建Agent, 自动化 | [打开](https://github.com/AgriciDaniel/codex-seo) | | compliant-llm - AI Agent 合规审计工具 | AI Agent 合规审计工具 | agent, multi-agent, tool, 构建Agent, 部署运维 | [打开](https://github.com/fiddlecube/compliant-llm) | | CopyWeb.ai - AI内容生成器 | AI驱动的内容生成平台,用于营销文案、文章和创意写作 | ai, copywriting, content-generation, marketing, automation, web-tool | - | | crawl4ai - 面向LLM的开源友好网页爬虫与数据提取工具 | 面向LLM的开源友好网页爬虫与数据提取工具 | tool, llm, open-source, 自动化, 需要开发 | [打开](https://github.com/unclecode/crawl4ai) | | DeepSeek-TUI - DeepSeek模型的终端编码Agent,TUI界面,在终端中直接使用AI编程 | DeepSeek模型的终端编码Agent,TUI界面,在终端中直接使用AI编程 | coding, agent, multi-agent, 拿来就用, 构建Agent | [打开](https://github.com/Hmbown/DeepSeek-TUI) | | DeepTutor - 深度学习辅导工具 | 深度学习辅导工具。 | tool, 学习参考 | [打开](https://github.com/HKUDS/DeepTutor) | | dflash - DFlash: 基于块扩散的Flash推测解码,加速LLM推理 | DFlash: 基于块扩散的Flash推测解码,加速LLM推理 | llm, 需要开发, 部署运维 | [打开](https://github.com/z-lab/dflash) | | firecrawl - 为 AI 提供网页搜索、抓取和交互的 API 服务 | 为 AI 提供网页搜索、抓取和交互的 API 服务 | ai, 自动化, 需要开发 | [打开](https://github.com/firecrawl/firecrawl) | | formative-memory - 受心理学启发的 AI Agent 记忆系统,包含联想、遗忘和综合功能 | 受心理学启发的 AI Agent 记忆系统,包含联想、遗忘和综合功能 | multi-agent, agent, 构建Agent | [打开](https://github.com/jarimustonen/formative-memory) | | free-claude-code - 免费使用 Claude Code 的工具,支持终端、VSCode 扩展和 Discord 等多种方式 | 免费使用 Claude Code 的工具,支持终端、VSCode 扩展和 Discord 等多种方式。 | tool, coding, 拿来就用 | [打开](https://github.com/Alishahryar1/free-claude-code) | | free-llm-api-resources - 免费 LLM API 资源列表 | 免费 LLM API 资源列表 | llm, 学习参考 | [打开](https://github.com/cheahjs/free-llm-api-resources) | | freemocap - 开源免费动作捕捉系统 | 开源免费动作捕捉系统 | open-source, 拿来就用 | [打开](https://github.com/freemocap/freemocap) | | GhostTrack - OSINT/安全追踪工具 | OSINT/安全追踪工具。 | tool, 自动化 | [打开](https://github.com/HunxByts/GhostTrack) | | Google Gemini全栈LangGraph快速入门 | Google构建Gemini和LangGraph全栈应用的快速入门模板 | google, gemini, fullstack, langgraph, quickstart, ai | - | | goose-skills - 125个开源 GTM Skills,让 Claude Code/Codex/Cursor 等 AI 编码 agent 自动化销售和营销任务 | 125个开源 GTM Skills,让 Claude Code/Codex/Cursor 等 AI 编码 agent 自动化销售和营销任务 | agent, open-source, multi-agent, 直接变现, 自动化 | [打开](https://github.com/athina-ai/goose-skills) | | hyperframes - 用 HTML 编写、渲染为视频的工具,专为 Agent 设计 | 用 HTML 编写、渲染为视频的工具,专为 Agent 设计 | multi-agent, agent, tool, 构建Agent, 需要开发 | [打开](https://github.com/heygen-com/hyperframes) | | Kotaemon - 文档问答系统 | 基于RAG的高级文档问答系统,支持多模态 | ai, knowledge, management, rag, document, qa | - | | langextract - Google 开源的语言提取工具 | Google 开源的语言提取工具。 | open-source, tool, 需要开发 | [打开](https://github.com/google/langextract) | | langflow - 构建和部署 AI Agent 和工作流的强大工具平台 | 构建和部署 AI Agent 和工作流的强大工具平台 | tool, agent, multi-agent, 构建Agent, 需要开发 | [打开](https://github.com/langflow-ai/langflow) | | LibreChat - 增强版ChatGPT克隆,支持多智能体、MCP、多AI提供商集成 | 增强版ChatGPT克隆,支持多智能体、MCP、多AI提供商集成 | mcp, 拿来就用 | [打开](https://github.com/danny-avila/LibreChat) | | local-deep-research - 本地深度研究工具,支持 Ollama/Google/Anthropic 等多种 LLM,搜索 arXiv/PubMed 等多源 | 本地深度研究工具,支持 Ollama/Google/Anthropic 等多种 LLM,搜索 arXiv/PubMed 等多源 | tool, llm, 拿来就用, 学习参考 | [打开](https://github.com/LearningCircuit/local-deep-research) | | LTX-2 - LTX-2音视频生成模型推理和LoRA训练 | LTX-2音视频生成模型推理和LoRA训练 | ai, 需要开发 | [打开](https://github.com/Lightricks/LTX-2) | | magic - Magicrew — 开源一体化 AI 生产力平台(Agent + 工作流 + IM + 协作) | Magicrew — 开源一体化 AI 生产力平台(Agent + 工作流 + IM + 协作) | agent, open-source, multi-agent, 构建Agent, 拿来就用 | [打开](https://github.com/dtyq/magic) | | maigret - OSINT 工具,通过用户名在大量网站中收集信息 | OSINT 工具,通过用户名在大量网站中收集信息。 | tool, 自动化, 拿来就用 | [打开](https://github.com/soxoj/maigret) | | manifest - Agent智能模型路由,降低成本最高70%,GitHub Trending项目 | Agent智能模型路由,降低成本最高70%,GitHub Trending项目 | agent, multi-agent, 构建Agent, 部署运维 | [打开](https://github.com/mnfst/manifest) | | MASPO - MASPO: LLM多智能体系统联合提示优化框架,ICML 2026接收 | MASPO: LLM多智能体系统联合提示优化框架,ICML 2026接收 | framework, llm, 需要开发, 学习参考 | [打开](https://github.com/wangzx1219/MASPO) | | MaxKB - 企业知识库 | 具有RAG功能的企业级AI知识库系统 | knowledge-base, ai, enterprise, rag, chinese | - | | mcp-memory-service - AI Agent 持久记忆服务,支持 LangGraph/CrewAI/AutoGen | AI Agent 持久记忆服务,支持 LangGraph/CrewAI/AutoGen | multi-agent, agent, 构建Agent | [打开](https://github.com/doobidoo/mcp-memory-service) | | memoirs - 基于 SQLite 的本地 AI Agent 长期记忆引擎,支持 MCP/HTTP/CLI | 基于 SQLite 的本地 AI Agent 长期记忆引擎,支持 MCP/HTTP/CLI | mcp, multi-agent, agent, 构建Agent | [打开](https://github.com/misaelzapata/memoirs) | | memvid - AI Agent 记忆层,用单文件替代复杂 RAG 管道 | AI Agent 记忆层,用单文件替代复杂 RAG 管道 | agent, rag, multi-agent, 构建Agent | [打开](https://github.com/memvid/memvid) | | metorial - MCP 服务器部署平台,类似 Vercel for MCP | MCP 服务器部署平台,类似 Vercel for MCP | mcp, 部署运维, 需要开发 | [打开](https://github.com/metorial/metorial) | | nanobrowser - 开源 Chrome 扩展,AI 驱动的多 Agent 网页自动化,BYOK 模式 | 开源 Chrome 扩展,AI 驱动的多 Agent 网页自动化,BYOK 模式 | agent, multi-agent, open-source, 自动化, 拿来就用 | [打开](https://github.com/nanobrowser/nanobrowser) | | nexent - 模型引擎框架 | 模型引擎框架。 | framework, 需要开发, 部署运维 | [打开](https://github.com/ModelEngine-Group/nexent) | | nocobase - 开源 AI + 无代码平台,快速构建业务系统 | 开源 AI + 无代码平台,快速构建业务系统 | open-source, 拿来就用 | [打开](https://github.com/nocobase/nocobase) | | omi - 开源 AI 可穿戴设备项目 | 开源 AI 可穿戴设备项目。 | open-source, 需要开发 | [打开](https://github.com/BasedHardware/omi) | | omlx - 面向 Apple Silicon 的 LLM 推理服务器,支持连续批处理与 SSD 缓存,macOS 菜单栏管理 | 面向 Apple Silicon 的 LLM 推理服务器,支持连续批处理与 SSD 缓存,macOS 菜单栏管理 | llm, 部署运维, 拿来就用 | [打开](https://github.com/jundot/omlx) | | onyx - 开源AI聊天平台,支持所有LLM | 开源AI聊天平台,支持所有LLM | open-source, llm, 拿来就用 | [打开](https://github.com/onyx-dot-app/onyx) | | open-webui - 开源 Web UI for LLMs,支持 Ollama、OpenAI 兼容 API 等 | 开源 Web UI for LLMs,支持 Ollama、OpenAI 兼容 API 等。 | open-source, llm, 拿来就用 | [打开](https://github.com/open-webui/open-webui) | | openpets - AI 编码 Agent 的桌面宠物,MCP 连接 | AI 编码 Agent 的桌面宠物,MCP 连接 | multi-agent, mcp, agent, 构建Agent, 拿来就用 | [打开](https://github.com/alvinunreal/openpets) | | Parlant - 对话AI框架 | 用于构建复杂对话系统的高级对话AI框架 | ai, conversation, nlp, chatbot, dialogue | - | | Personal_AI_Infrastructure - 增强人类能力的 Agentic AI 基础设施 | 增强人类能力的 Agentic AI 基础设施 | multi-agent, agent, 构建Agent, 学习参考 | [打开](https://github.com/danielmiessler/Personal_AI_Infrastructure) | | pi - AI Agent 工具包:编程 Agent CLI、统一 LLM API、TUI/Web UI 库、Slack Bot、vLLM 部署 | AI Agent 工具包:编程 Agent CLI、统一 LLM API、TUI/Web UI 库、Slack Bot、vLLM 部署 | multi-agent, llm, tool, agent, coding, 构建Agent, 部署运维 | [打开](https://github.com/earendil-works/pi) | | Pixelle-Video - AI 视频生成项目 | AI 视频生成项目。 | ai, 拿来就用 | [打开](https://github.com/AIDC-AI/Pixelle-Video) | | plur - AI Agent 本地优先、零成本的持久化记忆系统 | AI Agent 本地优先、零成本的持久化记忆系统 | multi-agent, agent, 构建Agent | [打开](https://github.com/plur-ai/plur) | | pocket-tts - 可在CPU上运行的轻量级TTS语音合成 | 可在CPU上运行的轻量级TTS语音合成 | ai, 需要开发 | [打开](https://github.com/kyutai-labs/pocket-tts) | | posthog - 开源产品分析平台,支持事件追踪、用户行为分析、功能开关等 | 开源产品分析平台,支持事件追踪、用户行为分析、功能开关等。 | open-source, 拿来就用, 部署运维 | [打开](https://github.com/PostHog/posthog) | | ppt-master - AI 从任何文档生成原生可编辑 PPTX | AI 从任何文档生成原生可编辑 PPTX | ai, 拿来就用 | [打开](https://github.com/hugohe3/ppt-master) | | Rapid-MLX - Apple Silicon最快本地AI推理引擎,比Ollama快4.2倍 | Apple Silicon最快本地AI推理引擎,比Ollama快4.2倍 | ai, 拿来就用, 部署运维 | [打开](https://github.com/raullenchai/Rapid-MLX) | | rf-detr - Roboflow 的实时目标检测模型 | Roboflow 的实时目标检测模型。 | ai, 需要开发 | [打开](https://github.com/roboflow/rf-detr) | | sglang - 高性能 LLM 和多模态模型推理服务框架 | 高性能 LLM 和多模态模型推理服务框架 | llm, framework, 需要开发, 部署运维 | [打开](https://github.com/sgl-project/sglang) | | Shadow - AI Agent 提示词变更调试工具,追踪导致 Agent 失败的 prompt 修改 | AI Agent 提示词变更调试工具,追踪导致 Agent 失败的 prompt 修改 | multi-agent, tool, agent, 构建Agent, 部署运维 | [打开](https://github.com/manav8498/Shadow) | | symbiopulse - MCP 优先的 AI 编程 Agent 上下文层,持久化项目记忆 | MCP 优先的 AI 编程 Agent 上下文层,持久化项目记忆 | mcp, coding, multi-agent, agent, 构建Agent | [打开](https://github.com/lzk-gh/symbiopulse) | | TabPFN - 表格数据基础模型 | 表格数据基础模型 | ai, 需要开发 | [打开](https://github.com/PriorLabs/TabPFN) | | thunderbolt - 强调用户自主控制的 AI 平台,自选模型、数据自主、消除供应商锁定 | 强调用户自主控制的 AI 平台,自选模型、数据自主、消除供应商锁定。 | ai, 拿来就用 | [打开](https://github.com/thunderbird/thunderbolt) | | TrendRadar - AI驱动舆情监控与热点筛选工具,支持多平台聚合 | AI驱动舆情监控与热点筛选工具,支持多平台聚合 | tool, 自动化, 拿来就用 | [打开](https://github.com/sansan0/TrendRadar) | | unsloth - 本地训练/运行开源模型(Gemma 4、Qwen3.5、DeepSeek 等),新增 Web UI,降低本地训练门槛 | 本地训练/运行开源模型(Gemma 4、Qwen3.5、DeepSeek 等),新增 Web UI,降低本地训练门槛。 | open-source, 学习参考, 拿来就用 | [打开](https://github.com/unslothai/unsloth) | | vellum-assistant - 个人 AI 助手,能随用户成长,具备记忆、人格、主动联系能力,跨 macOS/Telegram/Slack | 个人 AI 助手,能随用户成长,具备记忆、人格、主动联系能力,跨 macOS/Telegram/Slack | ai, 拿来就用 | [打开](https://github.com/vellum-ai/vellum-assistant) | | vexa - 开源会议转录API,支持Google Meet/Teams/Zoom | 开源会议转录API,支持Google Meet/Teams/Zoom | open-source, 拿来就用, 自动化 | [打开](https://github.com/Vexa-ai/vexa) | | VibeVoice - 统一语音转文本模型,单次处理 60 分钟长音频,自动生成结构化转录,支持 50+ 语言 | 统一语音转文本模型,单次处理 60 分钟长音频,自动生成结构化转录,支持 50+ 语言。 | ai, 需要开发 | [打开](https://github.com/microsoft/VibeVoice) | | ViMax - Agentic 视频生成框架,集成导演、编剧、制片和视频生成于一体 | Agentic 视频生成框架,集成导演、编剧、制片和视频生成于一体 | multi-agent, agent, framework, 构建Agent, 需要开发 | [打开](https://github.com/HKUDS/ViMax) | | voicebox - 开源语音合成工作室,用于语音生成和编辑 | 开源语音合成工作室,用于语音生成和编辑。 | open-source, 拿来就用 | [打开](https://github.com/jamiepine/voicebox) | | VoxCPM - 无 Tokenizer 的多语言语音生成 TTS 模型,支持创意声音设计和语音克隆 | 无 Tokenizer 的多语言语音生成 TTS 模型,支持创意声音设计和语音克隆 | ai, 需要开发 | [打开](https://github.com/OpenBMB/VoxCPM) | | Weknora - AI知识管理 | AI驱动的知识管理和研究协作平台 | knowledge-management, ai, research, collaboration, productivity | - | | worldmonitor - 世界监控工具,AI 驱动的信息聚合 | 世界监控工具,AI 驱动的信息聚合。 | tool, 自动化, 拿来就用 | [打开](https://github.com/koala73/worldmonitor) |