[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"similar-screenpipe--screenpipe":3,"tool-screenpipe--screenpipe":62},[4,18,26,36,46,54],{"id":5,"name":6,"github_repo":7,"description_zh":8,"stars":9,"difficulty_score":10,"last_commit_at":11,"category_tags":12,"status":17},4358,"openclaw","openclaw\u002Fopenclaw","OpenClaw 是一款专为个人打造的本地化 AI 助手，旨在让你在自己的设备上拥有完全可控的智能伙伴。它打破了传统 AI 助手局限于特定网页或应用的束缚，能够直接接入你日常使用的各类通讯渠道，包括微信、WhatsApp、Telegram、Discord、iMessage 等数十种平台。无论你在哪个聊天软件中发送消息，OpenClaw 都能即时响应，甚至支持在 macOS、iOS 和 Android 设备上进行语音交互，并提供实时的画布渲染功能供你操控。\n\n这款工具主要解决了用户对数据隐私、响应速度以及“始终在线”体验的需求。通过将 AI 部署在本地，用户无需依赖云端服务即可享受快速、私密的智能辅助，真正实现了“你的数据，你做主”。其独特的技术亮点在于强大的网关架构，将控制平面与核心助手分离，确保跨平台通信的流畅性与扩展性。\n\nOpenClaw 非常适合希望构建个性化工作流的技术爱好者、开发者，以及注重隐私保护且不愿被单一生态绑定的普通用户。只要具备基础的终端操作能力（支持 macOS、Linux 及 Windows WSL2），即可通过简单的命令行引导完成部署。如果你渴望拥有一个懂你",349277,3,"2026-04-06T06:32:30",[13,14,15,16],"Agent","开发框架","图像","数据工具","ready",{"id":19,"name":20,"github_repo":21,"description_zh":22,"stars":23,"difficulty_score":10,"last_commit_at":24,"category_tags":25,"status":17},3808,"stable-diffusion-webui","AUTOMATIC1111\u002Fstable-diffusion-webui","stable-diffusion-webui 是一个基于 Gradio 构建的网页版操作界面，旨在让用户能够轻松地在本地运行和使用强大的 Stable Diffusion 图像生成模型。它解决了原始模型依赖命令行、操作门槛高且功能分散的痛点，将复杂的 AI 绘图流程整合进一个直观易用的图形化平台。\n\n无论是希望快速上手的普通创作者、需要精细控制画面细节的设计师，还是想要深入探索模型潜力的开发者与研究人员，都能从中获益。其核心亮点在于极高的功能丰富度：不仅支持文生图、图生图、局部重绘（Inpainting）和外绘（Outpainting）等基础模式，还独创了注意力机制调整、提示词矩阵、负向提示词以及“高清修复”等高级功能。此外，它内置了 GFPGAN 和 CodeFormer 等人脸修复工具，支持多种神经网络放大算法，并允许用户通过插件系统无限扩展能力。即使是显存有限的设备，stable-diffusion-webui 也提供了相应的优化选项，让高质量的 AI 艺术创作变得触手可及。",162132,"2026-04-05T11:01:52",[14,15,13],{"id":27,"name":28,"github_repo":29,"description_zh":30,"stars":31,"difficulty_score":32,"last_commit_at":33,"category_tags":34,"status":17},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",160015,2,"2026-04-18T11:30:52",[14,13,35],"语言模型",{"id":37,"name":38,"github_repo":39,"description_zh":40,"stars":41,"difficulty_score":42,"last_commit_at":43,"category_tags":44,"status":17},8272,"opencode","anomalyco\u002Fopencode","OpenCode 是一款开源的 AI 编程助手（Coding Agent），旨在像一位智能搭档一样融入您的开发流程。它不仅仅是一个代码补全插件，而是一个能够理解项目上下文、自主规划任务并执行复杂编码操作的智能体。无论是生成全新功能、重构现有代码，还是排查难以定位的 Bug，OpenCode 都能通过自然语言交互高效完成，显著减少开发者在重复性劳动和上下文切换上的时间消耗。\n\n这款工具专为软件开发者、工程师及技术研究人员设计，特别适合希望利用大模型能力来提升编码效率、加速原型开发或处理遗留代码维护的专业人群。其核心亮点在于完全开源的架构，这意味着用户可以审查代码逻辑、自定义行为策略，甚至私有化部署以保障数据安全，彻底打破了传统闭源 AI 助手的“黑盒”限制。\n\n在技术体验上，OpenCode 提供了灵活的终端界面（Terminal UI）和正在测试中的桌面应用程序，支持 macOS、Windows 及 Linux 全平台。它兼容多种包管理工具，安装便捷，并能无缝集成到现有的开发环境中。无论您是追求极致控制权的资深极客，还是渴望提升产出的独立开发者，OpenCode 都提供了一个透明、可信",144296,1,"2026-04-16T14:50:03",[13,45],"插件",{"id":47,"name":48,"github_repo":49,"description_zh":50,"stars":51,"difficulty_score":32,"last_commit_at":52,"category_tags":53,"status":17},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。",109154,"2026-04-18T11:18:24",[14,15,13],{"id":55,"name":56,"github_repo":57,"description_zh":58,"stars":59,"difficulty_score":32,"last_commit_at":60,"category_tags":61,"status":17},6121,"gemini-cli","google-gemini\u002Fgemini-cli","gemini-cli 是一款由谷歌推出的开源 AI 命令行工具，它将强大的 Gemini 大模型能力直接集成到用户的终端环境中。对于习惯在命令行工作的开发者而言，它提供了一条从输入提示词到获取模型响应的最短路径，无需切换窗口即可享受智能辅助。\n\n这款工具主要解决了开发过程中频繁上下文切换的痛点，让用户能在熟悉的终端界面内直接完成代码理解、生成、调试以及自动化运维任务。无论是查询大型代码库、根据草图生成应用，还是执行复杂的 Git 操作，gemini-cli 都能通过自然语言指令高效处理。\n\n它特别适合广大软件工程师、DevOps 人员及技术研究人员使用。其核心亮点包括支持高达 100 万 token 的超长上下文窗口，具备出色的逻辑推理能力；内置 Google 搜索、文件操作及 Shell 命令执行等实用工具；更独特的是，它支持 MCP（模型上下文协议），允许用户灵活扩展自定义集成，连接如图像生成等外部能力。此外，个人谷歌账号即可享受免费的额度支持，且项目基于 Apache 2.0 协议完全开源，是提升终端工作效率的理想助手。",100752,"2026-04-10T01:20:03",[45,13,15,14],{"id":63,"github_repo":64,"name":65,"description_en":66,"description_zh":67,"ai_summary_zh":67,"readme_en":68,"readme_zh":69,"quickstart_zh":70,"use_case_zh":71,"hero_image_url":72,"owner_login":65,"owner_name":65,"owner_avatar_url":73,"owner_bio":74,"owner_company":75,"owner_location":75,"owner_email":76,"owner_twitter":75,"owner_website":77,"owner_url":78,"languages":79,"stars":117,"forks":118,"last_commit_at":119,"license":120,"difficulty_score":32,"env_os":121,"env_gpu":122,"env_ram":123,"env_deps":124,"category_tags":131,"github_topics":133,"view_count":32,"oss_zip_url":75,"oss_zip_packed_at":75,"status":17,"created_at":143,"updated_at":144,"faqs":145,"releases":179},9137,"screenpipe\u002Fscreenpipe","screenpipe","Run agents that work for you based on what you do. AI finally knows what you are doing","screenpipe 是一款开源的本地 AI 记忆工具，旨在将你的电脑屏幕和音频转化为可搜索、可自动化的个人知识库。它通过持续记录用户的操作轨迹（包括屏幕内容、声音、键盘输入及应用切换），利用 AI 技术让用户能够像回忆往事一样，用自然语言轻松检索“过去几分钟看到了什么”或“总结今天的对话”，甚至基于当前工作状态自动触发智能代理（Agents）执行任务。\n\n这一工具主要解决了数字时代信息过载与记忆碎片化的痛点，帮助用户不再遗漏重要的灵感、会议细节或操作步骤，同时打破了传统自动化脚本需要手动设定的局限，实现了基于上下文的智能响应。\n\nscreenpipe 特别适合注重隐私的开发者、研究人员、设计师以及希望提升个人效率的普通用户。其核心技术亮点在于\"100% 本地运行”：所有数据均存储于用户设备，绝不上传云端，确保了绝对的隐私安全。此外，它具备极低的资源占用（仅 5-10% CPU），支持完整的无障碍树捕获与 OCR 识别，并允许用户灵活过滤敏感信息（如密码）。作为 Rewind.ai 的开源替代方案，screenpipe 让每个人都能拥有真正属于自己的、安全且强大的第二大脑。","\n\u003Cp align=\"center\">\n   \u003Ca href =\"https:\u002F\u002Fscreenpi.pe\">\n      \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fscreenpipe_screenpipe_readme_b540f3c24678.png\" alt=\"logo\" width=\"200\">\n   \u003C\u002Fa>\n\u003C\u002Fp>\n\n\u003Ch1 align=\"center\">[ screenpipe ]\u003C\u002Fh1>\n\n\n\u003Cp align=\"center\">AI memory for your screen\u003C\u002Fp>\n\u003Cp align=\"center\">run agents that work for you in the background based on what you do\u003C\u002Fp>\n\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fscreenpi.pe\u002Fonboarding\" target=\"_blank\">\n    \u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fdownload-desktop%20app-black?style=for-the-badge\" alt=\"download\">\n  \u003C\u002Fa>\n\u003C\u002Fp>\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fdiscord.gg\u002Fscreenpipe\">\n    \u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fdiscord\u002F823813159592001537?style=for-the-badge&logo=discord&logoColor=white\" alt=\"discord\">\n  \u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Ftwitter.com\u002Fscreenpipe\">\n    \u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Ftwitter\u002Ffollow\u002Fscreenpipe?style=for-the-badge&logo=x&logoColor=white&label=follow\" alt=\"twitter\">\n  \u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fwww.youtube.com\u002F@screenpipe\">\n    \u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fyoutube\u002Fchannel\u002Fsubscribers\u002FUCwjkpAsb70_mENKvy7hT5bw?style=for-the-badge&logo=youtube&logoColor=white&label=subscribers\" alt=\"youtube\">\n  \u003C\u002Fa>\n\u003C\u002Fp>\n\n![image](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fscreenpipe_screenpipe_readme_ac32e80c994c.png)\n![image](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fscreenpipe_screenpipe_readme_d470edb34371.png)\n\n\n---\n\n## what is this?\n\nscreenpipe turns your computer into a personal AI that knows everything you've done. record. search. automate. all local, all private, all yours\n\n```\n┌─────────────────────────────────────────┐\n│  screen + audio → local storage → ai   │\n└─────────────────────────────────────────┘\n```\n\n- **remember everything** - never forget what you saw, heard, or did\n- **run agents that work based on what you do** - pipes are agents triggered by your work activity\n\n\u003Cimg width=\"360\" height=\"311\" alt=\"image\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fscreenpipe_screenpipe_readme_be5ca8d3bc63.png\" \u002F>\n\n- **search with ai** - find anything using natural language\n- **100% local** - your data never leaves your machine\n- **open source** - inspect, modify, own\n\n\u003Cp align=\"center\">\n   \u003Ca href =\"https:\u002F\u002Fscreenpi.pe\">\n      \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fscreenpipe_screenpipe_readme_8ca9606e8e6b.png\">\n   \u003C\u002Fa>\n\u003C\u002Fp>\n\n\n## install\n\n[download the desktop app](https:\u002F\u002Fscreenpi.pe\u002Fonboarding) — one-time purchase, all features, auto-updates\n\nor run the CLI:\n\n```bash\nnpx screenpipe@latest record\n```\n\nthen \n\n```bash\nclaude mcp add screenpipe -- npx -y screenpipe-mcp\n```\n\nthen ask claude `what did i see in the last 5 mins?` or `summarize today conversations` or `create a pipe that updates linear every time i work on task X`\n\n\n## specs\n\n- captures full accessibility tree, OCR as fallback, transcription, speakers, keyboard inputs, app switches\n- 5-10% cpu usage\n- 0.5-3gb ram\n- ~20gb storage\u002Fmonth\n- filters (window, app, chrome extensions, passwords, PII)\n- optional encryption at rest\n- works offline\n\n---\n\n\u003Cp align=\"center\">\n    \u003Ca href=\"https:\u002F\u002Fdocs.screenpi.pe\">docs\u003C\u002Fa> ·\n    \u003Ca href=\"https:\u002F\u002Fdiscord.gg\u002Fscreenpipe\">discord\u003C\u002Fa> ·\n    \u003Ca href=\"https:\u002F\u002Ftwitter.com\u002Fscreenpipe\">x\u003C\u002Fa> ·\n    \u003Ca href=\"https:\u002F\u002Fwww.youtube.com\u002F@screenpipe\">youtube\u003C\u002Fa> ·\n    \u003Ca href=\"https:\u002F\u002Fwww.reddit.com\u002Fr\u002Fscreen_pipe\">reddit\u003C\u002Fa>\n\u003C\u002Fp>\n\n## Contributing\n\nSee [CONTRIBUTING.md](CONTRIBUTING.md) for guidelines, maintainers, and how to submit PRs. AI\u002Fvibe-coded PRs welcome!\n\nThanks to all contributors:\n\n\u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fgraphs\u002Fcontributors\">\n  \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fscreenpipe_screenpipe_readme_13474b9452cb.png\" \u002F>\n\u003C\u002Fa>\n\n\u003Cdetails>\n\u003Csummary>📖 LLM reference\u003C\u002Fsummary>\n\n## About screenpipe\n\nscreenpipe is an open source application (MIT license) that continuously captures your screen and audio, creating a searchable, AI-powered memory of everything you do on your computer. All data is stored locally on your device. It is the leading open source alternative to Rewind.ai (now Limitless), Microsoft Recall, Granola, and Otter.ai. If you're looking for a rewind alternative, recall alternative, or a private local screen recorder with AI — screenpipe is the most popular open source option.\n\n- **Website**: https:\u002F\u002Fscreenpi.pe\n- **Documentation**: https:\u002F\u002Fdocs.screenpi.pe\n- **Discord**: https:\u002F\u002Fdiscord.gg\u002Fscreenpipe\n- **License**: MIT\n\n## Who screenpipe is for\n\n- **Knowledge workers** who want to recall anything they've seen or heard on their computer\n- **Developers** who want to give AI coding assistants (Cursor, Claude Code, Cline, Continue) context about what they're working on\n- **Researchers** who need to search through large volumes of screen-based information\n- **People with ADHD** who frequently lose track of tabs, documents, and conversations\n- **Remote workers** who want automatic meeting transcription and notes\n- **Teams & enterprises** who want to deploy AI across their organization with deterministic data permissions and central config management ([screenpi.pe\u002Fteam](https:\u002F\u002Fscreenpi.pe\u002Fteam))\n- **Anyone** who wants a private, local-first alternative to cloud-based AI memory tools\n\n## Platform support\n\n| Platform | Support | Installation |\n|----------|---------|-------------|\n| macOS (Apple Silicon) | ✅ Full support | Native .dmg installer |\n| macOS (Intel) | ✅ Full support | Native .dmg installer |\n| Windows 10\u002F11 | ✅ Full support | Native .exe installer |\n| Linux | ✅ Supported | Build from source |\n\nMinimum requirements: 8 GB RAM recommended. ~5–10 GB disk space per month. CPU usage typically 5–10% on modern hardware thanks to event-driven capture.\n\n## Core features\n\n### Event-driven screen capture\nInstead of recording every second, screenpipe listens for meaningful events — app switches, clicks, typing pauses, scrolling — and captures a screenshot only when something actually changes. Each capture pairs a screenshot with the accessibility tree (the structured text the OS already knows about: buttons, labels, text fields). If accessibility data isn't available (e.g. remote desktops, games), it falls back to OCR. This gives you maximum data quality with minimal CPU and storage — no more processing thousands of identical frames.\n\n### Audio transcription\nCaptures system audio (what you hear) and microphone input (what you say). Real-time speech-to-text using OpenAI Whisper running locally on your device. Speaker identification and diarization. Works with any audio source — Zoom, Google Meet, Teams, or any other application.\n\n### AI-powered search\nNatural language search across all OCR text and audio transcriptions. Filter by application name, window title, browser URL, date range. Semantic search using embeddings. Returns screenshots and audio clips alongside text results.\n\n### Timeline view\nVisual timeline of your entire screen history. Scroll through your day like a DVR. Click any moment to see the full screenshot and extracted text. Play back audio from any time period.\n\n### Plugin system (Pipes)\nPipes are scheduled AI agents defined as markdown files. Each pipe is a `pipe.md` with a prompt and schedule — screenpipe runs an AI coding agent (like pi or claude-code) that queries your screen data, calls APIs, writes files, and takes actions. Built-in pipes include:\n- **Obsidian sync**: Automatically sync screen activity to Obsidian vault as daily logs\n- **Reminders**: Scan activity for todos and create Apple Reminders (macOS)\n- **Idea tracker**: Surface startup ideas from your browsing + market trends\n\nDevelopers can create pipes by writing a markdown file in `~\u002F.screenpipe\u002Fpipes\u002F`.\n\n#### Pipe data permissions\nEach pipe supports YAML frontmatter fields that give admins deterministic, OS-level control over what data AI agents can access:\n- **App & window filtering**: `allow-apps`, `deny-apps`, `deny-windows` (glob patterns)\n- **Content type control**: restrict to `ocr`, `audio`, `input`, or `accessibility`\n- **Time & day restrictions**: e.g. `time-range: 09:00-18:00`, `days: Mon,Tue,Wed,Thu,Fri`\n- **Endpoint gating**: `allow-raw-sql: false`, `allow-frames: false`\n\nEnforced at three layers — skill gating (AI never learns denied endpoints), agent interception (blocked before execution), and server middleware (per-pipe cryptographic tokens). Not prompt-based. Deterministic.\n\n### MCP server (Model Context Protocol)\nscreenpipe runs as an MCP server, allowing AI assistants to query your screen history:\n- Works with Claude Desktop, Cursor, VS Code (Cline, Continue), and any MCP-compatible client\n- AI assistants can search your screen history, get recent context, and access meeting transcriptions\n- Zero configuration: `claude mcp add screenpipe -- npx -y screenpipe-mcp`\n\n### Developer API\nFull REST API running on localhost (default port 3030). Endpoints for searching screen content, audio, frames. Raw SQL access to the underlying SQLite database. JavaScript\u002FTypeScript SDK available.\n\n### Apple Intelligence integration (macOS)\nOn supported Macs, screenpipe uses Apple Intelligence for on-device AI processing — daily summaries, action items, and reminders with zero cloud dependency and zero cost.\n\n## Privacy and security\n\n- **100% local by default**: All data stored on your device in a local SQLite database. Nothing sent to external servers.\n- **Open source**: MIT licensed, fully auditable codebase.\n- **Local AI support**: Use Ollama or any local model — no data sent to any cloud.\n- **No account required**: Core application works without any sign-up.\n- **You own your data**: Export, delete, or back up at any time.\n- **Optional encrypted sync**: End-to-end encrypted sync between devices (zero-knowledge encryption).\n- **AI data permissions**: Per-pipe YAML-based access control — deterministic enforcement at the OS level, not prompt-based. Three enforcement layers prevent AI agents from accessing unauthorized data.\n\n## How screenpipe compares to alternatives\n\n| Feature | screenpipe | Rewind \u002F Limitless | Microsoft Recall | Granola |\n|---------|-----------|-------------------|-----------------|---------|\n| Open source | ✅ MIT license | ❌ | ❌ | ❌ |\n| Platforms | macOS, Windows, Linux | macOS, Windows | Windows only | macOS only |\n| Data storage | 100% local | Cloud required | Local (Windows) | Cloud |\n| Multi-monitor | ✅ All monitors | ❌ Active window only | ✅ | ❌ Meetings only |\n| Audio transcription | ✅ Local Whisper | ✅ | ❌ | ✅ Cloud |\n| Developer API | ✅ Full REST API + SDK | Limited | ❌ | ❌ |\n| Plugin system | ✅ Pipes (AI agents) | ❌ | ❌ | ❌ |\n| AI model choice | Any (local or cloud) | Proprietary | Microsoft AI | Proprietary |\n| Team deployment | ✅ Central config, AI permissions | ❌ | ❌ | ❌ |\n| Pricing | One-time purchase | Subscription | Bundled with Windows | Subscription |\n\n## Pricing\n\n- **Lifetime**: $400 one-time purchase. All features, all future updates, forever.\n- **Lifetime + Pro 1 year**: $600 one-time. Includes lifetime app + 1 year of Pro (cloud sync, priority support).\n- **Pro subscription**: $39\u002Fmonth for cloud sync between devices, priority support, and pro AI models.\n- **Teams**: Custom pricing. Shared configs, shared pipes, per-pipe AI data permissions, admin dashboard, MDM ready (Intune \u002F SCCM). See [screenpi.pe\u002Fteam](https:\u002F\u002Fscreenpi.pe\u002Fteam).\n\n## Integrations\n\n- **AI coding assistants**: Cursor, Claude Code, Cline, Continue, OpenCode, Gemini CLI\n- **AI chat assistants**: ChatGPT (via MCP), Claude Desktop (via MCP), any MCP-compatible client\n- **Note-taking**: Obsidian, Notion\n- **Local AI**: Ollama, any OpenAI-compatible model server\n- **Automation**: Custom pipes (scheduled AI agents as markdown files)\n\n## Teams & enterprise\n\nscreenpipe Teams lets organizations deploy AI agents across their team with full control over what AI can access. See [screenpi.pe\u002Fteam](https:\u002F\u002Fscreenpi.pe\u002Fteam).\n\n- **Central config management**: Push capture settings (app filters, schedules, URL rules) to every device from an admin dashboard.\n- **Shared pipes**: Deploy AI workflows (auto-standups, meeting-to-tickets, time tracking) team-wide.\n- **Per-pipe AI data permissions**: YAML frontmatter controls what each pipe can access — apps, windows, content types, time ranges, endpoints. Enforced deterministically at the OS level via three layers (skill gating, agent interception, server middleware with per-pipe cryptographic tokens).\n- **Privacy boundary**: Admins control what gets captured and what AI accesses. They never see the actual data — everything stays on each employee's device.\n- **Override rules**: Employees can add stricter filters (e.g. also block personal email) but cannot weaken admin-set rules.\n- **MDM ready**: Deploy via Intune, SCCM, Robopack, or any MDM solution.\n- **Enterprise**: SSO\u002FSAML, audit logs, SLA, SOC 2 \u002F HIPAA compliance ready.\n\n## Technical architecture\n\n1. **Event-driven capture**: Listens for OS events (app switch, click, typing pause, scroll, clipboard). When something meaningful happens, captures a screenshot + accessibility tree together with the same timestamp. Falls back to OCR when accessibility data isn't available. Idle fallback captures periodically when nothing is happening.\n2. **Audio processing**: Whisper (local) or Deepgram (cloud) for speech-to-text. Speaker identification and diarization.\n3. **Storage**: Local SQLite with FTS5 full-text search. Screenshots saved as JPEGs on disk (~300 MB\u002F8hr vs ~2 GB with continuous recording).\n4. **API layer**: REST API on localhost:3030. Search, frames, audio, elements, health, pipe management.\n5. **Plugin layer**: Pipes — scheduled AI agents as markdown files. Agent executes prompts with access to screenpipe API.\n6. **UI layer**: Desktop app built with Tauri (Rust + TypeScript).\n\n## API examples\n\nSearch screen content:\n```\nGET http:\u002F\u002Flocalhost:3030\u002Fsearch?q=meeting+notes&content_type=ocr&limit=10\n```\n\nSearch audio transcriptions:\n```\nGET http:\u002F\u002Flocalhost:3030\u002Fsearch?q=budget+discussion&content_type=audio&limit=10\n```\n\nJavaScript SDK:\n```javascript\nimport { pipe } from \"@screenpipe\u002Fjs\";\n\nconst results = await pipe.queryScreenpipe({\n  q: \"project deadline\",\n  contentType: \"all\",\n  limit: 20,\n  startTime: new Date(Date.now() - 24 * 60 * 60 * 1000).toISOString(),\n});\n```\n\n## Frequently asked questions\n\n**Is screenpipe free?**\nThe core engine is open source (MIT license). The desktop app is a one-time lifetime purchase ($400). No recurring subscription required for the core app.\n\n**Does screenpipe send my data to the cloud?**\nNo. All data is stored locally by default. You can use fully local AI models via Ollama for complete privacy.\n\n**How much disk space does it use?**\n~5–10 GB per month. Event-driven capture only stores frames when something changes, dramatically reducing storage compared to continuous recording.\n\n**Does it slow down my computer?**\nTypical CPU usage is 5–10% on modern hardware. Event-driven capture only processes frames when something changes, and accessibility tree extraction is much lighter than OCR.\n\n**Can I use it with ChatGPT\u002FClaude\u002FCursor?**\nYes. screenpipe runs as an MCP server, allowing Claude Desktop, Cursor, and other AI assistants to directly query your screen history.\n\n**Can it record multiple monitors?**\nYes. screenpipe captures all connected monitors simultaneously.\n\n**How does text extraction work?**\nscreenpipe primarily uses the OS accessibility tree to get structured text (buttons, labels, text fields) — this is faster and more accurate than OCR. When accessibility data isn't available (remote desktops, games, some Linux apps), it falls back to OCR: Apple Vision on macOS, Windows native OCR, or Tesseract on Linux.\n\n**Can I deploy screenpipe to my team?**\nYes. Screenpipe Teams provides central config management, shared AI pipes, and per-pipe data permissions. Admins control what gets captured and what AI can access — employees' actual data never leaves their devices. See [screenpi.pe\u002Fteam](https:\u002F\u002Fscreenpi.pe\u002Fteam).\n\n**How do AI data permissions work?**\nEach pipe supports YAML frontmatter fields (allow-apps, deny-apps, deny-windows, allow-content-types, time-range, days, allow-raw-sql, allow-frames) that deterministically control what data the AI agent can access. Enforcement happens at three OS-level layers — not by prompting the AI to behave. Even a compromised agent cannot access denied data.\n\n## Company\n\nBuilt by screenpipe (Mediar, Inc.). Founded 2024. Based in San Francisco, CA.\n\n- Founder: Louis Beaumont (@louis030195)\n- Twitter: @screenpipe\n- Email: louis@screenpi.pe\n\n\u003C\u002Fdetails>\n","\u003Cp align=\"center\">\n   \u003Ca href =\"https:\u002F\u002Fscreenpi.pe\">\n      \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fscreenpipe_screenpipe_readme_b540f3c24678.png\" alt=\"logo\" width=\"200\">\n   \u003C\u002Fa>\n\u003C\u002Fp>\n\n\u003Ch1 align=\"center\">[ screenpipe ]\u003C\u002Fh1>\n\n\n\u003Cp align=\"center\">你屏幕的AI记忆\u003C\u002Fp>\n\u003Cp align=\"center\">根据你的操作，在后台运行为你服务的智能代理\u003C\u002Fp>\n\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fscreenpi.pe\u002Fonboarding\" target=\"_blank\">\n    \u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fdownload-desktop%20app-black?style=for-the-badge\" alt=\"download\">\n  \u003C\u002Fa>\n\u003C\u002Fp>\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fdiscord.gg\u002Fscreenpipe\">\n    \u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fdiscord\u002F823813159592001537?style=for-the-badge&logo=discord&logoColor=white\" alt=\"discord\">\n  \u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Ftwitter.com\u002Fscreenpipe\">\n    \u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Ftwitter\u002Ffollow\u002Fscreenpipe?style=for-the-badge&logo=x&logoColor=white&label=follow\" alt=\"twitter\">\n  \u003C\u002FA>\n  \u003Ca href=\"https:\u002F\u002Fwww.youtube.com\u002F@screenpipe\">\n    \u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fyoutube\u002Fchannel\u002Fsubscribers\u002FUCwjkpAsb70_mENKvy7hT5bw?style=for-the-badge&logo=youtube&logoColor=white&label=subscribers\" alt=\"youtube\">\n  \u003C\u002Fa>\n\u003C\u002Fp>\n\n![image](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fscreenpipe_screenpipe_readme_ac32e80c994c.png)\n![image](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fscreenpipe_screenpipe_readme_d470edb34371.png)\n\n\n---\n\n## 这是什么？\n\nscreenpipe 将你的电脑变成一个了解你所有操作的个人AI。记录、搜索、自动化——全部本地化、完全私密、专属于你。\n\n```\n┌─────────────────────────────────────────┐\n│  屏幕 + 音频 → 本地存储 → AI   │\n└─────────────────────────────────────────┘\n```\n\n- **记住一切** - 再也不会忘记你看到、听到或做过的事情\n- **根据你的操作运行智能代理** - 管道是基于你的工作活动触发的智能代理\n\n\u003Cimg width=\"360\" height=\"311\" alt=\"image\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fscreenpipe_screenpipe_readme_be5ca8d3bc63.png\" \u002F>\n\n- **用AI搜索** - 使用自然语言查找任何内容\n- **100% 本地化** - 你的数据永远不会离开你的设备\n- **开源** - 可以检查、修改、完全掌控\n\n\u003Cp align=\"center\">\n   \u003Ca href =\"https:\u002F\u002Fscreenpi.pe\">\n      \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fscreenpipe_screenpipe_readme_8ca9606e8e6b.png\">\n   \u003C\u002Fa>\n\u003C\u002Fp>\n\n\n## 安装\n\n[下载桌面应用](https:\u002F\u002Fscreenpi.pe\u002Fonboarding) — 一次性购买，功能齐全，自动更新\n\n或者运行命令行：\n\n```bash\nnpx screenpipe@latest record\n```\n\n然后\n\n```bash\nclaude mcp add screenpipe -- npx -y screenpipe-mcp\n```\n\n接着你可以问 Claude：“我最近5分钟看到了什么？”、“总结一下今天的对话” 或者“创建一个管道，每当我处理任务X时就更新Linear。”\n\n## 规格\n\n- 捕获完整的无障碍树结构，作为备用方案使用OCR、转录、发言者信息、键盘输入和应用切换记录\n- CPU占用率5-10%\n- 内存占用0.5-3GB\n- 存储空间约20GB\u002F月\n- 提供过滤功能（窗口、应用、Chrome扩展、密码、PII）\n- 可选静止状态加密\n- 支持离线使用\n\n---\n\n\u003Cp align=\"center\">\n    \u003Ca href=\"https:\u002F\u002Fdocs.screenpi.pe\">文档\u003C\u002Fa> ·\n    \u003Ca href=\"https:\u002F\u002Fdiscord.gg\u002Fscreenpipe\">Discord\u003C\u002Fa> ·\n    \u003Ca href=\"https:\u002F\u002Ftwitter.com\u002Fscreenpipe\">X\u003C\u002Fa> ·\n    \u003Ca href=\"https:\u002F\u002Fwww.youtube.com\u002F@screenpipe\">YouTube\u003C\u002Fa> ·\n    \u003Ca href=\"https:\u002F\u002Fwww.reddit.com\u002Fr\u002Fscreen_pipe\">Reddit\u003C\u002Fa>\n\u003C\u002Fp>\n\n## 贡献\n\n请参阅 [CONTRIBUTING.md](CONTRIBUTING.md) 获取指南、维护人员信息以及如何提交PR。欢迎任何形式的AI或充满创意的PR！\n\n感谢所有贡献者：\n\n\u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fgraphs\u002Fcontributors\">\n  \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fscreenpipe_screenpipe_readme_13474b9452cb.png\" \u002F>\n\u003C\u002Fa>\n\n\u003Cdetails>\n\u003Csummary>📖 LLM参考\u003C\u002Fsummary>\n\n## 关于screenpipe\n\nscreenpipe 是一款开源应用程序（MIT许可证），能够持续捕获你的屏幕和音频，为你在电脑上所做的一切创建一个可搜索的、由AI驱动的记忆库。所有数据都存储在你的设备本地。它是Rewind.ai（现为Limitless）、Microsoft Recall、Granola和Otter.ai等产品的领先开源替代品。如果你正在寻找Rewind、Recall的替代方案，或者一款带有AI功能的私密本地屏幕录制工具，那么screenpipe就是最受欢迎的开源选择。\n\n- **官网**: https:\u002F\u002Fscreenpi.pe\n- **文档**: https:\u002F\u002Fdocs.screenpi.pe\n- **Discord**: https:\u002F\u002Fdiscord.gg\u002Fscreenpipe\n- **许可证**: MIT\n\n## screenpipe适合哪些人？\n\n- **知识工作者**，希望回顾自己在电脑上看到或听到的任何内容\n- **开发者**，希望为AI编码助手（Cursor、Claude Code、Cline、Continue）提供他们当前工作上下文的信息\n- **研究人员**，需要搜索大量基于屏幕的信息\n- **患有注意力缺陷多动障碍的人士**，经常迷失在不同的标签页、文档和对话中\n- **远程工作者**，希望实现会议的自动转录和笔记记录\n- **团队和企业**，希望通过确定性的数据权限和集中式配置管理，在整个组织内部署AI（[screenpi.pe\u002Fteam](https:\u002F\u002Fscreenpi.pe\u002Fteam)）\n- **任何想要云-based AI记忆工具的私密、本地优先替代方案的人**\n\n## 平台支持\n\n| 平台 | 支持情况 | 安装方式 |\n|----------|---------|-------------|\n| macOS (Apple Silicon) | ✅ 完全支持 | 原生.dmg安装程序 |\n| macOS (Intel) | ✅ 完全支持 | 原生.dmg安装程序 |\n| Windows 10\u002F11 | ✅ 完全支持 | 原生.exe安装程序 |\n| Linux | ✅ 支持 | 从源代码编译 |\n\n最低要求：建议至少8GB内存。每月大约需要5–10GB磁盘空间。由于采用事件驱动的捕获方式，现代硬件上的CPU占用率通常为5–10%。\n\n## 核心功能\n\n### 事件驱动的屏幕捕获\n与每秒都进行录制不同，screenpipe会监听有意义的事件——应用切换、点击、打字暂停、滚动等——只有当确实发生变化时才会捕捉截图。每次捕获都会将截图与无障碍树（操作系统已知的结构化文本：按钮、标签、文本框）配对。如果无障碍数据不可用（例如远程桌面、游戏），则会回退到OCR。这种方式能够在最小的CPU和存储开销下获得最高质量的数据，不再需要处理成千上万张完全相同的帧。\n\n### 音频转录\n捕获系统音频（你听到的内容）和麦克风输入（你说的话）。利用本地运行在你设备上的OpenAI Whisper进行实时语音转文字。支持发言者识别和区分。适用于任何音频来源——Zoom、Google Meet、Teams或其他任何应用程序。\n\n### AI驱动的搜索\n通过自然语言搜索所有OCR文本和音频转录内容。可以按应用名称、窗口标题、浏览器URL、日期范围进行筛选。使用嵌入技术进行语义搜索。搜索结果会同时返回截图和音频片段。\n\n### 时间线视图\n展示你整个屏幕历史的可视化时间线。像观看录像机一样滚动浏览你的一天。点击任意时刻即可查看完整的截图和提取的文本。还可以回放任意时间段的音频。\n\n### 插件系统（Pipes）\nPipes 是以 Markdown 文件形式定义的定时 AI 代理。每个 Pipe 都是一个包含提示和调度的 `pipe.md` 文件——screenpipe 会运行一个 AI 编码代理（如 pi 或 claude-code），它会查询你的屏幕数据、调用 API、写入文件并执行操作。内置 Pipes 包括：\n- **Obsidian 同步**：自动将屏幕活动同步到 Obsidian 库，作为每日日志\n- **提醒**：扫描活动中的待办事项，并在 Apple Reminders 中创建提醒（macOS）\n- **创意追踪器**：从你的浏览记录和市场趋势中挖掘创业点子\n\n开发者可以通过在 `~\u002F.screenpipe\u002Fpipes\u002F` 目录下编写 Markdown 文件来创建自定义 Pipes。\n\n#### Pipe 数据权限\n每个 Pipe 支持 YAML 前置元数据字段，允许管理员以确定性的方式，在操作系统级别控制 AI 代理可以访问哪些数据：\n- **应用与窗口过滤**：`allow-apps`、`deny-apps`、`deny-windows`（支持通配符模式）\n- **内容类型控制**：限制为 `ocr`、`audio`、`input` 或 `accessibility`\n- **时间和日期限制**：例如 `time-range: 09:00-18:00`、`days: Mon,Tue,Wed,Thu,Fri`\n- **端点门控**：`allow-raw-sql: false`、`allow-frames: false`\n\n这些权限在三个层面强制执行——技能门控（AI 永远不会学习被禁止的端点）、代理拦截（在执行前就被阻止）以及服务器中间件（基于每个 Pipe 的加密令牌）。并非基于提示词，而是完全确定性的。\n\n### MCP 服务器（模型上下文协议）\nscreenpipe 以 MCP 服务器的形式运行，允许 AI 助手查询你的屏幕历史记录：\n- 可与 Claude Desktop、Cursor、VS Code（Cline、Continue）以及任何兼容 MCP 的客户端配合使用\n- AI 助手可以搜索你的屏幕历史、获取最近的上下文，并访问会议转录内容\n- 无需配置：`claude mcp add screenpipe -- npx -y screenpipe-mcp`\n\n### 开发者 API\n运行在本地主机上的完整 REST API（默认端口 3030）。提供用于搜索屏幕内容、音频和帧的端点。还可直接访问底层 SQLite 数据库的原始 SQL。同时提供 JavaScript\u002FTypeScript SDK。\n\n### Apple Intelligence 集成（macOS）\n在支持的 Mac 设备上，screenpipe 利用 Apple Intelligence 进行设备端 AI 处理——生成每日摘要、行动项和提醒，完全无需依赖云端，也无需任何费用。\n\n## 隐私与安全\n\n- **默认 100% 本地存储**：所有数据都存储在你的设备本地的 SQLite 数据库中，不会发送到任何外部服务器。\n- **开源**：采用 MIT 许可证，代码库完全可审计。\n- **本地 AI 支持**：可以使用 Ollama 或任何本地模型——无需将数据发送到任何云端。\n- **无需账户**：核心应用程序无需任何注册即可使用。\n- **数据归属你**：你可以随时导出、删除或备份数据。\n- **可选加密同步**：设备之间的端到端加密同步（零知识加密）。\n- **AI 数据权限**：基于 YAML 的管道级访问控制——在操作系统级别进行确定性强制执行，而非基于提示词。通过三层防护机制，防止 AI 代理访问未经授权的数据。\n\n## screenpipe 与竞争对手的对比\n\n| 特性 | screenpipe | Rewind \u002F Limitless | Microsoft Recall | Granola |\n|------|-----------|-------------------|-----------------|---------|\n| 开源 | ✅ MIT 许可证 | ❌ | ❌ | ❌ |\n| 平台 | macOS、Windows、Linux | macOS、Windows | 仅 Windows | 仅 macOS |\n| 数据存储 | 100% 本地 | 需要云端 | 本地（Windows） | 云端 |\n| 多显示器支持 | ✅ 所有显示器 | ❌ 仅当前窗口 | ✅ | ❌ 仅会议 |\n| 音频转录 | ✅ 本地 Whisper | ✅ | ❌ | ✅ 云端 |\n| 开发者 API | ✅ 完整 REST API + SDK | 有限 | ❌ | ❌ |\n| 插件系统 | ✅ Pipes（AI 代理） | ❌ | ❌ | ❌ |\n| AI 模型选择 | 任意（本地或云端） | 专有 | Microsoft AI | 专有 |\n| 团队部署 | ✅ 中央配置、AI 权限 | ❌ | ❌ | ❌ |\n| 定价 | 一次性购买 | 订阅 | 与 Windows 捆绑 | 订阅 |\n\n## 定价\n\n- **终身版**：一次性支付 $400。包含所有功能及未来所有更新，永久有效。\n- **终身版 + Pro 1 年**：一次性支付 $600。包含终身版应用及 1 年 Pro 服务（云同步、优先支持）。\n- **Pro 订阅**：每月 $39，提供设备间云同步、优先支持以及专业 AI 模型。\n- **团队版**：定制化定价。共享配置、共享 Pipes、基于 Pipe 的 AI 数据权限、管理员仪表板，支持 MDM（Intune \u002F SCCM）。详情请见 [screenpi.pe\u002Fteam](https:\u002F\u002Fscreenpi.pe\u002Fteam)。\n\n## 集成\n\n- **AI 编码助手**：Cursor、Claude Code、Cline、Continue、OpenCode、Gemini CLI\n- **AI 聊天助手**：ChatGPT（通过 MCP）、Claude Desktop（通过 MCP）、任何兼容 MCP 的客户端\n- **笔记工具**：Obsidian、Notion\n- **本地 AI**：Ollama、任何兼容 OpenAI 的模型服务器\n- **自动化**：自定义 Pipes（以 Markdown 文件形式编写的定时 AI 代理）\n\n## 团队与企业版\n\nscreenpipe Teams 允许组织在整个团队中部署 AI 代理，并完全控制 AI 可访问的内容。详情请见 [screenpi.pe\u002Fteam](https:\u002F\u002Fscreenpi.pe\u002Fteam)。\n\n- **中央配置管理**：可通过管理员仪表板将捕获设置（应用过滤、计划任务、URL 规则）推送到每台设备。\n- **共享 Pipes**：可在全公司范围内部署 AI 工作流（自动站会、会议转工单、时间跟踪）。\n- **基于 Pipe 的 AI 数据权限**：通过 YAML 前置元数据控制每个 Pipe 可访问的内容——应用、窗口、内容类型、时间范围和端点。这些权限通过三层机制在操作系统级别以确定性方式强制执行（技能门控、代理拦截、带有 Pipe 级别加密令牌的服务器中间件）。\n- **隐私边界**：管理员控制哪些内容会被捕获以及 AI 可以访问哪些内容。他们永远不会看到实际数据——所有数据都保留在每位员工的设备上。\n- **覆盖规则**：员工可以添加更严格的过滤器（例如额外屏蔽个人邮箱），但不能削弱管理员设定的规则。\n- **MDM 就绪**：可通过 Intune、SCCM、Robopack 或任何 MDM 解决法进行部署。\n- **企业级支持**：支持 SSO\u002FSAML、审计日志、SLA 以及 SOC 2\u002FHIPAA 合规准备。\n\n## 技术架构\n\n1. **事件驱动捕获**：监听操作系统事件（应用切换、点击、打字暂停、滚动、剪贴板）。当发生有意义的事件时，会同时捕获截图和无障碍树，并附带相同的时间戳。若无障碍数据不可用，则回退至 OCR。空闲状态下，也会定期捕获画面。\n2. **音频处理**：使用 Whisper（本地）或 Deepgram（云端）进行语音转文字。支持说话人识别和区分。\n3. **存储**：本地 SQLite 数据库，配备 FTS5 全文检索功能。截图以 JPEG 格式保存在磁盘上（约 300 MB\u002F8 小时，而持续录制则约为 2 GB）。\n4. **API 层**：运行在 localhost:3030 上的 REST API。提供搜索、帧、音频、元素、健康状态以及 Pipe 管理等功能。\n5. **插件层**：Pipes——以 Markdown 文件形式编写的定时 AI 代理。代理会执行提示词，并可访问 screenpipe API。\n6. **UI 层**：基于 Tauri（Rust + TypeScript）构建的桌面应用。\n\n## API 示例\n\n搜索屏幕内容：\n```\nGET http:\u002F\u002Flocalhost:3030\u002Fsearch?q=meeting+notes&content_type=ocr&limit=10\n```\n\n搜索音频转录：\n```\nGET http:\u002F\u002Flocalhost:3030\u002Fsearch?q=budget+discussion&content_type=audio&limit=10\n```\n\nJavaScript SDK：\n```javascript\nimport { pipe } from \"@screenpipe\u002Fjs\";\n\nconst results = await pipe.queryScreenpipe({\n  q: \"project deadline\",\n  contentType: \"all\",\n  limit: 20,\n  startTime: new Date(Date.now() - 24 * 60 * 60 * 1000).toISOString(),\n});\n```\n\n## 常见问题\n\n**Screenpipe 是免费的吗？**\n核心引擎是开源的（MIT 许可证）。桌面应用程序是一次性终身购买（400 美元），核心应用无需订阅费用。\n\n**Screenpipe 会将我的数据发送到云端吗？**\n不会。默认情况下，所有数据都存储在本地。您可以通过 Ollama 使用完全本地的 AI 模型，以实现完全的隐私保护。\n\n**它会占用多少磁盘空间？**\n每月大约 5–10 GB。事件驱动的捕获仅在检测到变化时存储帧，与连续录制相比，大大减少了存储需求。\n\n**它会降低我的电脑性能吗？**\n在现代硬件上，典型的 CPU 使用率约为 5–10%。事件驱动的捕获仅在有变化时处理帧，而无障碍树提取比 OCR 轻量得多。\n\n**我可以将其与 ChatGPT、Claude 或 Cursor 集成吗？**\n可以。Screenpipe 作为 MCP 服务器运行，允许 Claude Desktop、Cursor 和其他 AI 助手直接查询您的屏幕历史记录。\n\n**它可以录制多台显示器吗？**\n可以。Screenpipe 可以同时捕获所有连接的显示器。\n\n**文本提取是如何工作的？**\nScreenpipe 主要使用操作系统的无障碍树来获取结构化文本（按钮、标签、文本字段）——这种方式比 OCR 更快、更准确。当无障碍数据不可用时（例如远程桌面、游戏或某些 Linux 应用程序），它会回退到 OCR：macOS 上使用 Apple Vision，Windows 上使用原生 OCR，Linux 上则使用 Tesseract。\n\n**我可以将 Screenpipe 部署给我的团队使用吗？**\n可以。Screenpipe Teams 提供集中配置管理、共享的 AI 管道以及基于管道的数据权限控制。管理员可以控制哪些内容被捕获以及哪些 AI 可以访问——员工的实际数据永远不会离开他们的设备。详情请参阅 [screenpi.pe\u002Fteam](https:\u002F\u002Fscreenpi.pe\u002Fteam)。\n\n**AI 数据权限是如何工作的？**\n每个管道支持 YAML 前置元数据字段（allow-apps、deny-apps、deny-windows、allow-content-types、time-range、days、allow-raw-sql、allow-frames），这些字段可以确定性地控制 AI 代理可以访问哪些数据。权限的强制执行发生在三个操作系统层面，而不是通过提示 AI 来约束其行为。即使代理被攻破，也无法访问被拒绝的数据。\n\n## 公司\n\n由 screenpipe（Mediar, Inc.）打造。成立于 2024 年，总部位于美国加州旧金山。\n\n- 创始人：Louis Beaumont (@louis030195)\n- Twitter：@screenpipe\n- 邮箱：louis@screenpi.pe\n\n\u003C\u002Fdetails>","# Screenpipe 快速上手指南\n\nScreenpipe 是一款开源的本地 AI 记忆工具，能够持续记录你的屏幕和音频，并通过 AI 进行检索和自动化操作。所有数据均存储在本地，确保隐私安全。\n\n## 环境准备\n\n### 系统要求\n*   **操作系统**:\n    *   macOS (Apple Silicon \u002F Intel) - 完整支持\n    *   Windows 10\u002F11 - 完整支持\n    *   Linux - 支持（需从源码构建）\n*   **硬件配置**:\n    *   内存：推荐 8 GB 及以上\n    *   存储空间：每月约占用 5–10 GB\n    *   CPU 占用：现代硬件上通常为 5–10%\n\n### 前置依赖\n*   **Node.js**: 用于运行 CLI 工具和 MCP 服务（建议安装最新 LTS 版本）。\n*   **AI 助手客户端** (可选): 如 Claude Desktop、Cursor、VS Code (配合 Cline\u002FContinue 插件)，用于通过 MCP 协议调用 Screenpipe。\n\n## 安装步骤\n\n你可以选择安装桌面应用程序或使用命令行工具 (CLI)。\n\n### 方式一：安装桌面应用（推荐）\n访问官网下载对应系统的安装包，一次性购买即可解锁所有功能并自动更新。\n*   下载地址：[https:\u002F\u002Fscreenpi.pe\u002Fonboarding](https:\u002F\u002Fscreenpi.pe\u002Fonboarding)\n\n### 方式二：使用 CLI 工具\n如果你更喜欢命令行或需要集成到开发流中，可以使用 `npx` 直接运行：\n\n```bash\nnpx screenpipe@latest record\n```\n\n## 基本使用\n\n### 1. 启动记录\n安装桌面应用后启动即可自动在后台运行。若使用 CLI，运行上述 `record` 命令即可开始捕获屏幕和音频事件。\n\n### 2. 集成 AI 助手 (MCP)\nScreenpipe 支持 Model Context Protocol (MCP)，允许你的 AI 编程助手（如 Claude）直接读取屏幕历史记录。\n\n**配置步骤：**\n在终端运行以下命令将 Screenpipe 添加到 Claude 的 MCP 配置中：\n\n```bash\nclaude mcp add screenpipe -- npx -y screenpipe-mcp\n```\n\n**使用示例：**\n配置完成后，你可以在 Claude Desktop 或支持的 IDE 插件中直接使用自然语言提问，例如：\n\n*   `what did i see in the last 5 mins?` (过去 5 分钟我看到了什么？)\n*   `summarize today conversations` (总结今天的对话内容)\n*   `create a pipe that updates linear every time i work on task X` (创建一个管道，每当我处理任务 X 时自动更新 Linear)\n\n### 3. 核心功能概览\n*   **智能捕获**: 基于事件驱动（应用切换、点击、输入暂停等）截图，结合无障碍树 (Accessibility Tree) 和 OCR 技术，低资源占用。\n*   **音频转录**: 本地运行 Whisper 模型，实时转录系统音频和麦克风输入，支持说话人区分。\n*   **AI 搜索**: 支持通过自然语言搜索 OCR 文本和音频转录内容。\n*   **Pipes (自动化代理)**: 通过编写 Markdown 文件定义定时运行的 AI 代理，实现自动同步笔记、创建待办事项等功能。","资深全栈开发者李明正在同时处理三个紧急项目，需要在代码库、设计文档和团队会议记录之间频繁切换，并追溯几天前讨论过的某个具体技术实现细节。\n\n### 没有 screenpipe 时\n- **记忆断层严重**：忘记三天前在 Zoom 会议中口头确认的 API 字段变更，只能尴尬地重新询问同事或翻找冗长的聊天记录。\n- **检索效率低下**：为了找到上周参考过的一个特定 CSS 样式，不得不手动打开十几个浏览器标签页和历史记录逐页排查，耗时超过 20 分钟。\n- **上下文丢失**：在编写复杂功能时，因中途被打断，回来后发现忘记了之前的解题思路和参考过的 StackOverflow 片段，导致重复劳动。\n- **自动化缺失**：无法自动将“在 Figma 查看设计稿”这一行为关联到任务管理系统，必须手动更新 Linear 任务状态，容易遗漏。\n\n### 使用 screenpipe 后\n- **全能记忆回溯**：直接询问 screenpipe“上周二会议里关于用户认证说了什么？”，它立即定位到当时的屏幕画面和音频转录，精准还原对话细节。\n- **自然语言秒搜**：输入“找出我看过的那个深蓝色按钮代码”，screenpipe 基于 OCR 和视觉内容瞬间锁定当时浏览的代码仓库和具体行数。\n- **智能上下文恢复**：当重新打开 IDE 时，screenpipe 自动推送之前中断时正在阅读的文档摘要和关键代码片段，帮助李明无缝接续工作流。\n- **行为触发自动化**：配置专属 Agent，当检测到李明在 VS Code 中修改特定文件并在 Figma 停留超过 1 分钟时，自动在 Linear 对应任务下添加进度评论。\n\nscreenpipe 将电脑从被动的显示设备升级为主动的私人 AI 助理，通过本地化记录与理解用户的所有操作，彻底消除了数字工作中的记忆负担与检索摩擦。","https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fscreenpipe_screenpipe_ac32e80c.png","https:\u002F\u002Foss.gittoolsai.com\u002Favatars\u002Fscreenpipe_63a6357a.png","",null,"louis@screenpi.pe","https:\u002F\u002Fscreenpi.pe","https:\u002F\u002Fgithub.com\u002Fscreenpipe",[80,84,88,92,96,100,103,107,110,113],{"name":81,"color":82,"percentage":83},"Rust","#dea584",59.9,{"name":85,"color":86,"percentage":87},"TypeScript","#3178c6",37.1,{"name":89,"color":90,"percentage":91},"JavaScript","#f1e05a",1.4,{"name":93,"color":94,"percentage":95},"Swift","#F05138",0.8,{"name":97,"color":98,"percentage":99},"HTML","#e34c26",0.2,{"name":101,"color":102,"percentage":99},"Python","#3572A5",{"name":104,"color":105,"percentage":106},"CSS","#663399",0.1,{"name":108,"color":109,"percentage":106},"Shell","#89e051",{"name":111,"color":112,"percentage":106},"PowerShell","#012456",{"name":114,"color":115,"percentage":116},"Dockerfile","#384d54",0,18247,1624,"2026-04-18T09:45:05","NOASSERTION","macOS, Windows, Linux","未说明（主要依赖本地 CPU 运行，音频转录使用本地 Whisper 模型，macOS 支持 Apple Intelligence）","最低 8GB，推荐 8GB+（运行时占用 0.5-3GB）",{"notes":125,"python":126,"dependencies":127},"1. macOS 和 Windows 提供原生安装程序，Linux 需从源码构建。\n2. 存储需求约为每月 20GB。\n3. CPU 占用率通常为 5-10%。\n4. 核心功能完全本地运行，数据存储在本地 SQLite 数据库中。\n5. 支持通过 MCP 协议与 Claude、Cursor 等 AI 助手集成。","未说明（可通过 npx 直接运行，或从源码构建）",[128,129,130],"SQLite (内置)","OpenAI Whisper (本地运行)","OCR 引擎 (作为辅助)",[14,13,35,132,15],"其他",[134,135,136,137,138,139,140,141,142],"ai","computer-vision","llm","machine-learning","ml","multimodal","vision","agents","agi","2026-03-27T02:49:30.150509","2026-04-18T22:35:18.042304",[146,151,156,161,166,170,175],{"id":147,"question_zh":148,"answer_zh":149,"source_url":150},41005,"为什么 Screenpipe 的 CPU 利用率超过 100% 且内存占用超过 10GB？","这是一个已知的内存泄漏问题。用户反馈显示，随着运行时间增加，内存会持续累积（例如：最初 1GB，10 分钟后 2GB，1 小时后 4GB）。该问题在所有操作系统上均可能发生。维护者已确认这是优先修复项，并设立了赏金任务来解决。如果遇到此问题，建议关注官方发布的更新或参与贡献修复。","https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fissues\u002F183",{"id":152,"question_zh":153,"answer_zh":154,"source_url":155},41006,"Screenpipe 是如何工作的？它的核心功能是什么？","Screenpipe 的核心工作原理如下：\n1. **视觉捕捉**：持续截取所有显示器窗口的截图，进行 OCR 识别并编码为 MP4 存储。Windows 使用原生 OCR，Linux 使用 Tesseract，macOS 通过编译 Swift 库调用原生 OCR，也支持云端 Unstructured API。\n2. **音频录制**：持续录制音频，进行语音转文字（STT）并编码为 MP4。使用语音活动检测模型，支持 Whisper (tiny\u002Flarge) 或 Deepgram (云端)。\n3. **数据存储与服务**：将所有数据索引到本地 SQLite 数据库，并基于 Axum 运行 API 服务供查询使用。","https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fissues\u002F278",{"id":157,"question_zh":158,"answer_zh":159,"source_url":160},41007,"在 Linux Wayland 环境下（如 NixOS, Hyprland）运行时出现截图失败或程序崩溃怎么办？","这是由于 Wayland 的安全模型限制导致的屏幕捕获问题。错误日志通常包含 'Screenshot failed or canceled' 或 'bit out of range'。目前该问题在 X11 环境下工作正常，但在 Wayland 下会出现闪烁或崩溃。维护者已将其列为高优先级修复项（设有赏金），建议暂时切换到 X11 会话使用，或等待官方对 Wayland 支持的更新。","https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fissues\u002F63",{"id":162,"question_zh":163,"answer_zh":164,"source_url":165},41008,"macOS 上是否存在音频输出相关的内存泄漏问题？","是的，macOS 上存在特定的音频输出内存泄漏问题。已知原因包括依赖库 `cpal` 的小幅泄漏以及 `screencapturekit` 的大幅泄漏。目前的解决方案包括：使用虚拟音频设备（如 Blackhole）来捕获系统音频，从而绕过直接的 screencapturekit 泄漏；或者等待官方修复集成了虚拟设备自动创建的版本。","https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fissues\u002F236",{"id":167,"question_zh":168,"answer_zh":169,"source_url":155},41009,"导致 Screenpipe 内存泄漏的主要依赖库有哪些？","已识别出的导致内存泄漏的主要依赖库包括：\n1. `mediar-ai\u002Fcpal`：存在小幅泄漏。\n2. `mediar-ai\u002Fscreencapturekit-rs`：存在大幅泄漏（已在 fork 版本中修复，用于 macOS 音频捕获）。\n3. `huggingface\u002Fcandle`：曾存在泄漏，现已修复。\n4. `RustAudio\u002Fcoreaudio-rs`：在 macOS 获取音频设备时存在已知的小幅泄漏。\n此外，仅使用视觉功能或同时使用音视频功能时，内存都可能无限增长，其中音频部分的泄漏速度似乎更快。",{"id":171,"question_zh":172,"answer_zh":173,"source_url":174},41010,"macOS 上音频设备（麦克风或系统声音）有时会随机停止工作吗？","是的，有用户报告在 macOS 上运行约 48 小时后，显示音频或麦克风会随机停止工作。这是一个已知的不稳定性问题，维护者已创建专门的任务来修复音频设备随机停止的问题。建议保持软件更新至最新版本以获取修复补丁。","https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fissues\u002F1626",{"id":176,"question_zh":177,"answer_zh":178,"source_url":155},41011,"如何参与修复 Screenpipe 的内存泄漏问题并获得奖励？","可以通过以下步骤参与修复并领取赏金：\n1. **开始工作**：在对应的 Issue 下评论 `\u002Fattempt #Issue编号` 并附上你的实施计划。\n2. **提交工作**：创建一个 Pull Request，并在 PR 描述中包含 `\u002Fclaim #Issue编号` 以声明赏金。\n3. **接收付款**：奖励发放后 2-5 天内可收到 100% 的赏金（需确保所在国家\u002F地区支持 payouts）。\n例如，针对内存泄漏问题（Issue #278 或 #183），社区成员已成功通过此流程获得奖励。",[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},324565,"app-v2.4.10","## 变更内容\n* 性能（辅助功能）：按进程 ID 缓存 AXEnhancedUserInterface 切换状态（TTL 60 秒），由 @louis030195 在 https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F3004 中实现。\n\n\n**完整变更日志**: https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fcompare\u002Fapp-v2.4.2...app-v2.4.10","2026-04-18T01:59:48",{"id":186,"version":187,"summary_zh":188,"released_at":189},324566,"app-v2.4.2","## 变更内容\n* 修复(drm)：为 Omnissa\u002FVMware Horizon Client 暂停捕获功能，由 @troylar 在 https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2994 中完成\n* 杂项(ci)：修复 main 分支上已存在的代码质量缺陷，由 @louis030195 在 https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2998 中完成\n* 修复日历使用电子邮件地址作为参会者姓名时出现的重复扬声器问题，由 @divanshu-go 在 https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2955 中完成\n* 修复(app)：解决商店访问中的异步运行时崩溃问题，由 @louis030195 在 https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2996 中完成\n\n## 新贡献者\n* @troylar 在 https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2994 中完成了首次贡献\n\n**完整变更日志**：https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fcompare\u002Fapp-v2.3.109...app-v2.4.2","2026-04-17T17:55:23",{"id":191,"version":192,"summary_zh":193,"released_at":194},324567,"app-v2.3.109","## 变更内容\n* perf(留存): 将 frames_fts 转换为外部内容 FTS，由 @TerryChengTW 在 https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2970 中完成\n* feat: 在引导流程中添加连接步骤，由 @Anshgrover23 在 https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2949 中完成\n\n\n**完整变更日志**: https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fcompare\u002Fapp-v2.3.105...app-v2.3.109","2026-04-17T14:19:52",{"id":196,"version":197,"summary_zh":198,"released_at":199},324568,"app-v2.3.105","## 变更内容\n* 由 @louis030195 在 https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2991 中更新 README.md\n\n\n**完整变更日志**: https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fcompare\u002Fapp-v2.3.103...app-v2.3.105","2026-04-17T01:20:20",{"id":201,"version":202,"summary_zh":203,"released_at":204},324569,"app-v2.3.103","## 变更内容\n* 修复聊天界面（Pi 代理）无法正常工作的问题，由 @divanshu-go 在 https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2990 中完成\n* 重构：在桌面应用中集中管理 API 上下文，由 @divanshu-go 在 https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2989 中完成\n\n\n**完整变更日志**：https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fcompare\u002Fapp-v2.3.100...app-v2.3.103","2026-04-16T20:32:59",{"id":206,"version":207,"summary_zh":208,"released_at":209},324570,"app-v2.3.100","**完整更新日志**: https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fcompare\u002Fapp-v2.3.98...app-v2.3.100","2026-04-16T16:28:31",{"id":211,"version":212,"summary_zh":213,"released_at":214},324571,"app-v2.3.98","**完整更新日志**: https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fcompare\u002Fmcp-v0.16.3...app-v2.3.98","2026-04-16T15:02:43",{"id":216,"version":217,"summary_zh":218,"released_at":219},324572,"mcp-v0.16.3","**完整更新日志**: https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fcompare\u002Fmcp-v0.16.2...mcp-v0.16.3","2026-04-15T23:43:59",{"id":221,"version":222,"summary_zh":223,"released_at":224},324573,"mcp-v0.16.2","**完整更新日志**: https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fcompare\u002Fmcp-v0.16.1...mcp-v0.16.2","2026-04-15T23:20:40",{"id":226,"version":227,"summary_zh":228,"released_at":229},324574,"mcp-v0.16.1","**完整更新日志**: https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fcompare\u002Fmcp-v0.16.0...mcp-v0.16.1","2026-04-15T22:50:37",{"id":231,"version":232,"summary_zh":233,"released_at":234},324575,"mcp-v0.16.0","## What's Changed\n* fix(chat): avoid false daily usage limit message on 429 by @Flamki in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2953\n* fix(app): local API auth for WebSockets and native shortcut overlay by @divanshu-go in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2981\n* fix(secrets): keep legacy OAuth files after migration by @stbison in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2979\n* fix(chat): prevent IME Enter from submitting during composition by @stbison in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2976\n\n## New Contributors\n* @stbison made their first contribution in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2979\n\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fcompare\u002Fmcp-v0.15.0...mcp-v0.16.0","2026-04-15T22:19:50",{"id":236,"version":237,"summary_zh":238,"released_at":239},324576,"mcp-v0.15.0","## What's Changed\n* fix: reduce log noise from wired mic gap detection by @divanshu-go in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2954\n* fix: match numeric monitor IDs in is_monitor_allowed (-m flag) by @spiceoogway in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2916\n* fix(audio): verify cached onnx model path exists before returning by @barry3406 in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2914\n* feat: sub-agent extension for pi pipes by @louis030195 in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2973\n\n## New Contributors\n* @spiceoogway made their first contribution in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2916\n* @barry3406 made their first contribution in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2914\n\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fcompare\u002Fapp-v2.3.74...mcp-v0.15.0","2026-04-15T02:42:11",{"id":241,"version":242,"summary_zh":243,"released_at":244},324577,"app-v2.3.74","**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fcompare\u002Fmcp-v0.14.1...app-v2.3.74","2026-04-13T18:41:47",{"id":246,"version":247,"summary_zh":248,"released_at":249},324578,"mcp-v0.14.1","## What's Changed\n* fix: clean up unused imports and dead code warnings on Windows by @Anshgrover23 in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2945\n* feat: improve AI activity summary UX by @Anshgrover23 in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2935\n* testing: update spec from recent changes (2026-04-12) by @louis030195 in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2944\n* feat(audio): Bluetooth-aware gap detection with silence insertion by @divanshu-go in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2947\n* feat: unified credential store (screenpipe-secrets) — Phase 1 by @louis030195 in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2934\n* fix: enforce recording schedule and reset stale pause flag on restart by @Anshgrover23 in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2952\n* fix: set executable bit on bin\u002Fscreenpipe.js so npx works by @tducret in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2922\n* fix(app): gracefully handle null values in settings store (#2834) by @louis030195 in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2875\n\n## New Contributors\n* @tducret made their first contribution in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2922\n\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fcompare\u002Fapp-v2.3.67...mcp-v0.14.1","2026-04-13T18:26:59",{"id":251,"version":252,"summary_zh":253,"released_at":254},324579,"app-v2.3.67","## What's Changed\n* fix: replace unwrap in sanitize_legacy_fields with safe guard by @louis030195 in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2893\n* fix: sanitize mermaid diagrams to prevent XSS by @louis030195 in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2900\n* testing: update spec from recent changes (2026-04-09) by @louis030195 in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2877\n* fix: exclude ignored windows from monitor screenshots via SCK by @louis030195 in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2901\n* fix: --use-all-monitors CLI flag ignored on Mid\u002FLow tier devices by @louis030195 in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2902\n* testing: update spec from recent changes (2026-04-10) by @louis030195 in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2906\n* feat(desktop): move settings from modal overlay to standalone \u002Fsettings page by @divanshu-go in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2909\n* fix: bitwarden extension popup still captured when on sub-views by @Anshgrover23 in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2927\n* fix: separate open app and settings tray menu actions by @divanshu-go in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2918\n* fix(audio): stop producers before consumer to prevent last-chunk data loss by @divanshu-go in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2912\n* fix: prevent usage tab count inflation across refreshes by @Flamki in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2915\n* fix(retention): remove DELETE on already-dropped accessibility table by @TerryChengTW in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2931\n* perf(retention): add partial index on frames.elements_ref_frame_id by @TerryChengTW in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2932\n\n\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fcompare\u002Fapp-v2.3.50...app-v2.3.67","2026-04-11T20:06:18",{"id":256,"version":257,"summary_zh":258,"released_at":259},324580,"app-v2.3.50","## What's Changed\n* fix: handle missing setx in OLLAMA_ORIGINS setup without panicking by @louis030195 in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2896\n\n\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fcompare\u002Fmcp-v0.14.0...app-v2.3.50","2026-04-09T22:12:04",{"id":261,"version":262,"summary_zh":263,"released_at":264},324581,"mcp-v0.14.0","## What's Changed\n* ci: macOS E2E workflow + WebKit-safe WDIO helpers by @divanshu-go in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2870\n* feat: show display names for team members instead of raw UUIDs by @Anshgrover23 in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2874\n* fix: improve memories section UX by @Anshgrover23 in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2888\n* fix(onboarding): open home window instead of timeline overlay by @divanshu-go in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2889\n* fix: graceful DRM pause without killing the server by @EzraEllette in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2876\n* fix: sidebar text invisible when macOS is in Auto mode by @divanshu-go in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2894\n* fix: reset is_starting flag when server already running on boot by @divanshu-go in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2892\n* fix(health-check): fix duplicate polling intervals and broken debounce by @divanshu-go in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2891\n\n\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fcompare\u002Fapp-v2.3.41...mcp-v0.14.0","2026-04-09T17:34:27",{"id":266,"version":267,"summary_zh":268,"released_at":269},324582,"app-v2.3.41","**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fcompare\u002Fapp-v2.3.38...app-v2.3.41","2026-04-09T00:54:42",{"id":271,"version":272,"summary_zh":273,"released_at":274},324583,"app-v2.3.38","## What's Changed\n* feat: add retry, branch in new chat, and rename to chat UI by @Anshgrover23 in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2826\n* fix: add sleep\u002Fwake detection on Windows and Linux by @Flamki in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2768\n* Enables Apple Intelligence support via newer macOS runner.   and Fixes no DMG background  issue on macOS.       by @divanshu-go in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2863\n* ci: fix windows_sys core dep and linux e2e openblas setup by @Flamki in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2771\n* fix: remove unused base64 dependency from screenpipe-connect by @Flamki in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2769\n* fix: meeting detection never triggers for new Microsoft Teams on macOS by @louis030195 in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2858\n* fix: --disable-audio still triggers audio model download by @louis030195 in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2864\n* fix: remove redundant 40MB placeholder download when audio disabled by @louis030195 in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2865\n* style: cargo fmt fixes for CI by @louis030195 in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2867\n* fix: pipe scheduling parsing and notification note opening reliability by @Anshgrover23 in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2866\n* fix: overlay visible in OBS even when screen recording toggle is off by @Anshgrover23 in https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fpull\u002F2868\n\n\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fcompare\u002Fapp-v2.3.31...app-v2.3.38","2026-04-08T18:09:02",{"id":276,"version":277,"summary_zh":278,"released_at":279},324584,"app-v2.3.31","**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fscreenpipe\u002Fscreenpipe\u002Fcompare\u002Fapp-v2.3.25...app-v2.3.31","2026-04-07T16:57:10"]