[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"similar-sgasser--pasteguard":3,"tool-sgasser--pasteguard":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":81,"owner_twitter":82,"owner_website":83,"owner_url":84,"languages":85,"stars":98,"forks":99,"last_commit_at":100,"license":101,"difficulty_score":10,"env_os":102,"env_gpu":102,"env_ram":102,"env_deps":103,"category_tags":110,"github_topics":111,"view_count":10,"oss_zip_url":82,"oss_zip_packed_at":82,"status":16,"created_at":125,"updated_at":126,"faqs":127,"releases":155},556,"sgasser\u002Fpasteguard","pasteguard","AI gets the context. Not your secrets. Open-source privacy proxy for LLMs.","Pasteguard 是一款专为大语言模型设计的开源隐私代理工具。它的核心理念是“让 AI 获取上下文，而非你的秘密”。在日常使用 AI 助手时，我们难免会输入包含个人信息、代码密钥或敏感业务数据的文本，直接发送存在泄露风险。Pasteguard 通过在本地运行一个代理服务器，自动识别并掩盖这些敏感信息（如姓名、邮箱、API Key），确保只有脱敏后的内容发送给云端 AI，而用户端依然能看到原始数据。\n\nPasteguard 特别适合开发者、研究人员以及注重数据安全的普通用户。无论是日常对话、编写代码时的 Copilot 辅助，还是自建 AI 应用，它都能无缝集成。它支持超过 30 种敏感数据类型和 24 种语言，且数据完全保留在本地机器上。此外，它还具备独特的路由模式，能将包含敏感信息的请求自动转发至本地运行的模型，进一步降低隐私风险。通过简单的配置，用户即可在不改变原有工作流的前提下，为 AI 交互加上安全锁。","\u003Cp align=\"center\">\n  \u003Cpicture>\n    \u003Csource media=\"(prefers-color-scheme: dark)\" srcset=\"assets\u002Fwordmark-dark.svg\">\n    \u003Csource media=\"(prefers-color-scheme: light)\" srcset=\"assets\u002Fwordmark-light.svg\">\n    \u003Cimg src=\"assets\u002Fwordmark-light.svg\" width=\"220\" height=\"44\" alt=\"PasteGuard\">\n  \u003C\u002Fpicture>\n\u003C\u002Fp>\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fsgasser\u002Fpasteguard\u002Factions\u002Fworkflows\u002Fci.yml\">\u003Cimg src=\"https:\u002F\u002Fgithub.com\u002Fsgasser\u002Fpasteguard\u002Factions\u002Fworkflows\u002Fci.yml\u002Fbadge.svg\" alt=\"CI\">\u003C\u002Fa>\n  \u003Ca href=\"LICENSE\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FLicense-Apache%202.0-blue.svg\" alt=\"License\">\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fsgasser\u002Fpasteguard\u002Freleases\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fv\u002Frelease\u002Fsgasser\u002Fpasteguard\" alt=\"Release\">\u003C\u002Fa>\n\u003C\u002Fp>\n\n\u003Cp align=\"center\">\n  \u003Cstrong>AI gets the context. Not your secrets.\u003C\u002Fstrong>\u003Cbr>\n  Automatically hides names, emails, and API keys before you send prompts to AI.\n\u003C\u002Fp>\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"#quick-start\">\u003Cstrong>Quick Start\u003C\u002Fstrong>\u003C\u002Fa> ·\n  \u003Ca href=\"#chat\">\u003Cstrong>Chat\u003C\u002Fstrong>\u003C\u002Fa> ·\n  \u003Ca href=\"#coding-tools\">\u003Cstrong>Coding Tools\u003C\u002Fstrong>\u003C\u002Fa> ·\n  \u003Ca href=\"https:\u002F\u002Fpasteguard.com\u002Fdocs\">\u003Cstrong>Documentation\u003C\u002Fstrong>\u003C\u002Fa>\n\u003C\u002Fp>\n\n\u003Cbr\u002F>\n\n\u003Cpicture>\n  \u003Csource media=\"(prefers-color-scheme: dark)\" srcset=\"assets\u002Fcomparison-dark.png\">\n  \u003Csource media=\"(prefers-color-scheme: light)\" srcset=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fsgasser_pasteguard_readme_fd745a0cfea4.png\">\n  \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fsgasser_pasteguard_readme_fd745a0cfea4.png\" width=\"100%\" alt=\"PasteGuard — Without vs. With: masks names, emails, and API keys before they reach AI\">\n\u003C\u002Fpicture>\n\n\u003Cp align=\"center\">\n  Detects 30+ types of sensitive data across 24 languages.\u003Cbr>\n  Your data never leaves your machine.\n\u003C\u002Fp>\n\n## Works Everywhere\n\n**[Chat](https:\u002F\u002Fpasteguard.com\u002Fdocs\u002Fuse-cases\u002Fchat)** — Masks PII and secrets when you paste into ChatGPT, Claude, and Gemini. You see originals, AI sees placeholders.\n\n**[Apps](https:\u002F\u002Fpasteguard.com\u002Fdocs\u002Fuse-cases\u002Fapps)** — Open WebUI, LibreChat, or any self-hosted AI setup. Optionally routes sensitive requests to a local model.\n\n**[Coding Tools](https:\u002F\u002Fpasteguard.com\u002Fdocs\u002Fuse-cases\u002Fcoding-tools)** — Cursor, Claude Code, Copilot, Windsurf — your codebase context flows to the provider. PasteGuard masks secrets and PII before they leave.\n\n**[API Integration](https:\u002F\u002Fpasteguard.com\u002Fdocs\u002Fuse-cases\u002Fapi-integration)** — Sits between your code and OpenAI or Anthropic. Change one URL, your users' data stays protected.\n\n## Quick Start\n\nRun PasteGuard as a local proxy:\n\n```bash\ndocker run --rm -p 3000:3000 ghcr.io\u002Fsgasser\u002Fpasteguard:en\n```\n\nPoint your tools or app to PasteGuard instead of the provider:\n\n| API | PasteGuard URL | Original URL |\n|----------|----------------|--------------|\n| OpenAI | `http:\u002F\u002Flocalhost:3000\u002Fopenai\u002Fv1` | `https:\u002F\u002Fapi.openai.com\u002Fv1` |\n| Anthropic | `http:\u002F\u002Flocalhost:3000\u002Fanthropic` | `https:\u002F\u002Fapi.anthropic.com` |\n\n```python\n# One line to protect your data\nclient = OpenAI(base_url=\"http:\u002F\u002Flocalhost:3000\u002Fopenai\u002Fv1\")\n```\n\n\u003Cdetails>\n\u003Csummary>\u003Cstrong>European Languages\u003C\u002Fstrong>\u003C\u002Fsummary>\n\nFor German, Spanish, French, Italian, Dutch, Polish, Portuguese, and Romanian:\n\n```bash\ndocker run --rm -p 3000:3000 ghcr.io\u002Fsgasser\u002Fpasteguard:eu\n```\n\nFor custom config, persistent logs, or other languages: **[Read the docs →](https:\u002F\u002Fpasteguard.com\u002Fdocs\u002Finstallation)**\n\n\u003C\u002Fdetails>\n\n\u003Cdetails>\n\u003Csummary>\u003Cstrong>Route Mode\u003C\u002Fstrong>\u003C\u002Fsummary>\n\nRoute Mode sends requests containing sensitive data to a local LLM (Ollama, vLLM, llama.cpp). Everything else goes to OpenAI or Anthropic. Sensitive data stays on your network.\n\n**[Route Mode docs →](https:\u002F\u002Fpasteguard.com\u002Fdocs\u002Fconcepts\u002Froute-mode)**\n\n\u003C\u002Fdetails>\n\n## Chat\n\nOpen-source browser extension for ChatGPT, Claude, and Gemini.\n\n- Paste customer data → masked before it reaches the AI\n- AI responds with placeholders → you see the originals\n- Works with the same detection engine as the proxy\n\nCurrently in beta. Apache 2.0.\n\n**[Join the Beta →](https:\u002F\u002Ftally.so\u002Fr\u002FJ9pNLr)** · **[Chat docs →](https:\u002F\u002Fpasteguard.com\u002Fdocs\u002Fuse-cases\u002Fchat)**\n\n## Coding Tools\n\nProtect your codebase context and secrets when using AI coding assistants.\n\n**Claude Code:**\n\n```bash\nANTHROPIC_BASE_URL=http:\u002F\u002Flocalhost:3000\u002Fanthropic claude\n```\n\n**Cursor:** Settings → Models → Enable \"Override OpenAI Base URL\" → `http:\u002F\u002Flocalhost:3000\u002Fopenai\u002Fv1`\n\n**[Coding Tools docs →](https:\u002F\u002Fpasteguard.com\u002Fdocs\u002Fuse-cases\u002Fcoding-tools)**\n\n## Dashboard\n\nEvery request is logged with masking details. See what was detected, what was masked, and what reached the provider.\n\n\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fsgasser_pasteguard_readme_d680249f6b90.png\" width=\"100%\" alt=\"PasteGuard Dashboard\">\n\n[localhost:3000\u002Fdashboard](http:\u002F\u002Flocalhost:3000\u002Fdashboard)\n\n## What it catches\n\n**Personal data** — Names, emails, phone numbers, credit cards, IBANs, IP addresses, locations. Powered by [Microsoft Presidio](https:\u002F\u002Fmicrosoft.github.io\u002Fpresidio\u002F). 24 languages.\n\n**Secrets** — API keys (OpenAI, Anthropic, Stripe, AWS, GitHub), SSH and PEM private keys, JWT tokens, bearer tokens, passwords, connection strings.\n\nBoth detected and masked in real time, including streaming responses.\n\n## Tech Stack\n\n[Bun](https:\u002F\u002Fbun.sh) · [Hono](https:\u002F\u002Fhono.dev) · [Microsoft Presidio](https:\u002F\u002Fmicrosoft.github.io\u002Fpresidio\u002F) · SQLite\n\n## Contributing\n\nSee [CONTRIBUTING.md](CONTRIBUTING.md) for guidelines on how to contribute.\n\n## License\n\n[Apache 2.0](LICENSE)\n","\u003Cp align=\"center\">\n  \u003Cpicture>\n    \u003Csource media=\"(prefers-color-scheme: dark)\" srcset=\"assets\u002Fwordmark-dark.svg\">\n    \u003Csource media=\"(prefers-color-scheme: light)\" srcset=\"assets\u002Fwordmark-light.svg\">\n    \u003Cimg src=\"assets\u002Fwordmark-light.svg\" width=\"220\" height=\"44\" alt=\"PasteGuard\">\n  \u003C\u002Fpicture>\n\u003C\u002Fp>\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fsgasser\u002Fpasteguard\u002Factions\u002Fworkflows\u002Fci.yml\">\u003Cimg src=\"https:\u002F\u002Fgithub.com\u002Fsgasser\u002Fpasteguard\u002Factions\u002Fworkflows\u002Fci.yml\u002Fbadge.svg\" alt=\"CI\">\u003C\u002Fa>\n  \u003Ca href=\"LICENSE\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FLicense-Apache%202.0-blue.svg\" alt=\"License\">\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fsgasser\u002Fpasteguard\u002Freleases\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fv\u002Frelease\u002Fsgasser\u002Fpasteguard\" alt=\"Release\">\u003C\u002Fa>\n\u003C\u002Fp>\n\n\u003Cp align=\"center\">\n  \u003Cstrong>AI 获取上下文。而非你的秘密。\u003C\u002Fstrong>\u003Cbr>\n  在将提示发送给 AI 之前，自动隐藏姓名、电子邮件和 API 密钥。\n\u003C\u002Fp>\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"#quick-start\">\u003Cstrong>快速开始\u003C\u002Fstrong>\u003C\u002Fa> ·\n  \u003Ca href=\"#chat\">\u003Cstrong>聊天\u003C\u002Fstrong>\u003C\u002Fa> ·\n  \u003Ca href=\"#coding-tools\">\u003Cstrong>编码工具\u003C\u002Fstrong>\u003C\u002Fa> ·\n  \u003Ca href=\"https:\u002F\u002Fpasteguard.com\u002Fdocs\">\u003Cstrong>文档\u003C\u002Fstrong>\u003C\u002Fa>\n\u003C\u002Fp>\n\n\u003Cbr\u002F>\n\n\u003Cpicture>\n  \u003Csource media=\"(prefers-color-scheme: dark)\" srcset=\"assets\u002Fcomparison-dark.png\">\n  \u003Csource media=\"(prefers-color-scheme: light)\" srcset=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fsgasser_pasteguard_readme_fd745a0cfea4.png\">\n  \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fsgasser_pasteguard_readme_fd745a0cfea4.png\" width=\"100%\" alt=\"PasteGuard — 无 vs. 有：在数据到达 AI 之前屏蔽姓名、电子邮件和 API 密钥\">\n\u003C\u002Fpicture>\n\n\u003Cp align=\"center\">\n  检测 24 种语言中的 30 多种敏感数据类型。\u003Cbr>\n  你的数据永远不会离开你的机器。\n\u003C\u002Fp>\n\n## 随处可用\n\n**[聊天](https:\u002F\u002Fpasteguard.com\u002Fdocs\u002Fuse-cases\u002Fchat)** —— 当你粘贴到 ChatGPT、Claude 和 Gemini 时，对 PII（个人身份信息）和密钥进行掩码处理。你看到的是原始内容，AI 看到的是占位符。\n\n**[应用](https:\u002F\u002Fpasteguard.com\u002Fdocs\u002Fuse-cases\u002Fapps)** —— Open WebUI、LibreChat 或任何自托管的 AI 设置。可选择将敏感请求路由到本地模型。\n\n**[编码工具](https:\u002F\u002Fpasteguard.com\u002Fdocs\u002Fuse-cases\u002Fcoding-tools)** —— Cursor、Claude Code、Copilot、Windsurf —— 你的代码库上下文流向提供商。PasteGuard 在数据离开前屏蔽密钥和个人身份信息。\n\n**[API 集成](https:\u002F\u002Fpasteguard.com\u002Fdocs\u002Fuse-cases\u002Fapi-integration)** —— 位于你的代码与 OpenAI 或 Anthropic 之间。更改一个 URL，你的用户数据就会得到保护。\n\n## 快速开始\n\n将 PasteGuard 作为本地代理运行：\n\n```bash\ndocker run --rm -p 3000:3000 ghcr.io\u002Fsgasser\u002Fpasteguard:en\n```\n\n将你的工具或应用指向 PasteGuard 而不是提供商：\n\n| API | PasteGuard 地址 | 原始地址 |\n|----------|----------------|--------------|\n| OpenAI | `http:\u002F\u002Flocalhost:3000\u002Fopenai\u002Fv1` | `https:\u002F\u002Fapi.openai.com\u002Fv1` |\n| Anthropic | `http:\u002F\u002Flocalhost:3000\u002Fanthropic` | `https:\u002F\u002Fapi.anthropic.com` |\n\n```python\n# 一行代码保护你的数据\nclient = OpenAI(base_url=\"http:\u002F\u002Flocalhost:3000\u002Fopenai\u002Fv1\")\n```\n\n\u003Cdetails>\n\u003Csummary>\u003Cstrong>欧洲语言\u003C\u002Fstrong>\u003C\u002Fsummary>\n\n对于德语、西班牙语、法语、意大利语、荷兰语、波兰语、葡萄牙语和罗马尼亚语：\n\n```bash\ndocker run --rm -p 3000:3000 ghcr.io\u002Fsgasser\u002Fpasteguard:eu\n```\n\n如需自定义配置、持久化日志或其他语言：**[阅读文档 →](https:\u002F\u002Fpasteguard.com\u002Fdocs\u002Finstallation)**\n\n\u003C\u002Fdetails>\n\n\u003Cdetails>\n\u003Csummary>\u003Cstrong>路由模式\u003C\u002Fstrong>\u003C\u002Fsummary>\n\n路由模式将包含敏感数据的请求发送到本地大型语言模型 (LLM)（如 Ollama、vLLM、llama.cpp）。其他请求则发送到 OpenAI 或 Anthropic。敏感数据保留在你的网络内。\n\n**[路由模式文档 →](https:\u002F\u002Fpasteguard.com\u002Fdocs\u002Fconcepts\u002Froute-mode)**\n\n\u003C\u002Fdetails>\n\n## 聊天\n\n用于 ChatGPT、Claude 和 Gemini 的开源浏览器扩展。\n\n- 粘贴客户数据 → 在到达 AI 之前被掩码\n- AI 使用占位符响应 → 你看到原始内容\n- 使用与代理相同的检测引擎\n\n目前处于测试版阶段。Apache 2.0 许可。\n\n**[加入测试版 →](https:\u002F\u002Ftally.so\u002Fr\u002FJ9pNLr)** · **[聊天文档 →](https:\u002F\u002Fpasteguard.com\u002Fdocs\u002Fuse-cases\u002Fchat)**\n\n## 编码工具\n\n在使用 AI 编码助手时保护你的代码库上下文和密钥。\n\n**Claude Code:**\n\n```bash\nANTHROPIC_BASE_URL=http:\u002F\u002Flocalhost:3000\u002Fanthropic claude\n```\n\n**Cursor:** 设置 → 模型 → 启用“覆盖 OpenAI 基础 URL\" → `http:\u002F\u002Flocalhost:3000\u002Fopenai\u002Fv1`\n\n**[编码工具文档 →](https:\u002F\u002Fpasteguard.com\u002Fdocs\u002Fuse-cases\u002Fcoding-tools)**\n\n## 仪表盘\n\n每个请求都会记录脱敏详情。查看检测到了什么、屏蔽了什么以及什么到达了提供商。\n\n\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fsgasser_pasteguard_readme_d680249f6b90.png\" width=\"100%\" alt=\"PasteGuard 仪表盘\">\n\n[localhost:3000\u002Fdashboard](http:\u002F\u002Flocalhost:3000\u002Fdashboard)\n\n## 它能捕获什么\n\n**个人数据** —— 姓名、电子邮件、电话号码、信用卡、IBAN（国际银行账户号码）、IP 地址、位置。由 [Microsoft Presidio](https:\u002F\u002Fmicrosoft.github.io\u002Fpresidio\u002F)（隐私数据保护工具）驱动。支持 24 种语言。\n\n**密钥** —— API 密钥（OpenAI、Anthropic、Stripe、AWS、GitHub）、SSH 和 PEM 私钥、JWT 令牌、Bearer 令牌、密码、连接字符串。\n\n两者均在实时检测并掩码，包括流式响应。\n\n## 技术栈\n\n[Bun](https:\u002F\u002Fbun.sh) · [Hono](https:\u002F\u002Fhono.dev) · [Microsoft Presidio](https:\u002F\u002Fmicrosoft.github.io\u002Fpresidio\u002F) · SQLite\n\n## 贡献\n\n有关如何贡献的指南，请参见 [CONTRIBUTING.md](CONTRIBUTING.md)。\n\n## 许可证\n\n[Apache 2.0](LICENSE)","# PasteGuard 快速上手指南\n\nPasteGuard 是一个本地隐私保护代理，能在将提示词发送给 AI 之前，自动检测并隐藏姓名、邮箱、API 密钥等敏感信息。所有数据处理均在本地完成，确保您的数据不会泄露。\n\n## 环境准备\n\n- **操作系统**: Windows, macOS, Linux\n- **前置依赖**: 已安装 [Docker](https:\u002F\u002Fwww.docker.com\u002F) 环境\n- **网络要求**: 能够访问 GitHub Container Registry (ghcr.io)\n\n## 安装步骤\n\n使用 Docker 运行 PasteGuard 作为本地代理。默认启动英文语言模型（支持 24 种语言）。\n\n```bash\ndocker run --rm -p 3000:3000 ghcr.io\u002Fsgasser\u002Fpasteguard:en\n```\n\n> **注意**: 如果您需要支持德语、西班牙语、法语等其他欧洲语言，请使用以下命令：\n> ```bash\n> docker run --rm -p 3000:3000 ghcr.io\u002Fsgasser\u002Fpasteguard:eu\n> ```\n\n## 基本使用\n\n### 配置 API 连接\n\n将您的 AI 应用或代码中的 API 地址指向 PasteGuard 的本地代理地址，而非原始服务商地址。\n\n| API 服务商 | PasteGuard 地址 | 原始地址 |\n| :--- | :--- | :--- |\n| OpenAI | `http:\u002F\u002Flocalhost:3000\u002Fopenai\u002Fv1` | `https:\u002F\u002Fapi.openai.com\u002Fv1` |\n| Anthropic | `http:\u002F\u002Flocalhost:3000\u002Fanthropic` | `https:\u002F\u002Fapi.anthropic.com` |\n\n**Python 示例：**\n```python\n# 一行代码即可保护数据\nclient = OpenAI(base_url=\"http:\u002F\u002Flocalhost:3000\u002Fopenai\u002Fv1\")\n```\n\n**CLI 工具示例 (Claude Code)：**\n```bash\nANTHROPIC_BASE_URL=http:\u002F\u002Flocalhost:3000\u002Fanthropic claude\n```\n\n### 查看监控面板\n\n启动后，可通过浏览器访问本地仪表盘，查看请求日志及脱敏详情：\n- 访问地址：[http:\u002F\u002Flocalhost:3000\u002Fdashboard](http:\u002F\u002Flocalhost:3000\u002Fdashboard)\n\n### 支持的编码工具\n\n- **Cursor**: 设置 → Models → 启用 \"Override OpenAI Base URL\" → 填入 `http:\u002F\u002Flocalhost:3000\u002Fopenai\u002Fv1`\n- **ChatGPT \u002F Claude \u002F Gemini**: 配合开源浏览器扩展使用（Beta 版），粘贴数据前自动脱敏。\n\n---\n*更多高级功能（如路由模式）请参考 [官方文档](https:\u002F\u002Fpasteguard.com\u002Fdocs)*","某后端工程师在排查线上故障时，需要将包含真实用户 ID 和支付密钥的代码片段发送给 AI 助手以获取修复建议。\n\n### 没有 pasteguard 时\n- 直接复制含敏感信息的代码可能导致用户隐私数据或 API 密钥泄露到公共云端。\n- 每次提问前必须人工逐行审查并抹除关键信息，严重拖慢调试进度。\n- 难以确认第三方 AI 服务商是否会利用这些数据训练模型，存在合规隐患。\n- 团队缺乏统一的安全防护层，个人疏忽极易造成大规模数据安全事故。\n\n### 使用 pasteguard 后\n- pasteguard 作为本地代理自动拦截请求，实时将姓名、邮箱和密钥替换为占位符。\n- 开发者无需中断工作流手动脱敏，AI 依然能理解完整的业务逻辑上下文。\n- 所有原始数据保留在本地机器，只有脱敏后的文本发送至大模型接口。\n- 支持自定义规则检测 30 多种敏感数据类型，大幅降低人为失误带来的安全风险。\n\npasteguard 在保障数据主权的前提下，让开发者能够毫无顾虑地享受 AI 编程带来的效率红利。","https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fsgasser_pasteguard_fd745a0c.png","sgasser","Stefan Gasser","https:\u002F\u002Foss.gittoolsai.com\u002Favatars\u002Fsgasser_d93f758c.png","Full-Stack Developer | Creator of @autoidle","Self-employed","Bolzano, Italy","stefan@stefangasser.com",null,"stefangasser.com","https:\u002F\u002Fgithub.com\u002Fsgasser",[86,90,94],{"name":87,"color":88,"percentage":89},"TypeScript","#3178c6",96.8,{"name":91,"color":92,"percentage":93},"Python","#3572A5",2.2,{"name":95,"color":96,"percentage":97},"Dockerfile","#384d54",1,575,21,"2026-04-04T07:33:43","Apache-2.0","未说明",{"notes":104,"python":102,"dependencies":105},"通过 Docker 容器化部署为本地代理；核心功能为隐私数据脱敏与流量拦截，非模型推理服务；可选启用路由模式对接本地 LLM（如 Ollama）；敏感信息检测基于 Microsoft Presidio 引擎，支持 24 种语言；服务端基于 Bun 运行，无需 Python 环境",[106,107,108,109],"Bun","Hono","Microsoft Presidio","SQLite",[53,13,26,51],[112,113,114,115,116,117,118,119,120,121,122,123,124],"llm","openai","pii","presidio","privacy","anthropic","claude","browser-extension","chatgpt","data-protection","open-webui","secrets","security","2026-03-27T02:49:30.150509","2026-04-06T07:11:49.746358",[128,133,137,142,146,151],{"id":129,"question_zh":130,"answer_zh":131,"source_url":132},2263,"如何在 LibreChat 中通过 PasteGuard 配置 Anthropic API？","无需修改代码即可实现。请在 PasteGuard 的 config.yaml 中将 openai provider 的 base_url 设置为 Anthropic 的 API 地址（如 https:\u002F\u002Fapi.anthropic.com\u002Fv1）。在 LibreChat 的 librechat.yaml 中，自定义 endpoint 使用 OpenAI 兼容格式（baseURL: http:\u002F\u002Flocalhost:3000\u002Fopenai\u002Fv1），不要指定 `provider: anthropic`，直接填入 Anthropic 的 API Key 和模型列表。这样可获得完整的 PII 检测和日志记录功能。","https:\u002F\u002Fgithub.com\u002Fsgasser\u002Fpasteguard\u002Fissues\u002F65",{"id":134,"question_zh":135,"answer_zh":136,"source_url":132},2264,"为什么 LibreChat 调用 Anthropic 时 PasteGuard 无法屏蔽敏感信息？","LibreChat 无论配置如何都始终使用 `\u002Fchat\u002Fcompletions` 路径，这会绕过 PasteGuard 的默认屏蔽逻辑。目前的变通方案是通过 OpenAI 兼容端点转发请求。项目计划后续添加对 `\u002Fanthropic\u002Fv1\u002Fchat\u002Fcompletions` 路径的支持以解决此问题。",{"id":138,"question_zh":139,"answer_zh":140,"source_url":141},2265,"PasteGuard 支持检测哪些类型的敏感凭证和环境变量？","除了私钥、特定 API 密钥和 Token 外，系统还支持检测常见环境变量模式。包括：1. `ENV_PASSWORD`：检测 `_PASSWORD` 和 `_PWD` 后缀的密码变量；2. `ENV_SECRET`：检测 `_SECRET` 后缀的通用秘密变量（排除已覆盖的 API 密钥）；3. `CONNECTION_STRING`：检测包含用户密码的数据库连接 URL。","https:\u002F\u002Fgithub.com\u002Fsgasser\u002Fpasteguard\u002Fissues\u002F15",{"id":143,"question_zh":144,"answer_zh":145,"source_url":141},2266,"PasteGuard 能否检测数据库连接字符串中嵌入的密码？","可以。它支持检测包含 `user:password@host` 格式的数据库连接 URL，且不受变量名限制。支持的协议包括 postgres\u002Fpostgresql, mysql, mariadb, mongodb\u002Fmongodb+srv, redis, amqp\u002Famqps。",{"id":147,"question_zh":148,"answer_zh":149,"source_url":150},2267,"Docker 运行 PasteGuard 时出现 SIGILL 错误或 FATAL 状态如何解决？","这是由于 Bun 二进制文件需要 AVX2 指令集而部分硬件（如 Intel Atom C3558R）不支持导致的。解决方法是从修复分支构建镜像，该分支使用基线构建版本（Baseline Builds），仅需 SSE4.2 支持。具体步骤：克隆仓库后检出分支 `fix\u002Fbun-baseline-avx2-issue-70`，然后执行 `docker build` 构建测试镜像。","https:\u002F\u002Fgithub.com\u002Fsgasser\u002Fpasteguard\u002Fissues\u002F70",{"id":44,"question_zh":152,"answer_zh":153,"source_url":154},"PasteGuard 支持哪些第三方大模型提供商？","PasteGuard 支持任何 OpenAI 兼容的 API 接口。官方明确支持的包括 OpenAI、Azure、OpenRouter、Groq、Together AI 等。对于非 OpenAI 兼容的 API（如 Gemini），需确认其是否兼容 OpenAI 格式才能正常工作。","https:\u002F\u002Fgithub.com\u002Fsgasser\u002Fpasteguard\u002Fissues\u002F21",[156,161,166,171,176,181,186,191,196],{"id":157,"version":158,"summary_zh":159,"released_at":160},101802,"v0.3.5","## Bug Fixes\n\n- Fix container UID from 1001 to 1000 to match typical Linux host users, resolving permission denied errors when bind-mounting volumes (#77, fixes #76)","2026-03-13T07:54:52",{"id":162,"version":163,"summary_zh":164,"released_at":165},101803,"v0.3.4","## Features\n\n⏱️ **Configurable request timeout** — The request timeout is now configurable via `server.request_timeout` in `config.yaml` (#78)\n\n```yaml\nserver:\n  request_timeout: 600  # seconds (default: 600, 0 = no timeout)\n```\n\nThe default has been increased from 120 to 600 seconds (matching OpenAI and Anthropic SDK defaults). Set to `0` to disable the timeout entirely.\n\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fsgasser\u002Fpasteguard\u002Fcompare\u002Fv0.3.3...v0.3.4","2026-03-04T08:30:31",{"id":167,"version":168,"summary_zh":169,"released_at":170},101804,"v0.3.3","## Fixes\n\n🔧 **Prompt caching restored** — Fixed Zod schemas silently stripping `cache_control` fields from Anthropic requests (#74)\n\nWithout `.passthrough()` on nested schemas, fields like `cache_control: { type: \"ephemeral\" }` were removed before forwarding to the API. This caused all requests to be billed as uncached, significantly increasing token costs.\n\n### Also in this release\n\n- Added `.passthrough()` to OpenAI schemas for consistency (preserves `name`, `tool_calls`, `audio` fields)\n- Added regression tests for both providers\n\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fsgasser\u002Fpasteguard\u002Fcompare\u002Fv0.3.2...v0.3.3","2026-02-27T18:59:12",{"id":172,"version":173,"summary_zh":174,"released_at":175},101805,"v0.3.2","## Fixes\n\n🔧 **SIGILL crash on CPUs without AVX2** — Fixed crash on older\u002Flow-power x86_64 CPUs like Intel Atom C3558R (#70)\n\nThe Docker image now uses Bun's baseline build which only requires SSE4.2, supporting CPUs without AVX2 instructions.\n\n### Breaking Change\n\n**Volume mount paths changed** from `\u002Fapp\u002F` to `\u002Fpasteguard\u002F`:\n\n```yaml\n# Before\nvolumes:\n  - .\u002Fconfig.yaml:\u002Fapp\u002Fconfig.yaml:ro\n  - .\u002Fdata:\u002Fapp\u002Fdata\n\n# After\nvolumes:\n  - .\u002Fconfig.yaml:\u002Fpasteguard\u002Fconfig.yaml:ro\n  - .\u002Fdata:\u002Fpasteguard\u002Fdata\n```\n\n### Other Changes\n\n- Container now runs as non-root user (UID 1001) for improved security\n- Fixed compatibility with updated `presidio-analyzer` base image\n\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fsgasser\u002Fpasteguard\u002Fcompare\u002Fv0.3.1...v0.3.2","2026-02-20T20:58:19",{"id":177,"version":178,"summary_zh":179,"released_at":180},101806,"v0.3.1","## Fixes\n\n🔧 **Missing Presidio Recognizers** — Fixed detection failures for URL, US_SSN, CRYPTO and other entity types (#67)\n\nThe Presidio config generator was only including 6 recognizers, missing standard ones like `UrlRecognizer`, `UsSsnRecognizer`, `CryptoRecognizer`. Users who enabled these entity types in their config would not get any detection results.\n\n### Changes\n\n- Added global recognizers for pattern-based detection (7 recognizers)\n- Added language-specific recognizers that load only when that language is configured:\n  - EN: US + UK recognizers (SSN, driver license, passport, etc.)\n  - ES: Spanish NIF\u002FNIE\n  - IT: Italian documents (fiscal code, driver license, etc.)\n  - PL: Polish PESEL\n  - KO: Korean RRN\n\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fsgasser\u002Fpasteguard\u002Fcompare\u002Fv0.3.0...v0.3.1","2026-02-09T08:11:28",{"id":182,"version":183,"summary_zh":184,"released_at":185},101807,"v0.3.0","## Highlights\n\n🔧 **Generic Mask API** — New `\u002Fapi\u002Fmask` endpoint for standalone text masking without chat context\n\n📊 **Dashboard Source Tracking** — See which endpoint (OpenAI, Anthropic, Mask API) each request came from\n\n🏷️ **Cleaner Placeholders** — Secrets now use `[[TYPE_N]]` format (e.g., `[[API_KEY_SK_1]]`) matching PII placeholders\n\n## New Features\n\n### Generic Mask API\n\nMask any text without the chat completions structure:\n\n```bash\ncurl -X POST http:\u002F\u002Flocalhost:3000\u002Fapi\u002Fmask \\\n  -H \"Content-Type: application\u002Fjson\" \\\n  -d '{\"text\": \"Contact John at john@example.com\"}'\n```\n\nResponse:\n```json\n{\n  \"masked_text\": \"Contact [[PERSON_1]] at [[EMAIL_ADDRESS_1]]\",\n  \"entities\": [...]\n}\n```\n\n### Dashboard Improvements\n\n- Source column shows request origin (OpenAI, Anthropic, Mask API)\n- Better visibility into which integration is being used\n\n## Breaking Changes\n\n### Placeholder Format for Secrets\n\nSecrets now use `[[TYPE_N]]` format instead of `[[SECRET_MASKED_TYPE_N]]`:\n\n| Before | After |\n|--------|-------|\n| `[[SECRET_MASKED_API_KEY_SK_1]]` | `[[API_KEY_SK_1]]` |\n| `[[SECRET_MASKED_PEM_PRIVATE_KEY_1]]` | `[[PEM_PRIVATE_KEY_1]]` |\n\n### Secret Type Renamed\n\n`API_KEY_OPENAI` → `API_KEY_SK` (now covers OpenAI, Anthropic, Stripe, RevenueCat)\n\nUpdate your config if you explicitly list secret entities:\n\n```yaml\nsecrets_detection:\n  entities:\n    - API_KEY_SK  # was: API_KEY_OPENAI\n```\n\n## Fixes\n\n- Increased Presidio startup timeout for multi-language images\n- Fixed entity extraction for consistent API responses\n\n## Documentation\n\n📚 Full documentation: https:\u002F\u002Fpasteguard.com\u002Fdocs\n\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fsgasser\u002Fpasteguard\u002Fcompare\u002Fv0.2.1...v0.3.0\n","2026-01-26T10:28:14",{"id":187,"version":188,"summary_zh":189,"released_at":190},101808,"v0.2.1","### Changed\n- Replaced generic \"provider\" and \"LLM\" terminology with \"OpenAI or Anthropic\"\n- Improved phrasing: \"before they reach the API\"\n- README: demo gif at top, dashboard in Quick Start section\n\n### Fixed\n- Version test now uses regex instead of hardcoded version","2026-01-20T22:40:38",{"id":192,"version":193,"summary_zh":194,"released_at":195},101809,"v0.2.0","Anthropic API support is here! PasteGuard now works with Claude, Claude Code, and the full Anthropic ecosystem.\n\n## Highlights\n\n🤖 **Anthropic Support** — Native `\u002Fanthropic\u002Fv1\u002Fmessages` endpoint with full PII and secrets detection\n\n🔧 **Claude Code Integration** — One environment variable: `ANTHROPIC_BASE_URL=http:\u002F\u002Flocalhost:3000\u002Fanthropic claude`\n\n🎯 **Role-Based Filtering** — New `scan_roles` config to scan only user messages, reducing false positives on system prompts\n\n✅ **Whitelist** — Exclude known text patterns (company names, product IDs) from PII masking\n\n## New Features\n\n### Anthropic API Support\n\nFull support for the Anthropic Messages API:\n\n```bash\ncurl http:\u002F\u002Flocalhost:3000\u002Fanthropic\u002Fv1\u002Fmessages \\\n  -H \"x-api-key: $ANTHROPIC_API_KEY\" \\\n  -H \"anthropic-version: 2023-06-01\" \\\n  -H \"Content-Type: application\u002Fjson\" \\\n  -d '{\"model\": \"claude-sonnet-4-20250514\", \"max_tokens\": 1024, \"messages\": [{\"role\": \"user\", \"content\": \"Hello\"}]}'\n```\n\n- Transparent header forwarding (`x-api-key`, `Authorization`, `anthropic-beta`)\n- Streaming support with real-time unmasking\n- Scans system prompts, messages, thinking blocks, and tool results\n- Works with mask mode and route mode\n\n### Role-Based Filtering\n\nScan only user-controlled content to reduce Presidio API calls and avoid false positives:\n\n```yaml\npii_detection:\n  scan_roles:\n    - user\n    - tool\n    - function\n\nsecrets_detection:\n  scan_roles:\n    - user\n    - tool\n```\n\n### Whitelist\n\nPrevent false positives on known text patterns:\n\n```yaml\nmasking:\n  whitelist:\n    - \"Acme Corp\"\n    - \"Product XYZ\"\n```\n\n## Quick Start\n\n```bash\ndocker run --rm -p 3000:3000 ghcr.io\u002Fsgasser\u002Fpasteguard:en\n```\n\n| Provider | PasteGuard URL |\n|----------|----------------|\n| OpenAI | `http:\u002F\u002Flocalhost:3000\u002Fopenai\u002Fv1` |\n| Anthropic | `http:\u002F\u002Flocalhost:3000\u002Fanthropic` |\n\nDashboard: http:\u002F\u002Flocalhost:3000\u002Fdashboard\n\n## Breaking Changes\n\nNone. Fully backward compatible with v0.1.0.\n\n## Documentation\n\n📚 Full documentation: https:\u002F\u002Fpasteguard.com\u002Fdocs\n\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fsgasser\u002Fpasteguard\u002Fcompare\u002Fv0.1.0...v0.2.0","2026-01-20T22:13:31",{"id":197,"version":198,"summary_zh":199,"released_at":200},101810,"v0.1.0","The first public release of PasteGuard — a privacy proxy for LLMs that masks personal data and secrets before sending to your provider.\n\n## Highlights\n\n🔒 **PII Detection** — Detect and mask names, emails, phone numbers, credit cards, IBANs, IP addresses, and locations using Microsoft Presidio\n\n🔑 **Secrets Detection** — Catch private keys, API keys (OpenAI, AWS, GitHub), JWT tokens, and credentials before they reach the LLM\n\n🌍 **24 Languages** — Automatic language detection with support for English, German, French, Spanish, and 20 more\n\n🐳 **Prebuilt Docker Images** — Zero-config deployment with language-specific images\n\n📊 **Dashboard** — Real-time monitoring of protected requests\n\n🔄 **Streaming Support** — Real-time unmasking as tokens arrive from the LLM\n\n## Two Privacy Modes\n\n**Mask Mode** — Replace PII with placeholders like `[[PERSON_1]]`, send to your provider, restore in response. No local infrastructure needed.\n\n**Route Mode** — Send PII requests to a local LLM (Ollama, vLLM, llama.cpp), everything else to your cloud provider. Data never leaves your network.\n\n## Quick Start\n\n```bash\ndocker run --rm -p 3000:3000 ghcr.io\u002Fsgasser\u002Fpasteguard:en\n```\n\nPoint your app to `http:\u002F\u002Flocalhost:3000\u002Fopenai\u002Fv1` instead of `https:\u002F\u002Fapi.openai.com\u002Fv1`.\n\nDashboard: http:\u002F\u002Flocalhost:3000\u002Fdashboard\n\n### European Languages\n\nFor German, Spanish, French, Italian, Dutch, Polish, Portuguese, and Romanian:\n\n```bash\ndocker run --rm -p 3000:3000 ghcr.io\u002Fsgasser\u002Fpasteguard:eu\n```\n\n## Docker Images\n\n| Tag | Languages | Size |\n|-----|-----------|------|\n| `en` \u002F `latest` | English | ~2.7GB |\n| `eu` | English, German, Spanish, French, Italian, Dutch, Polish, Portuguese, Romanian | ~12GB |\n\nLanguages are auto-configured per image — no config changes needed.\n\n## Features\n\n### PII Detection\n- Names, emails, phone numbers\n- Credit cards, IBANs\n- IP addresses, locations\n- Configurable confidence threshold\n- Automatic language detection\n\n### Secrets Detection\n- OpenSSH and PEM private keys\n- API keys: OpenAI, AWS, GitHub\n- JWT tokens, Bearer tokens\n- Environment variable credentials\n- Configurable actions: redact, block, or route to local\n\n### Integrations\n\nWorks with any OpenAI-compatible tool:\n- OpenAI SDK (Python\u002FJS)\n- LangChain, LlamaIndex\n- Cursor, Open WebUI, LibreChat\n\n## Tech Stack\n\n- [Bun](https:\u002F\u002Fbun.sh) — JavaScript runtime\n- [Hono](https:\u002F\u002Fhono.dev) — Web framework\n- [Microsoft Presidio](https:\u002F\u002Fmicrosoft.github.io\u002Fpresidio\u002F) — PII detection\n- SQLite — Request logging\n\n## Documentation\n\n📚 Full documentation: https:\u002F\u002Fpasteguard.com\u002Fdocs\n\n## Contributors\n\nThanks to everyone who contributed to this release:\n\n- [@sgasser](https:\u002F\u002Fgithub.com\u002Fsgasser) (Stefan Gasser)\n- [@maximiliancw](https:\u002F\u002Fgithub.com\u002Fmaximiliancw) (Max Wolf)\n- [@mkroemer](https:\u002F\u002Fgithub.com\u002Fmkroemer) (Markus)\n\n## License\n\nApache 2.0\n\n---\n\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fsgasser\u002Fpasteguard\u002Fcommits\u002Fv0.1.0","2026-01-17T13:16:48"]