{ "generated_at": "2026-04-10T20:44:23-07:00", "previous_generated_at": "2026-04-10T08:42:49-07:00", "period": "composite", "source_url": { "daily": "https://github.com/trending?since=daily", "weekly": "https://github.com/trending?since=weekly", "monthly": "https://github.com/trending?since=monthly" }, "items": [ { "rank": 1, "repo_path": "affaan-m/everything-claude-code", "owner": "affaan-m", "name": "everything-claude-code", "repo_url": "https://github.com/affaan-m/everything-claude-code", "owner_url": "https://github.com/affaan-m", "description": "The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.", "language": "JavaScript", "stars_total": 150278, "forks_total": 23249, "stars_in_period": 80939, "period_label": "最近30天", "description_zh": "代理利用性能优化系统。 Claude Code、Codex、Opencode、Cursor 等的技能、直觉、记忆、安全性和研究优先的开发。", "readme_summary_zh": "这是一个面向 AI 代理运行框架的性能优化与工程化增强项目，目标是把 Claude Code、Codex、Cowork 等代理工具从“可用”提升到“可持续、高质量地用于真实产品开发”。它不只是提供零散配置，而是一套完整系统，涵盖技能与行为模式设计、记忆优化、持续学习、安全扫描、命令与规则体系、钩子机制以及 MCP 配置，并强调以研究驱动开发的方法。该项目的内容来自长期高强度实战打磨，已经在真实产品构建过程中验证，适合希望提升 AI 编程代理稳定性、效率、安全性和上下文管理能力的开发者、独立构建者和产品团队，尤其适用于将 AI 代理投入生产环境、建立标准化工作流或优化多代理协作体验的场景。", "source_period": "monthly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 12, "stars_in_period": 14569 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 8, "stars_in_period": 80939 } }, "composite_score": 0.442544, "delta_since_last_run": 541, "previous_rank": 1 }, { "rank": 2, "repo_path": "obra/superpowers", "owner": "obra", "name": "superpowers", "repo_url": "https://github.com/obra/superpowers", "owner_url": "https://github.com/obra", "description": "An agentic skills framework & software development methodology that works.", "language": "Shell", "stars_total": 145955, "forks_total": 12510, "stars_in_period": 70214, "period_label": "最近30天", "description_zh": "有效的代理技能框架和软件开发方法。", "readme_summary_zh": "Superpowers 是一套面向编码智能体的软件开发工作流，核心目标是让 AI 不再一上来就直接写代码，而是先通过对话澄清真实需求，逐步整理出可确认的规格说明，并以易于阅读的小段内容呈现给用户审核。在设计获得认可后，它会进一步生成清晰、可执行的实现计划，强调真正的红绿 TDD、YAGNI 和 DRY 等工程原则，帮助任务按正确节奏推进。随后系统会进入由子智能体驱动的开发流程，让多个代理分工完成编码、检查、评审和持续推进，从而在较长时间内也能较稳定地按照既定方案自主工作。由于这些技能会自动触发，用户几乎不需要额外操作，因此它尤其适合希望把 AI 编程从“临时补全工具”升级为“可协作、可审查、可持续执行的开发伙伴”的个人开发者、小团队，以及需要提高需求梳理、实现规划与自动执行质量的项目场景。", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 10, "stars_in_period": 2150 }, "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 2, "stars_in_period": 12052 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 14, "stars_in_period": 70214 } }, "composite_score": 0.681865, "delta_since_last_run": 674, "previous_rank": 2 }, { "rank": 3, "repo_path": "msitarzewski/agency-agents", "owner": "msitarzewski", "name": "agency-agents", "repo_url": "https://github.com/msitarzewski/agency-agents", "owner_url": "https://github.com/msitarzewski", "description": "A complete AI agency at your fingertips - From frontend wizards to Reddit community ninjas, from whimsy injectors to reality checkers. Each agent is a specialized expert with personality, processes, and proven deliverables.", "language": "Shell", "stars_total": 77883, "forks_total": 12341, "stars_in_period": 56117, "period_label": "最近30天", "description_zh": "触手可及的完整人工智能机构 - 从前端向导到 Reddit 社区忍者，从奇思妙想的注入者到现实检查者。每个代理都是一位具有个性、流程和经过验证的可交付成果的专业专家。", "readme_summary_zh": "Agency Agents 是一个面向实际工作的 AI 智能体集合项目，目标是把“AI 代理团队”直接交到用户手中。它并非简单堆叠通用提示词，而是为不同职责打造具有明确分工的专家型角色，例如前端实现、社区运营、创意增强与风险校验等，每个智能体都具备鲜明的人格设定、稳定的工作流程和可落地的交付结果。项目强调专业化、人格化、以成果为导向和可投入生产环境，能够帮助个人创作者、独立开发者、内容团队或小型业务快速组建一支持续在线的“数字团队”，尤其适合需要多角色协作、希望提升执行效率、同时又重视输出质量与流程可靠性的场景。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 15, "stars_in_period": 56117 } }, "composite_score": 0.122159, "delta_since_last_run": 247, "previous_rank": 3 }, { "rank": 4, "repo_path": "NousResearch/hermes-agent", "owner": "NousResearch", "name": "hermes-agent", "repo_url": "https://github.com/NousResearch/hermes-agent", "owner_url": "https://github.com/NousResearch", "description": "The agent that grows with you", "language": "Python", "stars_total": 52911, "forks_total": 6905, "stars_in_period": 46639, "period_label": "最近30天", "description_zh": "与您共同成长的代理商", "readme_summary_zh": "Hermes Agent 是 Nous Research 打造的一款可自我改进的 AI 智能代理，核心特点是在传统代理能力之外内置“学习闭环”：它能从实际使用经验中沉淀并生成新技能，在后续任务中持续优化这些技能，主动强化知识保存，还能检索自己的历史对话，并在跨会话交互中逐步建立对用户偏好、背景与习惯的更深层理解。项目在部署上非常灵活，既可运行于低成本 VPS，也可扩展到 GPU 集群或几乎空闲零成本的无服务器环境，而且不受本地设备限制，适合放在云端长期运行并通过如 Telegram 等方式远程交互。它同时支持多种模型与自定义推理端点，方便按成本、性能和可用性自由切换，特别适合需要长期记忆、持续学习、远程执行和避免平台锁定的个人助手、研究代理与自动化工作流场景。", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 3, "stars_in_period": 7671 }, "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 3, "stars_in_period": 26783 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 1, "stars_in_period": 46639 } }, "composite_score": 0.962823, "delta_since_last_run": 2367, "previous_rank": 5 }, { "rank": 5, "repo_path": "666ghj/MiroFish", "owner": "666ghj", "name": "MiroFish", "repo_url": "https://github.com/666ghj/MiroFish", "owner_url": "https://github.com/666ghj", "description": "A Simple and Universal Swarm Intelligence Engine, Predicting Anything. 简洁通用的群体智能引擎，预测万物", "language": "Python", "stars_total": 53280, "forks_total": 7975, "stars_in_period": 40273, "period_label": "最近30天", "description_zh": "A Simple and Universal Swarm Intelligence Engine, Predicting Anything. 简洁通用的群体智能引擎，预测万物", "readme_summary_zh": "MiroFish 是一个面向“预测与推演”场景的通用群体智能引擎，基于多智能体技术把新闻、政策草案、金融信号、分析报告甚至小说文本等“种子信息”转化为高保真的平行数字世界。在这个仿真空间中，大量具备独立人格、长期记忆和行为逻辑的智能体会持续互动、演化，用户还能以自然语言提出预测目标，并从更高视角动态注入变量，观察不同条件下的未来走向。项目的核心能力包括现实信息抽取、图谱与记忆构建、角色与环境生成、并行模拟、时序记忆更新、预测报告生成，以及与模拟世界或报告代理进行深度交互。它特别适合政策预演、舆情分析、金融趋势研判、公关试错等严肃决策场景，也适合小说结局推演、创意脑洞实验等趣味仿真场景，帮助用户在低风险甚至零风险环境中提前看到“如果发生会怎样”的结果。", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 14, "stars_in_period": 561 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 6, "stars_in_period": 40273 } }, "composite_score": 0.320764, "delta_since_last_run": 212, "previous_rank": 4 }, { "rank": 6, "repo_path": "anthropics/claude-code", "owner": "anthropics", "name": "claude-code", "repo_url": "https://github.com/anthropics/claude-code", "owner_url": "https://github.com/anthropics", "description": "Claude Code is an agentic coding tool that lives in your terminal, understands your codebase, and helps you code faster by executing routine tasks, explaining complex code, and handling git workflows - all through natural language commands.", "language": "Shell", "stars_total": 112272, "forks_total": 18740, "stars_in_period": 36740, "period_label": "最近30天", "description_zh": "Claude Code 是一种代理编码工具，它位于您的终端中，了解您的代码库，并通过执行例行任务、解释复杂代码和处理 git 工作流程来帮助您更快地编码 - 所有这些都通过自然语言命令进行。", "readme_summary_zh": "Claude Code 是一款运行在终端中的智能编程代理，能够理解整个代码库，并通过自然语言指令协助开发者更高效地完成日常开发工作。它的核心能力包括自动执行重复性编码任务、解释复杂代码逻辑、处理 Git 工作流，以及在终端、IDE 甚至 GitHub 场景中提供一致的交互体验。项目还提供了一组可扩展插件，用于增加自定义命令和专用代理能力，方便团队按需增强功能。它尤其适合需要频繁浏览和修改大型代码库、希望减少机械操作、加快排查与协作效率的个人开发者和工程团队使用。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 5, "stars_in_period": 36740 } }, "composite_score": 0.104223, "delta_since_last_run": 148, "previous_rank": 6 }, { "rank": 7, "repo_path": "bytedance/deer-flow", "owner": "bytedance", "name": "deer-flow", "repo_url": "https://github.com/bytedance/deer-flow", "owner_url": "https://github.com/bytedance", "description": "An open-source long-horizon SuperAgent harness that researches, codes, and creates. With the help of sandboxes, memories, tools, skill, subagents and message gateway, it handles different levels of tasks that could take minutes to hours.", "language": "Python", "stars_total": 60277, "forks_total": 7672, "stars_in_period": 32590, "period_label": "最近30天", "description_zh": "一个开源的长期 SuperAgent 工具，用于研究、编码和创建。借助沙箱、内存、工具、技能、子代理和消息网关，它可以处理可能需要几分钟到几小时的不同级别的任务。", "readme_summary_zh": "DeerFlow 是字节跳动开源的“超级智能体”框架，名称源自 Deep Exploration and Efficient Research Flow，面向需要把复杂任务自动拆解、执行与追踪的开发者和团队。2.0 版本为完全重写的新架构，与 1.x 的深度研究框架代码互不共享，重点通过可扩展技能体系，将子智能体、长期记忆、沙箱环境、文件系统、上下文工程以及工具调用统一编排起来，支持搜索、爬取、代码执行、研究分析等多类任务，还可结合 Claude Code、MCP Server、IM 渠道、LangSmith Tracing 和内嵌 Python 客户端进行扩展与观测。它尤其适合构建自动化研究助手、复杂工作流代理、需要安全隔离执行环境的开发平台，以及希望将检索、推理、执行和记忆整合为一体的企业级 AI 应用场景。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 5, "stars_in_period": 32590 } }, "composite_score": 0.099993, "delta_since_last_run": 95, "previous_rank": 7 }, { "rank": 8, "repo_path": "shareAI-lab/learn-claude-code", "owner": "shareAI-lab", "name": "learn-claude-code", "repo_url": "https://github.com/shareAI-lab/learn-claude-code", "owner_url": "https://github.com/shareAI-lab", "description": "Bash is all you need - A nano claude code–like 「agent harness」, built from 0 to 1", "language": "TypeScript", "stars_total": 51408, "forks_total": 8382, "stars_in_period": 26930, "period_label": "最近30天", "description_zh": "Bash 就是你所需要的 - 一个 Nano claude 代码——类似“代理工具”，从 0 到 1 构建", "readme_summary_zh": "这是一个围绕“什么才是真正的 AI Agent”展开的学习型项目，核心目的不是教人搭建提示词工作流或编排 API 调用，而是澄清代理本质上是经过训练、能够感知环境、围绕目标推理并采取行动的模型本身。README 通过回顾 DQN、OpenAI Five、AlphaStar、绝悟以及近年的 Claude、GPT、Gemini 等案例，强调真正的 agent 来自训练后的神经网络，而非外层的流程编排、规则树或拖拽式框架。项目适合想系统理解智能体概念、辨别“模型驱动代理”与“提示词管线”差异、从软件工程视角认识 LLM coding agent 的开发者、研究者和 AI 从业者阅读，用于建立更准确的技术认知与方法判断。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 12, "stars_in_period": 26930 } }, "composite_score": 0.092953, "delta_since_last_run": 118, "previous_rank": 8 }, { "rank": 9, "repo_path": "gsd-build/get-shit-done", "owner": "gsd-build", "name": "get-shit-done", "repo_url": "https://github.com/gsd-build/get-shit-done", "owner_url": "https://github.com/gsd-build", "description": "A light-weight and powerful meta-prompting, context engineering and spec-driven development system for Claude Code by TÂCHES.", "language": "JavaScript", "stars_total": 50389, "forks_total": 4199, "stars_in_period": 23111, "period_label": "最近30天", "description_zh": "TâCHES 为 Claude Code 提供的轻量级且功能强大的元提示、上下文工程和规范驱动开发系统。", "readme_summary_zh": "GSD（get-shit-done）是一个面向 Claude Code、OpenCode、Gemini CLI、Codex、Copilot、Cursor、Windsurf 等工具的轻量级元提示与上下文工程系统，核心目标是解决 AI 编程过程中常见的“上下文腐化”问题，也就是随着会话变长，模型对需求理解和输出质量逐渐下降。它通过规格驱动开发、结构化提示、状态管理和子代理编排，在后台自动补足任务所需的关键信息、约束与验证步骤，让用户只需清晰描述想做什么，就能更稳定地得到可用、可扩展的实现结果。这个项目尤其适合独立开发者、小团队、创意型产品构建者以及不想引入繁琐企业流程的人，用更少的操作获得更可靠的 AI 协作开发体验，在 Mac、Windows 和 Linux 上都能使用。", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 18, "stars_in_period": 416 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 5, "stars_in_period": 23111 } }, "composite_score": 0.296473, "delta_since_last_run": 183, "previous_rank": 9 }, { "rank": 10, "repo_path": "shanraisshan/claude-code-best-practice", "owner": "shanraisshan", "name": "claude-code-best-practice", "repo_url": "https://github.com/shanraisshan/claude-code-best-practice", "owner_url": "https://github.com/shanraisshan", "description": "practice made claude perfect", "language": "HTML", "stars_total": 35943, "forks_total": 3388, "stars_in_period": 22777, "period_label": "最近30天", "description_zh": "练习使克劳德变得完美", "readme_summary_zh": "这是一个围绕 Claude Code 使用方法整理的最佳实践项目，核心目标是把日常开发中的提示与协作方式沉淀为一套更高效的工作模式。项目强调通过 Agents、Commands 和 Skills 的组合来组织任务，并提出一种由 Command 驱动、再调用 Agent、最终落到 Skill 的编排思路，用于把复杂开发流程拆分成更清晰、可复用的步骤。它适合希望系统化使用 Claude Code 的开发者、技术团队或工作流设计者，尤其适用于需要提升代码生成、任务分解、流程编排与多人协作一致性的场景，帮助用户建立更稳定、可扩展的智能编程实践。", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 12, "stars_in_period": 1251 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 1, "stars_in_period": 22777 } }, "composite_score": 0.327251, "delta_since_last_run": 640, "previous_rank": 10 }, { "rank": 11, "repo_path": "siddharthvaddem/openscreen", "owner": "siddharthvaddem", "name": "openscreen", "repo_url": "https://github.com/siddharthvaddem/openscreen", "owner_url": "https://github.com/siddharthvaddem", "description": "Create stunning demos for free. Open-source, no subscriptions, no watermarks, and free for commercial use. An alternative to Screen Studio.", "language": "TypeScript", "stars_total": 27677, "forks_total": 1853, "stars_in_period": 19712, "period_label": "最近30天", "description_zh": "免费创建令人惊叹的演示。开源、无订阅、无水印、免费用于商业用途。 Screen Studio 的替代品。", "readme_summary_zh": "OpenScreen 是一个仍处于测试阶段的免费开源录屏与演示视频制作工具，可以看作是 Screen Studio 的轻量替代方案，面向不想支付订阅费用、但又希望快速做出美观产品演示和操作讲解视频的用户。它主打“够用且可控”，支持录制整个屏幕或指定窗口，同时采集麦克风与系统声音，并提供自动或手动缩放、缩放位置与时长调整、视频裁剪、片段修剪、分段变速、运动模糊、文字箭头图片标注，以及自定义背景、导出不同画幅比例和分辨率等核心能力。虽然它并不追求完整复刻更成熟商业产品的高级功能，但对于制作软件介绍、功能 walkthrough、教学演示、产品宣传短片和日常内容录制等场景已经相当合适，尤其适合个人开发者、小团队和预算有限但希望拥有可修改、可分发工具的人使用。", "source_period": "monthly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 6, "stars_in_period": 10077 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 7, "stars_in_period": 19712 } }, "composite_score": 0.346615, "delta_since_last_run": 256, "previous_rank": 11 }, { "rank": 12, "repo_path": "Yeachan-Heo/oh-my-codex", "owner": "Yeachan-Heo", "name": "oh-my-codex", "repo_url": "https://github.com/Yeachan-Heo/oh-my-codex", "owner_url": "https://github.com/Yeachan-Heo", "description": "OmX - Oh My codeX: Your codex is not alone. Add hooks, agent teams, HUDs, and so much more.", "language": "TypeScript", "stars_total": 20686, "forks_total": 1843, "stars_in_period": 18775, "period_label": "最近30天", "description_zh": "OmX - 哦我的 codeX：你的法典并不孤单。添加挂钩、代理团队、HUD 等等。", "readme_summary_zh": "开始增强 Codex，然后让 OMX 在工作增长时添加更好的提示、工作流程和运行时帮助。网站：https://yechan-heo.github.io/oh-my-codex-website/ 文档：入门 · 代理 · 技能 · 集成 · 演示 · OpenClaw 指南", "source_period": "monthly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 8, "stars_in_period": 7276 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 2, "stars_in_period": 18775 } }, "composite_score": 0.322494, "delta_since_last_run": 262, "previous_rank": 12 }, { "rank": 13, "repo_path": "TauricResearch/TradingAgents", "owner": "TauricResearch", "name": "TradingAgents", "repo_url": "https://github.com/TauricResearch/TradingAgents", "owner_url": "https://github.com/TauricResearch", "description": "TradingAgents: Multi-Agents LLM Financial Trading Framework", "language": "Python", "stars_total": 49323, "forks_total": 8925, "stars_in_period": 17751, "period_label": "最近30天", "description_zh": "TradingAgents：多代理LLM金融交易框架", "readme_summary_zh": "TradingAgents 是一个面向研究场景的多智能体交易框架，尝试用贴近真实投资机构的协作方式，把复杂的市场分析与交易决策拆分给不同角色的 LLM 代理来完成。系统包含基本面、情绪、新闻、技术分析师，以及多空研究员、交易员、风控团队和投资组合经理，这些代理会分别从财务表现、社交情绪、宏观事件、技术指标等角度分析市场，再通过讨论、辩论与汇总形成交易建议，并由风控评估后交给组合经理决定是否在模拟交易所执行。项目近几个版本持续增强了多语言能力、多模型提供商支持、统一模型目录、回测日期一致性、代理与跨平台稳定性，适合用于量化交易研究、智能体协作实验、交易策略原型验证和教学演示，但并不构成任何投资建议。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 13, "stars_in_period": 17751 } }, "composite_score": 0.083416, "delta_since_last_run": 129, "previous_rank": 13 }, { "rank": 14, "repo_path": "volcengine/OpenViking", "owner": "volcengine", "name": "OpenViking", "repo_url": "https://github.com/volcengine/OpenViking", "owner_url": "https://github.com/volcengine", "description": "OpenViking is an open-source context database designed specifically for AI Agents(such as openclaw). OpenViking unifies the management of context (memory, resources, and skills) that Agents need through a file system paradigm, enabling hierarchical context delivery and self-evolving.", "language": "Python", "stars_total": 21983, "forks_total": 1582, "stars_in_period": 16520, "period_label": "最近30天", "description_zh": "OpenViking是一个专门为AI Agent（例如openclaw）设计的开源上下文数据库。 OpenViking 通过文件系统范例统一了代理所需的上下文（内存、资源和技能）管理，从而实现了分层上下文交付和自我演化。", "readme_summary_zh": "OpenViking 是火山引擎推出的一套面向 AI Agent 的上下文工程与知识增强框架，主要用于解决“数据很多但高质量上下文难获取”的问题。它围绕智能体在实际应用中常见的知识组织、检索、记忆与上下文管理需求，帮助开发者把分散的数据源转化为更适合大模型消费的高质量语境，从而提升问答、推理和任务执行的准确性与稳定性。项目的核心能力通常体现在对多源知识的接入与加工、面向检索增强生成的上下文构建、与 Agent 工作流结合的记忆管理，以及支撑复杂业务场景下的信息编排与调用。它尤其适合企业知识库问答、客服助手、办公协同、业务 Copilot 以及需要长期记忆和动态知识更新的智能体应用场景，能够帮助团队更高效地构建可落地、可维护的 AI 应用。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 9, "stars_in_period": 16520 } }, "composite_score": 0.082887, "delta_since_last_run": 28, "previous_rank": 14 }, { "rank": 15, "repo_path": "mvanhorn/last30days-skill", "owner": "mvanhorn", "name": "last30days-skill", "repo_url": "https://github.com/mvanhorn/last30days-skill", "owner_url": "https://github.com/mvanhorn", "description": "AI agent skill that researches any topic across Reddit, X, YouTube, HN, Polymarket, and the web - then synthesizes a grounded summary", "language": "Python", "stars_total": 20453, "forks_total": 1672, "stars_in_period": 16503, "period_label": "最近30天", "description_zh": "AI 代理技能可研究 Reddit、X、YouTube、HN、Polymarket 和网络上的任何主题，然后综合得出有依据的摘要", "readme_summary_zh": "last30days-skill 是一个面向近30天热点研究的 AI 技能，帮助用户围绕某个主题快速梳理 Reddit、X、YouTube、Bluesky、Hacker News、Polymarket，以及 TikTok、Instagram 等平台上的最新讨论、传播内容与真实市场信号，并生成带引用的研究简报。它的核心能力包括并行搜索多源内容、智能扩展查询、去重与相关性评分、社区热度与跨平台共识分析、账号解析、自动发现合适社群，以及支持“X vs Y”式比较研究，输出优劣势对照和数据驱动结论；同时还能自动保存报告，便于积累个人研究资料库。这个项目特别适合追踪 AI 工具、提示词方法、产品动态、技术趋势或舆情变化的人使用，尤其适用于想了解“最近一个月真正被社区关注、验证和讨论的内容”而不满足于零散搜索结果的研究、内容策划和产品分析场景。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 11, "stars_in_period": 16503 } }, "composite_score": 0.082507, "delta_since_last_run": 125, "previous_rank": 16 }, { "rank": 16, "repo_path": "farion1231/cc-switch", "owner": "farion1231", "name": "cc-switch", "repo_url": "https://github.com/farion1231/cc-switch", "owner_url": "https://github.com/farion1231", "description": "A cross-platform desktop All-in-One assistant tool for Claude Code, Codex, OpenCode, openclaw & Gemini CLI.", "language": "Rust", "stars_total": 42520, "forks_total": 2681, "stars_in_period": 16378, "period_label": "最近30天", "description_zh": "一款适用于 Claude Code、Codex、OpenCode、openclaw 和 Gemini CLI 的跨平台桌面一体式辅助工具。", "readme_summary_zh": "CC Switch 是一款面向 AI 编程工作流的跨平台桌面管理工具，主要用于统一管理 Claude Code、Codex、Gemini CLI、OpenCode 和 OpenClaw 这五类常见命令行客户端，解决不同工具配置格式分散、切换 API 提供商繁琐、MCP 与 Skills 无法集中维护的问题。它提供可视化界面与 50 多个内置提供商预设，支持一键导入、快速切换、系统托盘切换、本地代理热切换、故障转移、健康监测，以及基于 SQLite 的安全配置存储；同时还能统一管理 MCP 服务器、提示词文件和 Skills，并支持双向同步、云端同步、导入导出及插件扩展相关辅助功能。这个项目尤其适合同时使用多个 AI CLI 工具的开发者、技术团队和重度自动化用户，在需要频繁更换模型服务、跨设备保持一致配置、搭建稳定 AI 编码环境时价值最明显。", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 10, "stars_in_period": 633 }, "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 6, "stars_in_period": 4055 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 2, "stars_in_period": 16378 } }, "composite_score": 0.512404, "delta_since_last_run": 160, "previous_rank": 15 }, { "rank": 17, "repo_path": "lightpanda-io/browser", "owner": "lightpanda-io", "name": "browser", "repo_url": "https://github.com/lightpanda-io/browser", "owner_url": "https://github.com/lightpanda-io", "description": "Lightpanda: the headless browser designed for AI and automation", "language": "Zig", "stars_total": 28300, "forks_total": 1194, "stars_in_period": 16279, "period_label": "最近30天", "description_zh": "Lightpanda：专为人工智能和自动化设计的无头浏览器", "readme_summary_zh": "Lightpanda 是一个专为无头使用场景打造的开源浏览器，完全从零开始开发，并非基于 Chromium 或 WebKit 的改造版本，而是使用 Zig 编写的新浏览器内核。它面向 AI 代理与自动化任务，核心能力包括执行 JavaScript、提供部分 Web API 支持，并通过 CDP 兼容 Playwright、Puppeteer 和 chromedp 等常见自动化工具。该项目强调极致性能与资源效率，具备更低的内存占用、更快的执行速度和几乎即时的启动体验，因此特别适合需要大规模网页操作的场景，如 AI 智能体驱动的浏览自动化、LLM 训练数据采集、网页抓取以及自动化测试等，尤其适用于对速度、成本和部署密度有较高要求的服务端环境。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 10, "stars_in_period": 16279 } }, "composite_score": 0.08246, "delta_since_last_run": 92, "previous_rank": 17 }, { "rank": 18, "repo_path": "microsoft/VibeVoice", "owner": "microsoft", "name": "VibeVoice", "repo_url": "https://github.com/microsoft/VibeVoice", "owner_url": "https://github.com/microsoft", "description": "Open-Source Frontier Voice AI", "language": "Python", "stars_total": 38353, "forks_total": 4426, "stars_in_period": 14865, "period_label": "最近30天", "description_zh": "开源前沿语音人工智能", "readme_summary_zh": "VibeVoice 是微软开源的一组前沿语音 AI 模型与研究框架，覆盖语音识别与语音合成两大方向，重点突出其中的 VibeVoice-ASR 和实时语音能力。项目的核心价值在于面向长音频处理：VibeVoice-ASR 可在单次推理中处理最长 60 分钟连续音频，不必像传统 ASR 那样切片，从而更好地保持全局语义连贯与说话人一致性；同时它还能输出结构化转写结果，明确“谁在什么时间说了什么”，把语音识别、说话人分离和时间戳对齐结合在一起，并支持自定义热词或上下文信息，以提升专有名词、术语和特定场景下的识别准确率。项目还具备原生多语言支持、可微调、支持更快推理集成，以及实时文本转语音模型，可处理流式文本输入并生成稳健的长段语音。因此，它尤其适合会议纪要、访谈播客、课堂录音、客服质检、多语种转写和需要结构化语音内容理解的应用场景。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 10, "stars_in_period": 14865 } }, "composite_score": 0.081019, "delta_since_last_run": 91, "previous_rank": 20 }, { "rank": 19, "repo_path": "abhigyanpatwari/GitNexus", "owner": "abhigyanpatwari", "name": "GitNexus", "repo_url": "https://github.com/abhigyanpatwari/GitNexus", "owner_url": "https://github.com/abhigyanpatwari", "description": "GitNexus: The Zero-Server Code Intelligence Engine - GitNexus is a client-side knowledge graph creator that runs entirely in your browser. Drop in a GitHub repo or ZIP file, and get an interactive knowledge graph wit a built in Graph RAG Agent. Perfect for code exploration", "language": "TypeScript", "stars_total": 26329, "forks_total": 2964, "stars_in_period": 14773, "period_label": "最近30天", "description_zh": "GitNexus：零服务器代码智能引擎 - GitNexus 是一个完全在浏览器中运行的客户端知识图创建器。放入 GitHub 存储库或 ZIP 文件，即可获得带有内置 Graph RAG 代理的交互式知识图。非常适合代码探索", "readme_summary_zh": "GitNexus 是一个面向代码库分析与 AI 编程代理协作的知识图谱工具，目标是把任意代码仓库索引成包含依赖关系、调用链、模块聚类和执行流的结构化上下文，让模型不只“读懂”代码说明，还能真正理解系统架构与影响范围。它提供 Web 界面用于快速与仓库对话，也通过 CLI 与 MCP 将深层代码视图提供给 Cursor、Claude Code、Codex 等智能代理，减少遗漏依赖、破坏调用链和盲目修改的问题；桥接模式还能让本地索引结果直接被 Web UI 发现和复用，无需重复上传或重新索引。项目尤其适合大型、多模块、跨仓库或需要高可靠 AI 辅助开发的场景，例如架构分析、代码审查、变更影响评估、测试生成和持续维护等。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 5, "stars_in_period": 14773 } }, "composite_score": 0.081832, "delta_since_last_run": 179, "previous_rank": 19 }, { "rank": 20, "repo_path": "pbakaus/impeccable", "owner": "pbakaus", "name": "impeccable", "repo_url": "https://github.com/pbakaus/impeccable", "owner_url": "https://github.com/pbakaus", "description": "The design language that makes your AI harness better at design.", "language": "JavaScript", "stars_total": 18258, "forks_total": 810, "stars_in_period": 14691, "period_label": "最近30天", "description_zh": "使您的 AI 线束在设计上更加出色的设计语言。", "readme_summary_zh": "Impeccable 是一个面向前端设计与 AI 协作的提示词/技能包项目，目标是帮助大语言模型摆脱千篇一律的网页审美和常见 UI 失误，生成更讲究、更可控的界面设计。它建立在 Anthropic 的 frontend-design 技能基础上，进一步补充了 7 份面向具体设计领域的参考规范、20 个可组合的引导命令，以及一套明确列出的反模式规则，用来告诉模型不仅该怎么做，也要避免哪些陈词滥调，例如滥用常见字体、套娃式卡片布局或在彩色背景上使用低对比灰字。项目的核心能力包括设计审计、风格统一、体验批评、上线前润色，以及对界面细节、层次、动效和表达的精炼控制，特别适合用在 AI 参与网页或产品界面设计的场景中，例如在改版前先做质量检查、在多页面项目中校准设计系统，或在发布前对视觉与交互进行最后一轮打磨。", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 5, "stars_in_period": 453 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 3, "stars_in_period": 14691 } }, "composite_score": 0.296926, "delta_since_last_run": 142, "previous_rank": 18 }, { "rank": 21, "repo_path": "FujiwaraChoki/MoneyPrinterV2", "owner": "FujiwaraChoki", "name": "MoneyPrinterV2", "repo_url": "https://github.com/FujiwaraChoki/MoneyPrinterV2", "owner_url": "https://github.com/FujiwaraChoki", "description": "Automate the process of making money online.", "language": "Python", "stars_total": 29107, "forks_total": 3143, "stars_in_period": 14315, "period_label": "最近30天", "description_zh": "自动化在线赚钱的过程。", "readme_summary_zh": "MoneyPrinterV2 是一个用于自动化在线赚钱流程的开源项目，也是 MoneyPrinter 的第二代重写版本。它以更模块化的架构为基础，扩展了原项目的能力，能够支持多种自动化营销与内容分发场景，例如结合定时调度执行 Twitter 机器人、自动化生成与发布 YouTube Shorts、开展 Amazon 与 Twitter 联动的联盟营销，以及搜索本地商家并进行冷启动外联。这个项目适合希望把内容运营、流量获取和商业转化流程尽量自动化的个人创作者、独立开发者、增长营销人员和副业探索者使用，尤其适用于需要将多个线上获客环节串联起来、持续批量运行的场景；同时它依赖 Python 3.12 运行，并且也启发了社区在不同语言中的衍生版本。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 4, "stars_in_period": 14315 } }, "composite_score": 0.081547, "delta_since_last_run": 37, "previous_rank": 22 }, { "rank": 22, "repo_path": "jarrodwatts/claude-hud", "owner": "jarrodwatts", "name": "claude-hud", "repo_url": "https://github.com/jarrodwatts/claude-hud", "owner_url": "https://github.com/jarrodwatts", "description": "A Claude Code plugin that shows what's happening - context usage, active tools, running agents, and todo progress", "language": "JavaScript", "stars_total": 18243, "forks_total": 794, "stars_in_period": 13975, "period_label": "最近30天", "description_zh": "一个 Claude Code 插件，显示正在发生的事情 - 上下文使用、活动工具、正在运行的代理和待办事项进度", "readme_summary_zh": "claude-hud 是一个面向 Claude Code 的可视化插件，用来在你编写提示或与模型协作时，持续显示当前执行状态。它会固定出现在输入框下方，集中展示上下文使用情况、正在调用的工具、运行中的代理，以及待办事项的进度，让原本不易察觉的内部流程变得更直观、可追踪。这个项目的核心价值在于提升人机协作时的透明度，帮助开发者及时了解系统正在做什么、资源消耗到什么程度，以及任务是否按预期推进。它尤其适合需要长时间交互、频繁调用工具、同时管理多个子任务的编码场景，例如调试复杂项目、执行多步骤开发流程或观察代理式工作流时使用。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 9, "stars_in_period": 13975 } }, "composite_score": 0.080293, "delta_since_last_run": 81, "previous_rank": 23 }, { "rank": 23, "repo_path": "alibaba/page-agent", "owner": "alibaba", "name": "page-agent", "repo_url": "https://github.com/alibaba/page-agent", "owner_url": "https://github.com/alibaba", "description": "JavaScript in-page GUI agent. Control web interfaces with natural language.", "language": "TypeScript", "stars_total": 16727, "forks_total": 1356, "stars_in_period": 13647, "period_label": "最近30天", "description_zh": "JavaScript 页内 GUI 代理。使用自然语言控制 Web 界面。", "readme_summary_zh": "The GUI Agent Living in Your Webpage. Control web interfaces with natural language. 🌐 English | 中文", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 4, "stars_in_period": 13647 } }, "composite_score": 0.080866, "delta_since_last_run": 24, "previous_rank": 21 }, { "rank": 24, "repo_path": "mattpocock/skills", "owner": "mattpocock", "name": "skills", "repo_url": "https://github.com/mattpocock/skills", "owner_url": "https://github.com/mattpocock", "description": "My personal directory of skills, straight from my .claude directory.", "language": "Shell", "stars_total": 13881, "forks_total": 1142, "stars_in_period": 12774, "period_label": "最近30天", "description_zh": "我的个人技能目录，直接来自我的 .claude 目录。", "readme_summary_zh": "代理技能的集合，可扩展规划、开发和工具方面的能力。这些技能可以帮助您在编写代码之前思考问题。", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 2, "stars_in_period": 354 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 3, "stars_in_period": 12774 } }, "composite_score": 0.293504, "delta_since_last_run": 149, "previous_rank": 24 }, { "rank": 25, "repo_path": "opendataloader-project/opendataloader-pdf", "owner": "opendataloader-project", "name": "opendataloader-pdf", "repo_url": "https://github.com/opendataloader-project/opendataloader-pdf", "owner_url": "https://github.com/opendataloader-project", "description": "PDF Parser for AI-ready data. Automate PDF accessibility. Open-source.", "language": "Java", "stars_total": 14884, "forks_total": 1259, "stars_in_period": 12746, "period_label": "最近30天", "description_zh": "用于 AI 就绪数据的 PDF 解析器。自动化 PDF 辅助功能。开源。", "readme_summary_zh": "opendataloader-pdf 是一个面向 AI 数据提取与 PDF 无障碍自动化的开源 PDF 解析项目，目标是把各种普通、扫描版或版式复杂的 PDF 转成更适合机器理解和后续处理的结构化内容。它可输出 Markdown、带边界框坐标的 JSON、HTML 以及带标注的版面元素结果，能够识别标题、段落、表格、图片、公式等内容；其中本地确定性模式适合稳定、可控的解析，混合 AI 模式则擅长处理多栏论文、复杂或无边框表格、低质量扫描件、OCR 文本、公式和图表说明等难题，并提供 80 多种语言识别能力。项目也支持 RAG 场景，可用于切分、检索、引用定位与 LangChain 集成；同时其版面分析引擎还服务于 PDF 自动加标签，帮助将未标记 PDF 转为可访问性更强的 Tagged PDF，特别适合文档智能处理、知识库构建、科研论文解析、企业批量归档以及需要兼顾合规与无障碍改造的场景。", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 9, "stars_in_period": 1306 }, "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 1, "stars_in_period": 3613 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 1, "stars_in_period": 12746 } }, "composite_score": 0.530205, "delta_since_last_run": 320, "previous_rank": 26 }, { "rank": 26, "repo_path": "ruvnet/ruflo", "owner": "ruvnet", "name": "ruflo", "repo_url": "https://github.com/ruvnet/ruflo", "owner_url": "https://github.com/ruvnet", "description": "🌊 The leading agent orchestration platform for Claude. Deploy intelligent multi-agent swarms, coordinate autonomous workflows, and build conversational AI systems. Features enterprise-grade architecture, distributed swarm intelligence, RAG integration, and native Claude Code / Codex Integration", "language": "TypeScript", "stars_total": 31136, "forks_total": 3485, "stars_in_period": 10964, "period_label": "最近30天", "description_zh": "🌊 Claude 领先的代理编排平台。部署智能多代理群、协调自主工作流程并构建对话式 AI 系统。具有企业级架构、分布式集群智能、RAG 集成和本机 Claude Code/Codex 集成", "readme_summary_zh": null, "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 10, "stars_in_period": 10964 } }, "composite_score": 0.077042, "delta_since_last_run": 93, "previous_rank": 27 }, { "rank": 27, "repo_path": "hesreallyhim/awesome-claude-code", "owner": "hesreallyhim", "name": "awesome-claude-code", "repo_url": "https://github.com/hesreallyhim/awesome-claude-code", "owner_url": "https://github.com/hesreallyhim", "description": "A curated list of awesome skills, hooks, slash-commands, agent orchestrators, applications, and plugins for Claude Code by Anthropic", "language": "Python", "stars_total": 37935, "forks_total": 3071, "stars_in_period": 10742, "period_label": "最近30天", "description_zh": "Anthropic 为 Claude Code 精心策划的技能、挂钩、斜线命令、代理编排器、应用程序和插件列表", "readme_summary_zh": "awesome-claude-code 是一个面向 Claude Code 用户的精选资源集合，核心目标是帮助开发者系统化提升基于 Claude Code 的开发工作流。它汇集了多类实用内容，包括技能模板、智能体、插件、钩子、斜杠命令以及配套工具，并按 IDE 集成、使用监控、编排平台、配置管理、状态展示、版本控制、代码分析与测试、上下文加载、文档与变更记录、CI 与部署、项目任务管理等场景分类。项目中既有用于观察会话日志、分析上下文与子代理执行过程的可视化工具，也有可校验并自动修复 Claude Code 相关配置与代理文件的 lint 工具，还收录了将代码库转成交互式课程、以及用于多智能体协作编排的进阶平台。它最适合希望快速搭建高质量 Claude Code 工具链、比较不同实践方案、或为团队寻找规范化与自动化工作方式的开发者与高级用户参考使用。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 16, "stars_in_period": 10742 } }, "composite_score": 0.075727, "delta_since_last_run": 67, "previous_rank": 28 }, { "rank": 28, "repo_path": "openai/codex", "owner": "openai", "name": "codex", "repo_url": "https://github.com/openai/codex", "owner_url": "https://github.com/openai", "description": "Lightweight coding agent that runs in your terminal", "language": "Rust", "stars_total": 74454, "forks_total": 10507, "stars_in_period": 10498, "period_label": "最近30天", "description_zh": "在您的终端中运行的轻量级编码代理", "readme_summary_zh": "OpenAI 的 Codex CLI 是一款在本地电脑上运行的命令行编码代理，面向希望直接在终端中借助 AI 提升开发效率的用户。它的主要用途是把自然语言需求转化为实际的软件开发操作，在本地环境中协助理解代码、生成与修改文件、执行与迭代编程任务，从而让开发者能够以更少的手工切换完成编码流程。由于它以 CLI 形式工作，尤其适合习惯终端、重视本地工作流与操作效率的工程师，用于日常编程辅助、快速原型开发、代码维护以及在现有项目中进行交互式开发都比较合适。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 3, "stars_in_period": 10498 } }, "composite_score": 0.077838, "delta_since_last_run": 83, "previous_rank": 29 }, { "rank": 29, "repo_path": "langchain-ai/deepagents", "owner": "langchain-ai", "name": "deepagents", "repo_url": "https://github.com/langchain-ai/deepagents", "owner_url": "https://github.com/langchain-ai", "description": "Agent harness built with LangChain and LangGraph. Equipped with a planning tool, a filesystem backend, and the ability to spawn subagents - well-equipped to handle complex agentic tasks.", "language": "Python", "stars_total": 20273, "forks_total": 2813, "stars_in_period": 10114, "period_label": "最近30天", "description_zh": "使用 LangChain 和 LangGraph 构建的代理工具。配备了规划工具、文件系统后端以及生成子代理的能力——足以处理复杂的代理任务。", "readme_summary_zh": "Deep Agents 是 LangChain 提供的一个开箱即用的智能体运行框架，目标是让开发者不必从零拼装提示词、工具调用和上下文管理，就能迅速得到可用的 AI Agent，并在此基础上按需扩展。它内置任务规划与待办拆解、文件系统读写与搜索、带沙箱的命令执行、可隔离上下文的子智能体协作，以及长对话自动摘要和大输出落盘等能力，还提供教会模型高效使用这些工具的默认提示。项目既可作为终端中的预构建编码助手使用，支持交互式界面、联网搜索和适合脚本或 CI 的无头模式，也可作为基于 LangGraph 的可编排底座，方便接入自定义工具、替换模型、配置记忆、审批流和远程沙箱。它尤其适合开发、调试、部署 AI 代理和 LLM 应用，以及需要在代码、文件和命令行环境中完成复杂多步骤任务的场景。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 11, "stars_in_period": 10114 } }, "composite_score": 0.075995, "delta_since_last_run": 49, "previous_rank": 30 }, { "rank": 30, "repo_path": "Fission-AI/OpenSpec", "owner": "Fission-AI", "name": "OpenSpec", "repo_url": "https://github.com/Fission-AI/OpenSpec", "owner_url": "https://github.com/Fission-AI", "description": "Spec-driven development (SDD) for AI coding assistants.", "language": "TypeScript", "stars_total": 39002, "forks_total": 2651, "stars_in_period": 10008, "period_label": "最近30天", "description_zh": "AI 编码助手的规范驱动开发 (SDD)。", "readme_summary_zh": "OpenSpec 是一个面向想法梳理与方案定义的工作流工具，现已重构为“由产物驱动”的新流程，核心用途是把一句模糊的创意逐步转化为结构化、可讨论的规范或提案。用户可以从输入一个想法开始，由系统围绕提案生成和组织关键产物，帮助团队更清晰地明确目标、需求与后续执行方向。它特别适合产品规划、功能设计、项目立项和团队协作等场景，尤其适用于需要把早期概念快速沉淀为可共享、可评审文档的个人或团队，从而降低沟通成本并提升方案成形效率。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 9, "stars_in_period": 10008 } }, "composite_score": 0.076249, "delta_since_last_run": 75, "previous_rank": 31 }, { "rank": 31, "repo_path": "hacksider/Deep-Live-Cam", "owner": "hacksider", "name": "Deep-Live-Cam", "repo_url": "https://github.com/hacksider/Deep-Live-Cam", "owner_url": "https://github.com/hacksider", "description": "real time face swap and one-click video deepfake with only a single image", "language": "Python", "stars_total": 89532, "forks_total": 13007, "stars_in_period": 10005, "period_label": "最近30天", "description_zh": "仅使用单个图像进行实时换脸和一键视频深度换脸", "readme_summary_zh": "Deep-Live-Cam 是一个面向 AI 生成媒体创作的实时换脸与视频深度伪造工具，主打“单击操作”和“仅需一张人脸图片”即可完成面部替换，适合希望快速生成角色演绎、短视频内容或服装展示效果的创作者与视觉艺术从业者使用。项目的核心能力在于将目标人脸高效映射到实时画面或视频中，以较低门槛实现自定义角色动画和内容制作，同时强调将其作为提升生产力的创作工具而非无约束的伪造系统。README 也明确提出了严格的伦理与合规立场，内置对裸体、血腥画面及战争影像等敏感内容的检测与拦截机制，并要求用户在使用真实人物面孔时事先取得同意、公开传播时明确标注为深度伪造内容；如法律要求，项目还可能停止服务或对输出添加水印，因此它最适合用于合法、经授权且需要高效率人脸驱动生成的内容创作场景。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 12, "stars_in_period": 10005 } }, "composite_score": 0.075702, "delta_since_last_run": 20, "previous_rank": 32 }, { "rank": 32, "repo_path": "andrewyng/context-hub", "owner": "andrewyng", "name": "context-hub", "repo_url": "https://github.com/andrewyng/context-hub", "owner_url": "https://github.com/andrewyng", "description": "暂无简介", "language": "JavaScript", "stars_total": 12792, "forks_total": 1120, "stars_in_period": 9948, "period_label": "最近30天", "description_zh": "暂无简介", "readme_summary_zh": "编码代理会对 API 产生幻觉，并忘记他们在会话中学到的内容。 Context Hub 为他们提供精心策划的版本化文档，并让他们能够更智能地完成每项任务。此存储库中的所有内容都是开放的并以降价形式维护 - 您可以准确检查代理阅读的内容并做出贡献。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 4, "stars_in_period": 9948 } }, "composite_score": 0.077096, "delta_since_last_run": 9, "previous_rank": 25 }, { "rank": 33, "repo_path": "manaflow-ai/cmux", "owner": "manaflow-ai", "name": "cmux", "repo_url": "https://github.com/manaflow-ai/cmux", "owner_url": "https://github.com/manaflow-ai", "description": "Ghostty-based macOS terminal with vertical tabs and notifications for AI coding agents", "language": "Swift", "stars_total": 13553, "forks_total": 979, "stars_in_period": 8751, "period_label": "最近30天", "description_zh": "基于 Ghostty 的 macOS 终端，带有垂直选项卡和 AI 编码代理通知", "readme_summary_zh": "cmux 是一款面向 macOS 的终端工具，基于 Ghostty 技术栈构建，特别适合与 AI 编程代理协同工作。它提供纵向标签页与通知机制，便于同时管理多个任务和会话；同时具备可脚本化能力，可通过命令行和 socket API 创建工作区、分割面板、发送按键并自动化浏览器操作，方便将终端纳入开发流程或智能代理编排中。作为原生 Swift 与 AppKit 应用，它不依赖 Electron，启动更快、内存占用更低，并通过 libghostty 获得 GPU 加速渲染，保证界面流畅；此外还能兼容现有 Ghostty 配置，直接复用主题、字体和颜色设置。它尤其适合需要多终端并行、自动化控制、追求性能与原生体验的 macOS 开发者和 AI 辅助编码场景。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 1, "stars_in_period": 8751 } }, "composite_score": 0.07642, "delta_since_last_run": 53, "previous_rank": 34 }, { "rank": 34, "repo_path": "onyx-dot-app/onyx", "owner": "onyx-dot-app", "name": "onyx", "repo_url": "https://github.com/onyx-dot-app/onyx", "owner_url": "https://github.com/onyx-dot-app", "description": "Open Source AI Platform - AI Chat with advanced features that works with every LLM", "language": "Python", "stars_total": 26442, "forks_total": 3514, "stars_in_period": 8661, "period_label": "最近30天", "description_zh": "开源人工智能平台-具有高级功能的人工智能聊天，适用于每个法学硕士", "readme_summary_zh": "Onyx 是一个开源、可自托管的 AI 平台，核心形态是可对接任意大语言模型的聊天式工作界面，既支持 OpenAI、Anthropic、Gemini 等云端模型，也兼容 Ollama、vLLM 等自部署模型，并且能够在完全隔离外网的环境中运行。它面向企业知识问答、内部智能助手和研究分析等场景，提供较完整的能力组合，包括可定制的 AI Agents、网页搜索、基于混合检索与知识图谱的 RAG、覆盖 40 多种应用的数据连接器、多步骤深度研究、MCP 与外部系统动作调用、代码解释执行、图像生成，以及分享协作、反馈收集、用户管理和使用分析等功能。对于希望把分散知识源、业务系统与多种模型统一到同一平台中的团队，Onyx 尤其适合用来构建安全、可控、可扩展的企业级 AI 助手。", "source_period": "monthly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 10, "stars_in_period": 3804 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 18, "stars_in_period": 8661 } }, "composite_score": 0.279697, "delta_since_last_run": 106, "previous_rank": 35 }, { "rank": 35, "repo_path": "f/prompts.chat", "owner": "f", "name": "prompts.chat", "repo_url": "https://github.com/f/prompts.chat", "owner_url": "https://github.com/f", "description": "f.k.a. Awesome ChatGPT Prompts. Share, discover, and collect prompts from the community. Free and open source — self-host for your organization with complete privacy.", "language": "HTML", "stars_total": 159192, "forks_total": 20847, "stars_in_period": 8360, "period_label": "最近30天", "description_zh": "F.K.A.很棒的 ChatGPT 提示。分享、发现并收集社区中的提示。免费和开源——为您的组织自行托管，并具有完全的隐私。", "readme_summary_zh": "prompts.chat 是一个面向各类 AI 对话模型的开源提示词库，前身为 Awesome ChatGPT Prompts，现已扩展到可适配 ChatGPT、Claude、Gemini、Llama、Mistral 等主流助手。项目的核心价值在于集中整理高质量提示词示例，帮助用户快速复用成熟的提问模板，同时提供系统化的提示工程学习内容，涵盖从基础方法到链式思维、少样本学习和 AI 代理等进阶技巧；此外还包含适合儿童的互动式 AI 沟通教学，以及可自托管的私有提示词库方案，支持品牌定制、主题配置和身份认证，并可作为 MCP server 集成到其他 AI 工具中。它尤其适合希望提升 AI 使用效率的个人用户、内容创作者、教育场景、团队知识库建设，以及需要统一管理和部署提示词资产的企业环境。", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 3, "stars_in_period": 519 }, "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 2, "stars_in_period": 2293 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 3, "stars_in_period": 8360 } }, "composite_score": 0.492056, "delta_since_last_run": 97, "previous_rank": 37 }, { "rank": 36, "repo_path": "promptfoo/promptfoo", "owner": "promptfoo", "name": "promptfoo", "repo_url": "https://github.com/promptfoo/promptfoo", "owner_url": "https://github.com/promptfoo", "description": "Test your prompts, agents, and RAGs. Red teaming/pentesting/vulnerability scanning for AI. Compare performance of GPT, Claude, Gemini, Llama, and more. Simple declarative configs with command line and CI/CD integration. Used by OpenAI and Anthropic.", "language": "TypeScript", "stars_total": 19914, "forks_total": 1706, "stars_in_period": 8307, "period_label": "最近30天", "description_zh": "测试您的提示、代理和 RAG。 AI 红队/渗透测试/漏洞扫描。比较 GPT、Claude、Gemini、Llama 等的性能。具有命令行和 CI/CD 集成的简单声明性配置。由 OpenAI 和 Anthropic 使用。", "readme_summary_zh": "promptfoo 是一个面向大语言模型应用的开源评测与红队测试工具，既可作为命令行工具使用，也可集成到代码库中作为库来运行，帮助开发者摆脱反复试错的低效流程，更系统地验证 AI 应用的质量、安全性与稳定性。它的核心能力在于对提示词、模型输出和应用行为进行自动化评估，并通过红队测试发现越狱、提示注入、敏感信息泄露等潜在风险，从而支持团队更有把握地交付可靠的 AI 产品。这个项目尤其适合正在开发聊天机器人、智能助手、工作流代理或其他依赖 LLM 的应用团队，在上线前后的测试、回归验证、安全审查和持续迭代场景中使用。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 7, "stars_in_period": 8307 } }, "composite_score": 0.074879, "delta_since_last_run": 31, "previous_rank": 33 }, { "rank": 37, "repo_path": "clash-verge-rev/clash-verge-rev", "owner": "clash-verge-rev", "name": "clash-verge-rev", "repo_url": "https://github.com/clash-verge-rev/clash-verge-rev", "owner_url": "https://github.com/clash-verge-rev", "description": "A modern GUI client based on Tauri, designed to run in Windows, macOS and Linux for tailored proxy experience", "language": "TypeScript", "stars_total": 109572, "forks_total": 7962, "stars_in_period": 8257, "period_label": "最近30天", "description_zh": "基于 Tauri 的现代 GUI 客户端，设计用于在 Windows、macOS 和 Linux 中运行，以提供定制的代理体验", "readme_summary_zh": "Languages: 简体中文 · English · Español · Русский · 日本語 · 한국어 · فارسی", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 12, "stars_in_period": 8257 } }, "composite_score": 0.07392, "delta_since_last_run": 94, "previous_rank": 36 }, { "rank": 38, "repo_path": "aaif-goose/goose", "owner": "aaif-goose", "name": "goose", "repo_url": "https://github.com/aaif-goose/goose", "owner_url": "https://github.com/aaif-goose", "description": "an open source, extensible AI agent that goes beyond code suggestions - install, execute, edit, and test with any LLM", "language": "Rust", "stars_total": 40974, "forks_total": 4071, "stars_in_period": 8256, "period_label": "最近30天", "description_zh": "开源、可扩展的 AI 代理，超越代码建议 - 使用任何 LLM 安装、执行、编辑和测试", "readme_summary_zh": "🦆鹅动了！该项目已从 block/goose 转移到 Linux 基金会的 Agentic AI Foundation (AAIF)。一些链接和参考仍在更新中——过渡期间请耐心等待。您的本机开源 AI 代理 - 桌面应用程序、CLI 和 API - 用于代码、工作流程以及其间的所有内容", "source_period": "monthly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 2, "stars_in_period": 6428 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 9, "stars_in_period": 8256 } }, "composite_score": 0.306732, "delta_since_last_run": 170, "previous_rank": 39 }, { "rank": 39, "repo_path": "coreyhaines31/marketingskills", "owner": "coreyhaines31", "name": "marketingskills", "repo_url": "https://github.com/coreyhaines31/marketingskills", "owner_url": "https://github.com/coreyhaines31", "description": "Marketing skills for Claude Code and AI agents. CRO, copywriting, SEO, analytics, and growth engineering.", "language": "JavaScript", "stars_total": 20197, "forks_total": 3157, "stars_in_period": 7714, "period_label": "最近30天", "description_zh": "Claude Code 和 AI 代理的营销技巧。 CRO、文案、SEO、分析和增长工程。", "readme_summary_zh": "这是一个面向营销工作的 AI Agent Skills 技能库，主要为技术型营销人员、创业者以及希望让 AI 编码代理参与增长工作的团队设计。项目通过一组 Markdown 技能文件，为代理提供在转化率优化、文案撰写、SEO、数据分析和增长工程等任务中的专业知识、工作流程与最佳实践，可运行在 Claude Code、OpenAI Codex、Cursor、Windsurf 以及任何支持 Agent Skills 规范的代理环境中。其核心机制是让代理在识别到营销相关任务时，自动调用合适的框架与方法，并结合技能之间的相互引用协同工作，其中 product-marketing-context 作为基础能力，先帮助代理理解产品、目标用户和市场定位，再开展后续分析与执行。这个项目特别适合需要把 AI 从“写代码”进一步扩展到“做营销”的场景，例如优化落地页、制定增长实验、提升搜索流量、完善信息传达和建立更系统的营销执行流程。", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 13, "stars_in_period": 216 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 11, "stars_in_period": 7714 } }, "composite_score": 0.276114, "delta_since_last_run": 97, "previous_rank": 38 }, { "rank": 40, "repo_path": "unslothai/unsloth", "owner": "unslothai", "name": "unsloth", "repo_url": "https://github.com/unslothai/unsloth", "owner_url": "https://github.com/unslothai", "description": "Unsloth Studio is a web UI for training and running open models like Gemma 4, Qwen3.5, DeepSeek, gpt-oss locally.", "language": "Python", "stars_total": 60947, "forks_total": 5252, "stars_in_period": 7472, "period_label": "最近30天", "description_zh": "Unsloth Studio 是一个 Web UI，用于在本地训练和运行 Gemma 4、Qwen3.5、DeepSeek、gpt-oss 等开放模型。", "readme_summary_zh": "Unsloth 是一个面向生成式 AI 模型训练与推理的一体化平台，可通过图形化的 Unsloth Studio（Beta）或代码式的 Unsloth Core 使用，支持在 Windows、Linux、WSL 和 macOS 上运行文本、音频、视觉与嵌入模型。它的核心能力包括搜索下载并运行 GGUF、LoRA、safetensors 等模型，导出为 GGUF 或 16 位 safetensors 等格式，支持工具调用、联网搜索、代码执行，以及对推理参数和聊天模板进行自动调优；同时还能上传图片、音频、PDF、代码、DOCX 等多种文件进行多模态交互。训练方面，它支持 500 多个模型的微调、预训练与强化学习，涵盖 4-bit、16-bit、FP8 等模式，在尽量不损失精度的情况下提升速度并显著降低显存占用，还提供数据集自动构建、可视化数据流程、训练监控和多 GPU 支持。这个项目尤其适合希望以较低硬件成本快速开展本地模型实验、企业内部知识问答、多模态助手开发和高效模型定制的团队与开发者。", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 12, "stars_in_period": 292 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 14, "stars_in_period": 7472 } }, "composite_score": 0.27835, "delta_since_last_run": 91, "previous_rank": 42 }, { "rank": 41, "repo_path": "Wei-Shaw/sub2api", "owner": "Wei-Shaw", "name": "sub2api", "repo_url": "https://github.com/Wei-Shaw/sub2api", "owner_url": "https://github.com/Wei-Shaw", "description": "Sub2API-CRS2 一站式开源中转服务，让 Claude、Openai 、Gemini、Antigravity订阅统一接入，支持拼车共享，更高效分摊成本，原生工具无缝使用。", "language": "Go", "stars_total": 11605, "forks_total": 1909, "stars_in_period": 7325, "period_label": "最近30天", "description_zh": "Sub2API-CRS2 一站式开源中转服务，让 Claude、Openai 、Gemini、Antigravity订阅统一接入，支持拼车共享，更高效分摊成本，原生工具无缝使用。", "readme_summary_zh": "Sub2API 是一个面向 AI 产品订阅额度分发与管理的 API 网关平台，适合把上游多个 AI 账号的调用能力统一封装后提供给团队、客户或内部系统使用。它的核心作用是由平台生成 API Key 供下游访问，而平台自身负责完成上游账号认证、请求转发、负载均衡、计费与配额控制。项目支持多账号管理，可接入 OAuth 或 API Key 等不同类型的上游账户，并通过智能调度与粘性会话在多账号场景下稳定分配请求；同时提供用户级和账户级并发限制、请求与 Token 速率限制，以及按 Token 精确统计用量和成本，便于精细化运营。配套的管理后台可用于监控和日常维护，还能通过 iframe 嵌入支付、工单等外部系统扩展管理能力，因此特别适合需要统一管理订阅资源、对外提供受控 AI 接口、或在多账户环境下实现稳定转发和计费的服务商与团队。", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 3, "stars_in_period": 267 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 1, "stars_in_period": 7325 } }, "composite_score": 0.28493, "delta_since_last_run": 77, "previous_rank": 40 }, { "rank": 42, "repo_path": "luongnv89/claude-howto", "owner": "luongnv89", "name": "claude-howto", "repo_url": "https://github.com/luongnv89/claude-howto", "owner_url": "https://github.com/luongnv89", "description": "A visual, example-driven guide to Claude Code — from basic concepts to advanced agents, with copy-paste templates that bring immediate value.", "language": "Python", "stars_total": 24779, "forks_total": 2964, "stars_in_period": 7215, "period_label": "最近7天", "description_zh": "克劳德代码的可视化、示例驱动指南——从基本概念到高级代理，以及可带来即时价值的复制粘贴模板。", "readme_summary_zh": "这是一个面向 Claude Code 用户的系统化学习与实战指南，目标是帮助已经会基本提问的人，进一步掌握如何把斜杠命令、hooks、memory、skills、subagents 与 MCP 服务器组合成真正高效的自动化开发工作流。项目并非单纯罗列功能，而是通过 10 个可视化教程模块、可直接复制的配置模板、Mermaid 原理图以及循序渐进的学习路径，解释各项能力如何协同工作，并提供适合落地到项目中的命令、脚本、配置与插件打包示例。它尤其适合想从入门快速进阶为高级用户的开发者、团队负责人或需要搭建代码审查、安全扫描、任务分派等生产级流程的人，也适合对 Claude Code 功能很多却不知道先学什么、怎样组合最有价值的用户。", "source_period": "weekly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 12, "stars_in_period": 7215 } }, "composite_score": 0.233407, "delta_since_last_run": 249, "previous_rank": 41 }, { "rank": 43, "repo_path": "anthropics/claude-plugins-official", "owner": "anthropics", "name": "claude-plugins-official", "repo_url": "https://github.com/anthropics/claude-plugins-official", "owner_url": "https://github.com/anthropics", "description": "Official, Anthropic-managed directory of high quality Claude Code Plugins.", "language": "Python", "stars_total": 16607, "forks_total": 1913, "stars_in_period": 6995, "period_label": "最近30天", "description_zh": "由 Anthropic 管理的官方高品质 Claude 代码插件目录。", "readme_summary_zh": "这是一个面向 Claude Code 的官方精选插件目录，旨在帮助用户更方便地发现和使用高质量插件。项目按来源分为两类：一类是由 Anthropic 自行开发和维护的内部插件，通常更适合希望获得与 Claude Code 深度配合能力的用户；另一类是来自合作伙伴和社区的第三方插件，用于扩展不同工具、服务或工作流的接入范围。它的核心价值在于集中整理可选插件，降低查找成本，让开发者、技术团队和高级用户能根据自身需求快速挑选合适扩展。不过 README 也明确强调，安装、更新或使用任何插件前都应先确认其可信度，因为插件中包含的 MCP 服务器、文件或其他软件并不由 Anthropic 直接控制，也无法保证其行为、稳定性或后续不变更，因此它最适合用于希望为 Claude Code 增强能力、并具备一定安全审查意识的使用场景。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 17, "stars_in_period": 6995 } }, "composite_score": 0.071727, "delta_since_last_run": 38, "previous_rank": 44 }, { "rank": 44, "repo_path": "vectorize-io/hindsight", "owner": "vectorize-io", "name": "hindsight", "repo_url": "https://github.com/vectorize-io/hindsight", "owner_url": "https://github.com/vectorize-io", "description": "Hindsight: Agent Memory That Learns", "language": "Python", "stars_total": 8862, "forks_total": 521, "stars_in_period": 6408, "period_label": "最近30天", "description_zh": "事后诸葛亮：智能体的学习记忆", "readme_summary_zh": "Hindsight 是一个面向智能体的长期记忆系统，目标不是简单保存对话历史，而是让智能体在持续交互中真正“学会”积累经验并提升表现。它强调对长期记忆任务的高精度处理，试图弥补传统 RAG 和知识图谱方案在记忆可靠性、召回效果与复杂场景适应性上的不足，并在 LongMemEval 等记忆基准测试中取得了领先成绩。项目既提供适合快速接入的 LLM Wrapper，可较低成本为现有 agent 自动加入记忆的存储与检索能力，也提供更灵活的 API、SDK 和 HTTP 集成方式，方便开发者精细控制记忆写入与调用流程。它尤其适合需要跨会话持续理解用户、长期跟踪上下文、不断优化响应质量的企业级 AI 助手、客服系统、编程助手和各类生产环境中的智能体应用。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 13, "stars_in_period": 6408 } }, "composite_score": 0.071854, "delta_since_last_run": 48, "previous_rank": 46 }, { "rank": 45, "repo_path": "louis-e/arnis", "owner": "louis-e", "name": "arnis", "repo_url": "https://github.com/louis-e/arnis", "owner_url": "https://github.com/louis-e", "description": "Generate any location from the real world in Minecraft with a high level of detail.", "language": "Rust", "stars_total": 14711, "forks_total": 1173, "stars_in_period": 6380, "period_label": "最近30天", "description_zh": "在 Minecraft 中生成现实世界中的任何位置，具有高水平的细节。", "readme_summary_zh": "Arnis 是一个免费开源的 Minecraft 世界生成项目，面向 Java 版 1.17 及以上以及基岩版，目标是把现实世界的地理环境高精度还原到游戏中。它能够处理大规模真实地理数据，结合 OpenStreetMap 的道路、建筑等空间信息与高程数据，自动生成贴近现实的地形起伏、城市结构和建筑布局，从而构建复杂且准确的 Minecraft 地图。这个项目特别适合想在游戏中重现家乡、知名城市或自然景观的玩家，也适用于地图创作者、教育展示和地理可视化等场景；如果希望在移动设备上使用或生成更大的地图，还可以通过其浏览器方案完成，无需本地安装。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 7, "stars_in_period": 6380 } }, "composite_score": 0.072914, "delta_since_last_run": 39, "previous_rank": 45 }, { "rank": 46, "repo_path": "google-research/timesfm", "owner": "google-research", "name": "timesfm", "repo_url": "https://github.com/google-research/timesfm", "owner_url": "https://github.com/google-research", "description": "TimesFM (Time Series Foundation Model) is a pretrained time-series foundation model developed by Google Research for time-series forecasting.", "language": "Python", "stars_total": 16288, "forks_total": 1511, "stars_in_period": 6288, "period_label": "最近30天", "description_zh": "TimesFM（时间序列基础模型）是 Google Research 开发的用于时间序列预测的预训练时间序列基础模型。", "readme_summary_zh": "TimesFM 是 Google Research 推出的预训练时间序列基础模型，主要用于时间序列预测，目标是以统一模型处理多种预测任务。该项目当前重点提供 TimesFM 2.5 版本，相比早期版本在模型设计与推理接口上都有升级：参数量从 5 亿降到 2 亿，在降低规模的同时支持最长 16k 的上下文输入，并可通过可选的分位数预测头实现最长 1k 预测步长的连续分位数预测，还恢复了通过 XReg 使用协变量的能力，去除了频率指示器并增加了一些新的预测选项。整体来看，它适合需要中长序列建模、批量预测、多步未来趋势估计以及不确定性评估的场景，尤其适用于研究人员、数据科学团队和企业在零售、运营、容量规划、监控指标分析等任务中快速应用通用时序基础模型。", "source_period": "monthly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 11, "stars_in_period": 2481 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 15, "stars_in_period": 6288 } }, "composite_score": 0.266423, "delta_since_last_run": 78, "previous_rank": 49 }, { "rank": 47, "repo_path": "QuantumNous/new-api", "owner": "QuantumNous", "name": "new-api", "repo_url": "https://github.com/QuantumNous/new-api", "owner_url": "https://github.com/QuantumNous", "description": "A unified AI model hub for aggregation & distribution. It supports cross-converting various LLMs into OpenAI-compatible, Claude-compatible, or Gemini-compatible formats. A centralized gateway for personal and enterprise model management. 🍥", "language": "Go", "stars_total": 26054, "forks_total": 5263, "stars_in_period": 6215, "period_label": "最近30天", "description_zh": "用于聚合和分发的统一人工智能模型中心。它支持将各种 LLM 交叉转换为 OpenAI 兼容、Claude 兼容或 Gemini 兼容格式。用于个人和企业模型管理的集中式网关。 🍥", "readme_summary_zh": "new-api 是一个面向大语言模型接入与统一管理的下一代网关及 AI 资产管理系统，主要用于帮助开发者在一个平台内集中处理模型调用、渠道接入与相关资源管理。它的核心价值在于把分散的 AI 服务入口整合为统一接口，便于个人或团队在学习、测试和内部使用场景中更高效地管理多种模型能力与资产。该项目更适合希望搭建自有 LLM 中间层、做接口聚合、进行实验性开发或内部系统集成的用户，但 README 也明确说明它主要用于个人学习，不承诺稳定性或技术支持，使用者需遵守 OpenAI 使用条款及相关法律法规，不得用于任何非法用途，并应注意在中国境内不得面向公众提供未备案的生成式人工智能服务。", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 17, "stars_in_period": 223 }, "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 11, "stars_in_period": 1365 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 2, "stars_in_period": 6215 } }, "composite_score": 0.460031, "delta_since_last_run": 55, "previous_rank": 47 }, { "rank": 48, "repo_path": "ggml-org/llama.cpp", "owner": "ggml-org", "name": "llama.cpp", "repo_url": "https://github.com/ggml-org/llama.cpp", "owner_url": "https://github.com/ggml-org", "description": "LLM inference in C/C++", "language": "C++", "stars_total": 103009, "forks_total": 16664, "stars_in_period": 6112, "period_label": "最近30天", "description_zh": "C/C++ 中的 LLM 推理", "readme_summary_zh": "llama.cpp 是一个以 C/C++ 实现的大语言模型推理项目，围绕 ggml 构建，目标是在尽量少依赖、较高效率和广泛硬件兼容性的前提下，本地运行与部署各类 LLM。它的核心能力包括对 GGUF 等模型格式的支持、面向 libllama 的推理接口、可通过 llama-server 提供 REST API 服务，以及持续完善的模型管理与生态集成，例如将通过 Hugging Face 下载的模型统一放入标准缓存目录，便于与其他 HF 工具共享。项目还支持包括 gpt-oss 在内的模型，并加入了原生 MXFP4 格式支持，同时提供 WebUI 使用方式，以及面向 VS Code、Vim/Neovim 的 FIM 补全扩展生态。它尤其适合需要在本地设备、边缘环境或自建服务中进行高性能模型推理、快速原型开发、代码补全集成和轻量化 API 部署的场景。", "source_period": "monthly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 2, "stars_in_period": 2013 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 2, "stars_in_period": 6112 } }, "composite_score": 0.269551, "delta_since_last_run": 69, "previous_rank": 48 }, { "rank": 49, "repo_path": "ChromeDevTools/chrome-devtools-mcp", "owner": "ChromeDevTools", "name": "chrome-devtools-mcp", "repo_url": "https://github.com/ChromeDevTools/chrome-devtools-mcp", "owner_url": "https://github.com/ChromeDevTools", "description": "Chrome DevTools for coding agents", "language": "TypeScript", "stars_total": 34002, "forks_total": 2036, "stars_in_period": 5834, "period_label": "最近30天", "description_zh": "用于编码代理的 Chrome DevTools", "readme_summary_zh": "chrome-devtools-mcp 是一个面向 AI 编码助手的 MCP 服务器，用来让 Gemini、Claude、Cursor、Copilot 等代理直接连接并操控正在运行的 Chrome 浏览器。它把 Chrome DevTools 的能力开放给这些智能代理，使其不仅能执行更可靠的浏览器自动化操作，还能实时检查页面状态、定位前端问题、分析网络与性能表现，并辅助完成更深入的调试工作。这个项目尤其适合需要让 AI 参与网页开发、交互测试、故障排查和性能优化的场景，能够帮助开发者在真实浏览器环境中获得更准确的上下文和更高效的问题解决能力。", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 12, "stars_in_period": 146 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 8, "stars_in_period": 5834 } }, "composite_score": 0.273059, "delta_since_last_run": 44, "previous_rank": 50 }, { "rank": 50, "repo_path": "vxcontrol/pentagi", "owner": "vxcontrol", "name": "pentagi", "repo_url": "https://github.com/vxcontrol/pentagi", "owner_url": "https://github.com/vxcontrol", "description": "Fully autonomous AI Agents system capable of performing complex penetration testing tasks", "language": "Go", "stars_total": 14736, "forks_total": 1930, "stars_in_period": 5485, "period_label": "最近30天", "description_zh": "完全自主的人工智能代理系统能够执行复杂的渗透测试任务", "readme_summary_zh": "PentAGI 是一个面向自动化安全测试与渗透测试的 AI 驱动平台，主要为信息安全从业者、研究人员和具备合规授权的测试团队提供高灵活度的实战环境。它在隔离的 Docker 沙箱中运行，可由智能代理自主规划并执行测试步骤，结合 20 多种专业安全工具完成资产探测、漏洞验证与利用分析，同时通过长期记忆、知识图谱和网页情报采集持续积累上下文与经验。项目还支持多种外部搜索系统、面向研究与开发等任务的多代理协作、详细日志与 Grafana/Prometheus 监控、漏洞报告生成，以及用于集成自动化流程的 REST 和 GraphQL API。它最适合用于企业内网评估、研究型攻防实验、自动化漏洞验证和需要可观察、可扩展、可复用测试流程的安全场景。", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 5, "stars_in_period": 75 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 3, "stars_in_period": 5485 } }, "composite_score": 0.275347, "delta_since_last_run": 15, "previous_rank": 51 }, { "rank": 51, "repo_path": "microsoft/markitdown", "owner": "microsoft", "name": "markitdown", "repo_url": "https://github.com/microsoft/markitdown", "owner_url": "https://github.com/microsoft", "description": "Python tool for converting files and office documents to Markdown.", "language": "Python", "stars_total": 100025, "forks_total": 6127, "stars_in_period": 5214, "period_label": "最近7天", "description_zh": "用于将文件和 Office 文档转换为 Markdown 的 Python 工具。", "readme_summary_zh": "MarkItDown 现在提供 MCP（模型上下文协议）服务器，用于与 Claude Desktop 等 LLM 应用程序集成。有关更多信息，请参阅 markitdown-mcp。 0.0.1 到 0.1.0 之间的重大变化：", "source_period": "weekly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 1, "stars_in_period": 2352 }, "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 9, "stars_in_period": 5214 } }, "composite_score": 0.496954, "delta_since_last_run": 1374, "previous_rank": 53 }, { "rank": 52, "repo_path": "supermemoryai/supermemory", "owner": "supermemoryai", "name": "supermemory", "repo_url": "https://github.com/supermemoryai/supermemory", "owner_url": "https://github.com/supermemoryai", "description": "Memory engine and app that is extremely fast, scalable. The Memory API for the AI era.", "language": "TypeScript", "stars_total": 21637, "forks_total": 1983, "stars_in_period": 4868, "period_label": "最近30天", "description_zh": "内存引擎和应用程序速度极快，可扩展。 AI时代的内存API。", "readme_summary_zh": "Supermemory 是一个面向 AI 助手的先进记忆与上下文引擎，目标是解决模型在多轮、跨会话场景中“聊完就忘”的问题。它能够从对话中自动学习并提取事实，构建用户画像与持久化记忆图谱，处理知识更新与相互矛盾的信息，清理过期内容，并在合适的时机为模型提供最相关的上下文。项目同时整合了完整的上下文技术栈，包括 RAG、外部数据连接器和文件处理能力，并将这些能力统一在单一的记忆结构与本体体系中。它尤其适合需要长期记住用户偏好、项目背景和历史讨论的 AI 产品，例如个人 AI 助手、企业知识助手、客服与协作型智能体，能够让系统随着交互不断积累经验、提升理解与响应质量。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 6, "stars_in_period": 4868 } }, "composite_score": 0.071555, "delta_since_last_run": 12, "previous_rank": 52 }, { "rank": 53, "repo_path": "google-ai-edge/gallery", "owner": "google-ai-edge", "name": "gallery", "repo_url": "https://github.com/google-ai-edge/gallery", "owner_url": "https://github.com/google-ai-edge", "description": "A gallery that showcases on-device ML/GenAI use cases and allows people to try and use models locally.", "language": "Kotlin", "stars_total": 20249, "forks_total": 1889, "stars_in_period": 4846, "period_label": "最近30天", "description_zh": "展示设备上 ML/GenAI 用例并允许人们在本地尝试和使用模型的图库。", "readme_summary_zh": "使用 Google AI Edge 探索、体验和评估设备上生成式 AI 的未来。 AI Edge Gallery 是在移动设备上运行世界上最强大的开源大型语言模型 (LLM) 的首选目的地。直接在您的硬件上体验高性能生成式人工智能——完全离线、私密且快如闪电。", "source_period": "monthly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 1, "stars_in_period": 4409 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 16, "stars_in_period": 4846 } }, "composite_score": 0.285934, "delta_since_last_run": 138, "previous_rank": 54 }, { "rank": 54, "repo_path": "HKUDS/DeepTutor", "owner": "HKUDS", "name": "DeepTutor", "repo_url": "https://github.com/HKUDS/DeepTutor", "owner_url": "https://github.com/HKUDS", "description": "\"DeepTutor: Agent-Native Personalized Learning Assistant\"", "language": "Python", "stars_total": 16066, "forks_total": 2115, "stars_in_period": 4698, "period_label": "最近7天", "description_zh": "《DeepTutor：Agent-原生个性化学习助手》", "readme_summary_zh": "Features · Get Started · Explore · TutorBot · CLI · Community 🇨🇳 中文 · 🇯🇵 日本語 · 🇪🇸 Español · 🇫🇷 Français · 🇸🇦 العربية · 🇷🇺 Русский · 🇮🇳 हिन्दी · 🇵🇹 Português", "source_period": "weekly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 8, "stars_in_period": 1424 }, "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 2, "stars_in_period": 4698 } }, "composite_score": 0.462475, "delta_since_last_run": 376, "previous_rank": 62 }, { "rank": 55, "repo_path": "pathwaycom/llm-app", "owner": "pathwaycom", "name": "llm-app", "repo_url": "https://github.com/pathwaycom/llm-app", "owner_url": "https://github.com/pathwaycom", "description": "Ready-to-run cloud templates for RAG, AI pipelines, and enterprise search with live data. 🐳Docker-friendly.⚡Always in sync with Sharepoint, Google Drive, S3, Kafka, PostgreSQL, real-time data APIs, and more.", "language": "Jupyter Notebook", "stars_total": 59990, "forks_total": 1427, "stars_in_period": 4440, "period_label": "最近30天", "description_zh": "适用于 RAG、AI 管道和带有实时数据的企业搜索的可立即运行的云模板。 🐳Docker 友好。⚡始终与 Sharepoint、Google Drive、S3、Kafka、PostgreSQL、实时数据 API 等同步。", "readme_summary_zh": "这个项目提供了一套基于 Pathway 的现成 AI 应用模板，目标是帮助开发者快速构建并上线高精度的 RAG 与企业级 AI 搜索系统。它能够持续同步多种数据源中的新增、删除和更新内容，包括本地文件、Google Drive、SharePoint、S3、Kafka、PostgreSQL 以及实时数据 API，并在内存中完成带缓存的数据索引，支持向量检索、混合检索和全文检索，可扩展到数百万页文档规模。项目强调开箱即用，无需额外搭建向量数据库、缓存或独立 API 框架，同时又保留较强可定制性，开发者可以很方便地替换数据源或调整检索策略。应用可通过 Docker 部署到云端或本地环境，对外提供 HTTP API，部分模板还附带便于演示的 Streamlit 界面，因此特别适合需要快速验证、低运维成本、并且要求知识始终保持最新的企业知识库问答、内部搜索和生产级生成式 AI 场景。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 14, "stars_in_period": 4440 } }, "composite_score": 0.069667, "delta_since_last_run": -1, "previous_rank": 55 }, { "rank": 56, "repo_path": "langchain-ai/open-swe", "owner": "langchain-ai", "name": "open-swe", "repo_url": "https://github.com/langchain-ai/open-swe", "owner_url": "https://github.com/langchain-ai", "description": "An Open-Source Asynchronous Coding Agent", "language": "Python", "stars_total": 9428, "forks_total": 1061, "stars_in_period": 4187, "period_label": "最近30天", "description_zh": "开源异步编码代理", "readme_summary_zh": "open-swe 是一个面向企业工程团队的开源编码代理方案，目标是帮助组织构建类似 Stripe、Ramp、Coinbase 等公司内部使用的智能开发助手，并将其部署为 Slack 机器人、命令行工具或 Web 应用，直接融入工程师已经熟悉的工作流。它的核心价值在于把编码代理与企业内部系统连接起来，使其能够在具备正确上下文、权限控制和安全边界的前提下执行开发相关任务，从而减少人工介入并提升工程效率。这个项目尤其适合希望打造内部 AI 工程助手的团队，用于支持日常研发协作、代码处理与自动化工程流程等场景。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 17, "stars_in_period": 4187 } }, "composite_score": 0.068865, "delta_since_last_run": 29, "previous_rank": 56 }, { "rank": 57, "repo_path": "VoltAgent/awesome-claude-code-subagents", "owner": "VoltAgent", "name": "awesome-claude-code-subagents", "repo_url": "https://github.com/VoltAgent/awesome-claude-code-subagents", "owner_url": "https://github.com/VoltAgent", "description": "A collection of 100+ specialized Claude Code subagents covering a wide range of development use cases", "language": "Shell", "stars_total": 16920, "forks_total": 1928, "stars_in_period": 3841, "period_label": "最近30天", "description_zh": "100 多个专业 Claude Code 子代理的集合，涵盖广泛的开发用例", "readme_summary_zh": "这是一个专门收集 Claude Code 子代理的精选项目，目标是为开发者提供一套面向具体开发任务的专业化 AI 助手目录。项目中的子代理并非通用聊天机器人，而是围绕特定职责设计的协作单元，可用于承担不同软件开发环节中的细分工作，帮助团队更高效地组织、分派与执行任务。它的核心价值在于汇总和整理这些针对性强的智能助手，让用户能够更方便地发现适合自己流程的子代理方案，并在编码、协作和工程实践中引入更明确的角色分工。这个项目最适合希望借助 Claude Code 扩展开发工作流、为不同任务配置专门 AI 助手，以及寻找高质量子代理参考集合的开发者与技术团队使用。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 10, "stars_in_period": 3841 } }, "composite_score": 0.069782, "delta_since_last_run": 34, "previous_rank": 58 }, { "rank": 58, "repo_path": "forrestchang/andrej-karpathy-skills", "owner": "forrestchang", "name": "andrej-karpathy-skills", "repo_url": "https://github.com/forrestchang/andrej-karpathy-skills", "owner_url": "https://github.com/forrestchang", "description": "A single CLAUDE.md file to improve Claude Code behavior, derived from Andrej Karpathy's observations on LLM coding pitfalls.", "language": null, "stars_total": 11871, "forks_total": 785, "stars_in_period": 3741, "period_label": "最近7天", "description_zh": "用于改进 Claude 代码行为的单个 CLAUDE.md 文件，源自 Andrej Karpathy 对 LLM 编码陷阱的观察。", "readme_summary_zh": "用于改进 Claude 代码行为的单个 CLAUDE.md 文件，源自 Andrej Karpathy 对 LLM 编码陷阱的观察。来自安德烈的帖子：", "source_period": "weekly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 6, "stars_in_period": 1450 }, "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 7, "stars_in_period": 3741 } }, "composite_score": 0.453941, "delta_since_last_run": 473, "previous_rank": 79 }, { "rank": 59, "repo_path": "datalab-to/chandra", "owner": "datalab-to", "name": "chandra", "repo_url": "https://github.com/datalab-to/chandra", "owner_url": "https://github.com/datalab-to", "description": "OCR model that handles complex tables, forms, handwriting with full layout.", "language": "Python", "stars_total": 8507, "forks_total": 871, "stars_in_period": 3647, "period_label": "最近30天", "description_zh": "OCR 模型可处理复杂的表格、表格、具有完整布局的手写内容。", "readme_summary_zh": "Chandra 是一个面向文档智能的先进 OCR 项目，主要用于把图片和 PDF 转换为带有版面结构信息的 Markdown、HTML 或 JSON，方便后续检索、解析、归档与自动化处理。项目强调高精度的复杂文档理解能力，在表格、数学公式、多栏排版、表单重建以及复选框识别等场景中表现突出，同时支持 90 多种语言，并对手写内容也有很强的识别效果。除文本提取外，它还能抽取图片和图示，并生成说明与结构化数据。Chandra 提供本地推理和远程服务两种使用方式，适合需要将扫描件、学术论文、合同表单、财务报表、教育资料或多语言档案批量转成可机器处理格式的团队与应用。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 18, "stars_in_period": 3647 } }, "composite_score": 0.068133, "delta_since_last_run": 8, "previous_rank": 59 }, { "rank": 60, "repo_path": "multica-ai/multica", "owner": "multica-ai", "name": "multica", "repo_url": "https://github.com/multica-ai/multica", "owner_url": "https://github.com/multica-ai", "description": "The open-source managed agents platform. Turn coding agents into real teammates — assign tasks, track progress, compound skills.", "language": "TypeScript", "stars_total": 6323, "forks_total": 784, "stars_in_period": 3512, "period_label": "最近7天", "description_zh": "开源托管代理平台。将编码代理变成真正的队友 - 分配任务、跟踪进度、复合技能。", "readme_summary_zh": "你接下来的 10 名员工将不是人类。开源托管代理平台。将编码代理变成真正的队友 - 分配任务、跟踪进度、复合技能。", "source_period": "weekly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 5, "stars_in_period": 1506 }, "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 5, "stars_in_period": 3512 } }, "composite_score": 0.455507, "delta_since_last_run": 853, "previous_rank": 63 }, { "rank": 61, "repo_path": "2dust/v2rayN", "owner": "2dust", "name": "v2rayN", "repo_url": "https://github.com/2dust/v2rayN", "owner_url": "https://github.com/2dust", "description": "A GUI client for Windows, Linux and macOS, support Xray and sing-box and others", "language": "C#", "stars_total": 101556, "forks_total": 14549, "stars_in_period": 3500, "period_label": "最近30天", "description_zh": "适用于 Windows、Linux 和 macOS 的 GUI 客户端，支持 Xray 和 sing-box 等", "readme_summary_zh": "v2rayN 是一款面向 Windows、Linux 和 macOS 的跨平台图形界面客户端，主要用于管理和使用基于 Xray、sing-box 等内核的代理与网络转发能力。它的核心价值在于把原本较为复杂的代理配置、节点管理和连接切换过程做成更直观的桌面操作方式，方便用户在不同协议或后端之间进行统一使用与维护。对于需要在日常办公、开发测试、跨平台设备间保持一致网络代理体验的用户来说，这个项目尤其合适，既能降低命令行配置门槛，也便于在图形界面中完成常见的连接管理与使用。", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 5, "stars_in_period": 165 }, "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 3, "stars_in_period": 1011 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 3, "stars_in_period": 3500 } }, "composite_score": 0.463467, "delta_since_last_run": 40, "previous_rank": 60 }, { "rank": 62, "repo_path": "dmtrKovalenko/fff.nvim", "owner": "dmtrKovalenko", "name": "fff.nvim", "repo_url": "https://github.com/dmtrKovalenko/fff.nvim", "owner_url": "https://github.com/dmtrKovalenko", "description": "The fastest and the most accurate file search toolkit for AI agents, Neovim, Rust, C, and NodeJS", "language": "Rust", "stars_total": 4347, "forks_total": 178, "stars_in_period": 3449, "period_label": "最近30天", "description_zh": "适用于 AI 代理、Neovim、Rust、C 和 NodeJS 的最快、最准确的文件搜索工具包", "readme_summary_zh": "fff.nvim 是一个面向 Neovim 用户和 AI 代理的高性能文件搜索工具，专注把“找文件”这件事做到又快又准。它提供模糊文件匹配、内容 grep、glob 搜索和多关键字检索等能力，并特别强调性能与结果质量。对人类用户来说，它具备很强的拼写容错能力，输入不精确也能较容易找到目标；对 AI 代理来说，它在搜索过程中内置“记忆”机制，会综合使用频率与新近度、Git 状态、文件大小、定义匹配等因素对结果排序，帮助模型更快定位最相关的代码文件，减少无效读取、往返查询和 token 消耗。这个项目尤其适合大型代码库、文件数量庞杂的工程，以及希望在 Neovim 中提升导航效率、或为 Claude Code、Codex、OpenCode 等 AI 编程助手增强检索能力的场景。", "source_period": "monthly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 1, "stars_in_period": 1362 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 1, "stars_in_period": 3449 } }, "composite_score": 0.262203, "delta_since_last_run": 141, "previous_rank": 65 }, { "rank": 63, "repo_path": "anthropics/claude-cookbooks", "owner": "anthropics", "name": "claude-cookbooks", "repo_url": "https://github.com/anthropics/claude-cookbooks", "owner_url": "https://github.com/anthropics", "description": "A collection of notebooks/recipes showcasing some fun and effective ways of using Claude.", "language": "Jupyter Notebook", "stars_total": 37884, "forks_total": 4210, "stars_in_period": 3359, "period_label": "最近30天", "description_zh": "一系列笔记本/食谱展示了一些有趣且有效的使用克劳德的方法。", "readme_summary_zh": "Claude Cookbooks 是一套面向开发者的示例代码与实践指南，目标是帮助用户更快地基于 Claude 构建应用并把能力接入自己的项目。它提供了可直接复制和改造的代码片段，主要以 Python 编写，但核心思路同样适用于任何能够调用 Claude API 的编程语言。这个项目尤其适合想快速上手 Claude API、验证产品原型、学习常见集成方式，或在现有系统中加入 AI 助手能力的开发者使用。对于刚接触 Claude API 的用户，它也能作为入门和实践参考，帮助建立从基础调用到实际落地的开发路径。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 1, "stars_in_period": 3359 } }, "composite_score": 0.070924, "delta_since_last_run": 36, "previous_rank": 61 }, { "rank": 64, "repo_path": "sgoudelis/ground-station", "owner": "sgoudelis", "name": "ground-station", "repo_url": "https://github.com/sgoudelis/ground-station", "owner_url": "https://github.com/sgoudelis", "description": "Ground Station is all-in-one satellite monitoring suite", "language": "JavaScript", "stars_total": 3714, "forks_total": 645, "stars_in_period": 3142, "period_label": "最近30天", "description_zh": "地面站是一体化卫星监控套件", "readme_summary_zh": "Ground Station 是一款面向卫星跟踪与无线电通信的开源地面站软件，适合业余无线电爱好者、卫星观测者和科研用户使用。它将卫星轨道跟踪、过境预测、实时位置显示、天线旋转器自动控制、兼容 Hamlib 的电台控制与多种 SDR 设备接入整合到同一平台中，可对卫星信号进行实时接收、录制与回放，并支持 SSTV、FSK、GFSK、GMSK、BPSK 等多种调制与数据解码。项目还提供自动更新 TLE 数据、计划任务式观测、气象卫星图像解码、遥测与数据包查看，以及基于 Gemini Live 或 Deepgram 的实时语音转写能力，便于在卫星过境时自动完成监听、解码、转写和存档。它尤其适合需要长期值守、自动化观测、分析卫星下行信号或构建个人地面站系统的场景。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 9, "stars_in_period": 3142 } }, "composite_score": 0.069251, "delta_since_last_run": 5, "previous_rank": 64 }, { "rank": 65, "repo_path": "capcom6/android-sms-gateway", "owner": "capcom6", "name": "android-sms-gateway", "repo_url": "https://github.com/capcom6/android-sms-gateway", "owner_url": "https://github.com/capcom6", "description": "The SMS Gateway for Android™ app enables sending and receiving SMS messages through an API that can be accessed directly on the device or via a cloud server when direct device access is not possible.", "language": "Kotlin", "stars_total": 3963, "forks_total": 656, "stars_in_period": 3079, "period_label": "最近30天", "description_zh": "SMS Gateway for Android™ 应用程序可以通过 API 发送和接收 SMS 消息，该 API 可以直接在设备上访问，或者在无法直接访问设备时通过云服务器访问。", "readme_summary_zh": "将您的智能手机变成 SMS 网关，用于通过 API 发送和接收消息。浏览文档 » 报告错误·请求功能 SMS 网关可将您的 Android 智能手机变成 SMS 网关。它是一个轻量级应用程序，允许您通过 API 或 CLI 工具以编程方式发送 SMS 消息，并接收传入 SMS 上的 Webhooks。这使得它非常适合将 SMS 功能集成到您自己的应用程序中。", "source_period": "monthly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 2, "stars_in_period": 2466 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 10, "stars_in_period": 3079 } }, "composite_score": 0.268729, "delta_since_last_run": 16, "previous_rank": 66 }, { "rank": 66, "repo_path": "tobi/qmd", "owner": "tobi", "name": "qmd", "repo_url": "https://github.com/tobi/qmd", "owner_url": "https://github.com/tobi", "description": "mini cli search engine for your docs, knowledge bases, meeting notes, whatever. Tracking current sota approaches while being all local", "language": "TypeScript", "stars_total": 20715, "forks_total": 1268, "stars_in_period": 3018, "period_label": "最近7天", "description_zh": "迷你 cli 搜索引擎，用于搜索您的文档、知识库、会议记录等。跟踪当前的 sota 方法，同时都是本地的", "readme_summary_zh": "设备上的搜索引擎，可满足您需要记住的一切。索引您的 Markdown 笔记、会议记录、文档和知识库。使用关键字或自然语言进行搜索。非常适合您的代理流程。 QMD 结合了 BM25 全文搜索、向量语义搜索和 LLM 重新排名——所有这些都通过带有 GGUF 模型的 node-llama-cpp 在本地运行。", "source_period": "weekly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 6, "stars_in_period": 3018 } }, "composite_score": 0.202128, "delta_since_last_run": 186, "previous_rank": 67 }, { "rank": 67, "repo_path": "NVIDIA/personaplex", "owner": "NVIDIA", "name": "personaplex", "repo_url": "https://github.com/NVIDIA/personaplex", "owner_url": "https://github.com/NVIDIA", "description": "PersonaPlex code.", "language": "Python", "stars_total": 8932, "forks_total": 1259, "stars_in_period": 2939, "period_label": "最近7天", "description_zh": "PersonaPlex 代码。", "readme_summary_zh": "PersonaPlex 是一种实时、全双工语音到语音对话模型，可通过基于文本的角色提示和基于音频的语音调节来实现角色控制。它经过综合对话和真实对话的训练，可以产生自然、低延迟的语音交互，并具有一致的角色。 PersonaPlex 基于 Moshi 架构和权重。 PersonaPlex架构", "source_period": "weekly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 9, "stars_in_period": 2939 } }, "composite_score": 0.199881, "delta_since_last_run": 55, "previous_rank": 69 }, { "rank": 68, "repo_path": "sherlock-project/sherlock", "owner": "sherlock-project", "name": "sherlock", "repo_url": "https://github.com/sherlock-project/sherlock", "owner_url": "https://github.com/sherlock-project", "description": "Hunt down social media accounts by username across social networks", "language": "Python", "stars_total": 80776, "forks_total": 9372, "stars_in_period": 2874, "period_label": "最近7天", "description_zh": "通过社交网络上的用户名查找社交媒体帐户", "readme_summary_zh": "Sherlock 是一个用于按用户名在 400 多个社交网络和在线平台上查找账号的开源侦查工具，适合进行公开信息收集、身份关联分析和数字足迹排查。它的核心能力是一次性将同一用户名批量投递到大量网站进行检测，快速判断该用户名是否已被注册，并输出匹配结果，帮助用户发现可能属于同一人的社交媒体主页、论坛账号或其他网络身份。这个项目尤其适用于安全研究、调查取证、品牌与用户名占用检查，以及个人隐私自查等场景，能够显著减少人工逐站搜索的时间成本，但其结果通常仍需要结合人工进一步核验与分析。", "source_period": "weekly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 11, "stars_in_period": 2874 } }, "composite_score": 0.198282, "delta_since_last_run": 34, "previous_rank": 57 }, { "rank": 69, "repo_path": "chen08209/FlClash", "owner": "chen08209", "name": "FlClash", "repo_url": "https://github.com/chen08209/FlClash", "owner_url": "https://github.com/chen08209", "description": "A multi-platform proxy client based on ClashMeta,simple and easy to use, open-source and ad-free.", "language": "Dart", "stars_total": 35703, "forks_total": 2201, "stars_in_period": 2868, "period_label": "最近30天", "description_zh": "一款基于ClashMeta的多平台代理客户端，简单易用，开源无广告。", "readme_summary_zh": "FlClash 是一款基于 ClashMeta 的多平台代理客户端，主打简单易用、开源且无广告，适合希望在不同设备上统一管理代理配置的用户使用。它覆盖 Android、Windows、macOS 和 Linux，界面采用接近 Material You 与 Surfboard 风格的设计，能够自适应多种屏幕尺寸，并提供多种主题与深色模式，兼顾美观和日常使用体验。功能上支持通过订阅链接导入配置，也支持借助 WebDAV 进行数据同步，方便在多设备之间保持设置一致。因此，这个项目特别适合需要跨平台科学上网、切换代理节点、同步个人配置，且重视界面体验与轻量化使用感受的场景。", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 7, "stars_in_period": 115 }, "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 2, "stars_in_period": 715 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 2, "stars_in_period": 2868 } }, "composite_score": 0.458341, "delta_since_last_run": 40, "previous_rank": 68 }, { "rank": 70, "repo_path": "sxyazi/yazi", "owner": "sxyazi", "name": "yazi", "repo_url": "https://github.com/sxyazi/yazi", "owner_url": "https://github.com/sxyazi", "description": "💥 Blazing fast terminal file manager written in Rust, based on async I/O.", "language": "Rust", "stars_total": 36246, "forks_total": 803, "stars_in_period": 2644, "period_label": "最近30天", "description_zh": "💥 用 Rust 编写的基于异步 I/O 的超快终端文件管理器。", "readme_summary_zh": "Yazi 是一个用 Rust 编写的终端文件管理器，基于非阻塞异步 I/O，目标是在命令行环境中提供高效、顺手且可高度定制的文件管理体验。它的核心优势在于全面异步化设计，能够把 I/O 与 CPU 任务并发调度，并提供进度显示、任务取消和优先级管理，因此在浏览大量文件、加载预览和执行复杂操作时依然保持流畅。项目内置图片协议支持、代码高亮、图片解码和预加载机制，还具备多标签页、跨目录选择、批量重命名、压缩包解压、Git 集成、挂载管理、回收站和鼠标支持等功能；同时通过 Lua 插件体系、虚拟文件系统、数据分发服务以及包管理器，进一步扩展远程文件管理、自定义预览、跨实例通信和主题插件管理能力。它尤其适合长期在终端中工作的开发者、运维人员和高级用户，适用于需要快速浏览、检索、预览和批量处理本地或远程文件的日常场景。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 6, "stars_in_period": 2644 } }, "composite_score": 0.069288, "delta_since_last_run": 21, "previous_rank": 70 }, { "rank": 71, "repo_path": "jesseduffield/lazygit", "owner": "jesseduffield", "name": "lazygit", "repo_url": "https://github.com/jesseduffield/lazygit", "owner_url": "https://github.com/jesseduffield", "description": "simple terminal UI for git commands", "language": "Go", "stars_total": 76176, "forks_total": 2744, "stars_in_period": 2619, "period_label": "最近30天", "description_zh": "git 命令的简单终端 UI", "readme_summary_zh": "lazygit 是一个面向 Git 的终端图形界面工具，目标是把原本强大却繁琐、容易让人挫败的命令行操作变得更直观高效。它提供了丰富的交互式能力，例如按行暂存代码、交互式变基、拣选提交、二分排错、撤销与修订旧提交、比较两个提交、查看提交图谱、管理工作树，以及通过自定义补丁实现更灵活的 rebase 流程，还能在界面中调用自定义命令并快速清理工作区。这个项目尤其适合经常使用 Git、希望减少记忆复杂命令和手动编辑补丁文件的开发者，也很适合在终端环境中进行日常提交整理、分支切换、历史重写和问题定位等版本控制场景。", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 1, "stars_in_period": 144 }, "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 19, "stars_in_period": 655 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 14, "stars_in_period": 2619 } }, "composite_score": 0.450751, "delta_since_last_run": 40, "previous_rank": 71 }, { "rank": 72, "repo_path": "mergisi/awesome-openclaw-agents", "owner": "mergisi", "name": "awesome-openclaw-agents", "repo_url": "https://github.com/mergisi/awesome-openclaw-agents", "owner_url": "https://github.com/mergisi", "description": "162 production-ready AI agent templates for OpenClaw. SOUL.md configs across 19 categories. Submit yours!", "language": "HTML", "stars_total": 2786, "forks_total": 436, "stars_in_period": 2566, "period_label": "最近30天", "description_zh": "162 个适用于 OpenClaw 的生产就绪型 AI 代理模板。 SOUL.md 跨 19 个类别的配置。提交你的！", "readme_summary_zh": "适用于 OpenClaw 生态系统的 199 个可用于生产的 AI 代理模板的精选集合。每个模板都是一个复制粘贴就绪的 SOUL.md 文件。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 6, "stars_in_period": 2566 } }, "composite_score": 0.069208, "delta_since_last_run": 12, "previous_rank": 72 }, { "rank": 73, "repo_path": "kepano/defuddle", "owner": "kepano", "name": "defuddle", "repo_url": "https://github.com/kepano/defuddle", "owner_url": "https://github.com/kepano", "description": "Get the main content of any page as Markdown.", "language": "HTML", "stars_total": 6566, "forks_total": 259, "stars_in_period": 2536, "period_label": "最近30天", "description_zh": "以 Markdown 形式获取任何页面的主要内容。", "readme_summary_zh": "de·​fud·dle /diˈfʌdl/ 及物动词，用于从网页中删除不必要的元素，并使其易于阅读。提防！ Defuddle 是一项正在进行中的工作！", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 6, "stars_in_period": 68 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 2, "stars_in_period": 2536 } }, "composite_score": 0.271721, "delta_since_last_run": 18, "previous_rank": 73 }, { "rank": 74, "repo_path": "TheCraigHewitt/seomachine", "owner": "TheCraigHewitt", "name": "seomachine", "repo_url": "https://github.com/TheCraigHewitt/seomachine", "owner_url": "https://github.com/TheCraigHewitt", "description": "A specialized Claude Code workspace for creating long-form, SEO-optimized blog content for any business. This system helps you research, write, analyze, and optimize content that ranks well and serves your target audience.", "language": "Python", "stars_total": 5553, "forks_total": 793, "stars_in_period": 2526, "period_label": "最近7天", "description_zh": "专门的 Claude Code 工作区，用于为任何企业创建长篇、SEO 优化的博客内容。该系统可帮助您研究、撰写、分析和优化排名良好并服务于目标受众的内容。", "readme_summary_zh": "专门的 Claude Code 工作区，用于为任何企业创建长篇、SEO 优化的博客内容。该系统可帮助您研究、撰写、分析和优化排名良好并服务于目标受众的内容。 SEO Machine 基于 Claude Code 构建，并提供：", "source_period": "weekly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 8, "stars_in_period": 2526 } }, "composite_score": 0.197021, "delta_since_last_run": 92, "previous_rank": 82 }, { "rank": 75, "repo_path": "mobile-dev-inc/Maestro", "owner": "mobile-dev-inc", "name": "Maestro", "repo_url": "https://github.com/mobile-dev-inc/Maestro", "owner_url": "https://github.com/mobile-dev-inc", "description": "Painless E2E Automation for Mobile and Web", "language": "Kotlin", "stars_total": 13508, "forks_total": 781, "stars_in_period": 2494, "period_label": "最近30天", "description_zh": "适用于移动和 Web 的无痛 E2E 自动化", "readme_summary_zh": "Maestro 是一个面向 Android、iOS 与 Web 应用的开源 UI 和端到端测试框架，目标是让测试编写与执行更简单、更快速。它借鉴了 Appium、Espresso、XCTest、Selenium、Playwright 等工具的经验，使用人类可读的 YAML Flow 来描述操作与断言，例如启动应用、点击元素、检查可见性等，并通过解释执行方式实现无需编译的快速迭代。Maestro 支持跨平台测试，可运行在模拟器、仿真器、真实设备和浏览器上，适用于 React Native、Flutter 及混合应用；同时内置智能等待与抗波动能力，减少因动态界面带来的脆弱性和手动 sleep 调用。它尤其适合需要统一维护移动端和 Web 自动化测试、追求上手快、可扩展且稳定执行流程的团队与测试场景。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 2, "stars_in_period": 2494 } }, "composite_score": 0.069861, "delta_since_last_run": 9, "previous_rank": 74 }, { "rank": 76, "repo_path": "google-ai-edge/LiteRT-LM", "owner": "google-ai-edge", "name": "LiteRT-LM", "repo_url": "https://github.com/google-ai-edge/LiteRT-LM", "owner_url": "https://github.com/google-ai-edge", "description": "暂无简介", "language": "C++", "stars_total": 3335, "forks_total": 313, "stars_in_period": 2401, "period_label": "最近30天", "description_zh": "暂无简介", "readme_summary_zh": "LiteRT-LM 是 Google 的生产就绪、高性能、开源推理框架，用于在边缘设备上部署大型语言模型。 🔗 产品网站", "source_period": "monthly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 4, "stars_in_period": 2210 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 1, "stars_in_period": 2401 } }, "composite_score": 0.266503, "delta_since_last_run": 56, "previous_rank": 78 }, { "rank": 77, "repo_path": "krahets/hello-algo", "owner": "krahets", "name": "hello-algo", "repo_url": "https://github.com/krahets/hello-algo", "owner_url": "https://github.com/krahets", "description": "《Hello 算法》：动画图解、一键运行的数据结构与算法教程。支持简中、繁中、English、日本語，提供 Python, Java, C++, C, C#, JS, Go, Swift, Rust, Ruby, Kotlin, TS, Dart 等代码实现", "language": "Java", "stars_total": 125067, "forks_total": 15038, "stars_in_period": 2368, "period_label": "最近30天", "description_zh": "《Hello 算法》：动画图解、一键运行的数据结构与算法教程。支持简中、繁中、English、日本語，提供 Python, Java, C++, C, C#, JS, Go, Swift, Rust, Ruby, Kotlin, TS, Dart 等代码实现", "readme_summary_zh": "Hello 算法是一个面向初学者的开源免费数据结构与算法入门教程，目标是用更直观、低门槛的方式帮助读者建立完整的知识框架。项目以动画图解讲解核心概念，降低理解难度，并配套可一键运行的示例代码，让读者在动手实践中掌握常见数据结构、算法思想以及底层实现原理。它不仅适合零基础或刚接触算法的学生、自学者和转码人群，也适合希望用更清晰方式复习基础知识的开发者；在课堂辅助学习、面试准备、编程训练和入门教学等场景中都很实用，同时项目仍持续更新，并鼓励社区共同完善内容与多语言代码支持。", "source_period": "monthly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 7, "stars_in_period": 584 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 3, "stars_in_period": 2368 } }, "composite_score": 0.251153, "delta_since_last_run": 29, "previous_rank": 75 }, { "rank": 78, "repo_path": "jaywcjlove/awesome-mac", "owner": "jaywcjlove", "name": "awesome-mac", "repo_url": "https://github.com/jaywcjlove/awesome-mac", "owner_url": "https://github.com/jaywcjlove", "description": " This project is dedicated to collecting high-quality macOS software and organizing them systematically by different categories for easy search and use.", "language": "Swift", "stars_total": 101742, "forks_total": 7600, "stars_in_period": 2339, "period_label": "最近30天", "description_zh": " 本项目致力于收集优质 macOS 软件，并按不同类别进行系统整理，方便查找和使用。", "readme_summary_zh": "awesome-mac 是一个面向 macOS 用户的高质量软件精选清单，目标是把分散的优秀应用按用途系统化整理，帮助用户更高效地发现、筛选和使用合适的软件。项目覆盖范围非常广，核心能力在于以清晰分类汇总大量常用与专业工具，包括阅读与写作、文本编辑器、办公、Markdown、笔记、日记、电子书、RSS，以及开发领域的 IDE、开发辅助工具、正则编辑器、API 开发与分析、网络分析、混合应用框架、版本控制、虚拟化、数据库和终端应用，同时也涵盖设计、原型设计、思维导图、截图与录屏等类别。它特别适合希望快速搭建 macOS 软件工具箱的新用户，也适合开发者、设计师、内容创作者和重度效率用户按场景寻找替代方案、比较同类工具并持续关注优质应用生态。", "source_period": "monthly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 9, "stars_in_period": 533 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 5, "stars_in_period": 2339 } }, "composite_score": 0.249276, "delta_since_last_run": -34, "previous_rank": 77 }, { "rank": 79, "repo_path": "shadps4-emu/shadPS4", "owner": "shadps4-emu", "name": "shadPS4", "repo_url": "https://github.com/shadps4-emu/shadPS4", "owner_url": "https://github.com/shadps4-emu", "description": "PlayStation 4 emulator for Windows, Linux, macOS and FreeBSD written in C++", "language": "C++", "stars_total": 30715, "forks_total": 2086, "stars_in_period": 2297, "period_label": "最近30天", "description_zh": "用 C++ 编写的适用于 Windows、Linux、macOS 和 FreeBSD 的 PlayStation 4 模拟器", "readme_summary_zh": "shadPS4 是一个以 C++ 编写、面向 Windows、Linux 和 macOS 的早期 PlayStation 4 模拟器核心项目，主要目标是在 PC 上逐步实现 PS4 游戏的运行与兼容。该仓库提供的是底层模拟核心而非图形界面，更适合开发者、进阶用户或希望自行集成前端的人使用。项目当前仍处于早期开发阶段，体验尚不完美，但已经能够成功运行《血源诅咒》《黑暗之魂：重制版》《荒野大镖客》等部分游戏，体现出其在游戏执行、兼容性推进和跨平台支持方面的核心能力。总体来看，它最适合用于关注 PS4 模拟技术进展、测试游戏兼容性、参与开发与调试，或愿意尝试尚未完全成熟模拟器的用户。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 3, "stars_in_period": 2297 } }, "composite_score": 0.069478, "delta_since_last_run": 2, "previous_rank": 76 }, { "rank": 80, "repo_path": "MHSanaei/3x-ui", "owner": "MHSanaei", "name": "3x-ui", "repo_url": "https://github.com/MHSanaei/3x-ui", "owner_url": "https://github.com/MHSanaei", "description": "Xray panel supporting multi-protocol multi-user expire day & traffic & IP limit (Vmess, Vless, Trojan, ShadowSocks, Wireguard, Tunnel, Mixed, HTTP, Tun)", "language": "HTML", "stars_total": 33698, "forks_total": 6312, "stars_in_period": 2272, "period_label": "最近30天", "description_zh": "Xray面板支持多协议多用户过期日、流量和IP限制（ Vmess、Vless、特洛伊木马、ShadowSocks、Wireguard、Tunnel、Mixed、HTTP、TUN ）", "readme_summary_zh": "3X-UI 是一个面向 Xray-core 服务器的高级开源 Web 控制面板，主要用于更直观地管理和维护各类 VPN 与代理服务配置。它在原始 X-UI 的基础上增强而来，重点提供更友好的图形化操作界面、更好的稳定性、更广泛的协议支持，以及对服务配置与运行状态的监控能力，帮助用户减少手工编辑配置文件的复杂度，更方便地完成节点管理、参数调整和日常查看。这个项目更适合个人学习、测试和自用场景，尤其适用于希望通过网页方式管理 Xray-core 的用户；同时 README 明确提醒其仅供个人使用，不应用于非法用途，也不建议部署在生产环境中。", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 5, "stars_in_period": 126 }, "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 1, "stars_in_period": 636 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 9, "stars_in_period": 2272 } }, "composite_score": 0.457853, "delta_since_last_run": 26, "previous_rank": 80 }, { "rank": 81, "repo_path": "rowboatlabs/rowboat", "owner": "rowboatlabs", "name": "rowboat", "repo_url": "https://github.com/rowboatlabs/rowboat", "owner_url": "https://github.com/rowboatlabs", "description": "Open-source AI coworker, with memory", "language": "TypeScript", "stars_total": 11800, "forks_total": 1098, "stars_in_period": 2271, "period_label": "最近7天", "description_zh": "开源人工智能同事，有记忆", "readme_summary_zh": "开源 AI 同事将工作转化为知识图并对其进行操作 Rowboat 连接到您的电子邮件和会议记录，构建长期存在的知识图，并使用该上下文帮助您在您的机器上私下完成工作。", "source_period": "weekly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 4, "stars_in_period": 507 }, "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 9, "stars_in_period": 2271 } }, "composite_score": 0.411526, "delta_since_last_run": 280, "previous_rank": 84 }, { "rank": 82, "repo_path": "vercel-labs/agent-browser", "owner": "vercel-labs", "name": "agent-browser", "repo_url": "https://github.com/vercel-labs/agent-browser", "owner_url": "https://github.com/vercel-labs", "description": "Browser automation CLI for AI agents", "language": "Rust", "stars_total": 28548, "forks_total": 1732, "stars_in_period": 2152, "period_label": "最近7天", "description_zh": "用于 AI 代理的浏览器自动化 CLI", "readme_summary_zh": "agent-browser 是一个面向 AI 代理的浏览器自动化命令行工具，主打以原生 Rust 实现带来的速度、稳定性与较低资源开销，适合作为智能体执行网页交互任务时的底层能力。它的核心用途是让 AI 能通过 CLI 驱动浏览器完成页面访问、信息读取以及常见操作流程，从而把网页环境纳入自动化工作流之中。相比偏重通用测试框架的方案，这个项目更强调为代理式调用而设计，便于集成到需要快速、可靠浏览器控制的系统里。它尤其适合需要让大模型或自动化代理处理网站导航、页面操作、数据采集和基于浏览器的任务执行等场景，是构建 AI 驱动网页代理能力时很合适的基础工具。", "source_period": "weekly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 12, "stars_in_period": 2152 } }, "composite_score": 0.191819, "delta_since_last_run": 41, "previous_rank": 81 }, { "rank": 83, "repo_path": "google-labs-code/stitch-skills", "owner": "google-labs-code", "name": "stitch-skills", "repo_url": "https://github.com/google-labs-code/stitch-skills", "owner_url": "https://github.com/google-labs-code", "description": "暂无简介", "language": "TypeScript", "stars_total": 4279, "forks_total": 487, "stars_in_period": 1988, "period_label": "最近30天", "description_zh": "暂无简介", "readme_summary_zh": "旨在与 Stitch MCP 服务器配合使用的代理技能库。每个技能都遵循 Agent Skills 开放标准，以便与 Antigravity、Gemini CLI、Claude Code、Cursor 等编码代理兼容。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 13, "stars_in_period": 1988 } }, "composite_score": 0.067349, "delta_since_last_run": 22, "previous_rank": 85 }, { "rank": 84, "repo_path": "2dust/v2rayNG", "owner": "2dust", "name": "v2rayNG", "repo_url": "https://github.com/2dust/v2rayNG", "owner_url": "https://github.com/2dust", "description": "A V2Ray client for Android, support Xray core and v2fly core", "language": "Kotlin", "stars_total": 53729, "forks_total": 7167, "stars_in_period": 1970, "period_label": "最近30天", "description_zh": "一款适用于Android的V2Ray客户端，支持Xray核心和v2fly核心", "readme_summary_zh": "Android 版 V2Ray 客户端，支持 Xray 核心和 v2fly 核心 github_2dust", "source_period": "monthly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 3, "stars_in_period": 569 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 21, "stars_in_period": 1970 } }, "composite_score": 0.249488, "delta_since_last_run": 40, "previous_rank": 87 }, { "rank": 85, "repo_path": "localsend/localsend", "owner": "localsend", "name": "localsend", "repo_url": "https://github.com/localsend/localsend", "owner_url": "https://github.com/localsend", "description": "An open-source cross-platform alternative to AirDrop", "language": "Dart", "stars_total": 78003, "forks_total": 4165, "stars_in_period": 1951, "period_label": "最近30天", "description_zh": "AirDrop 的开源跨平台替代方案", "readme_summary_zh": "LocalSend 是一款免费开源的跨平台局域网传输工具，面向 Android、iOS、macOS、Windows 和 Linux 等设备，主要用于在同一本地网络内安全地分享文件与消息。它通过 REST API 与 HTTPS 加密在设备之间直接通信，不依赖互联网连接、云端账户或第三方服务器，因此相比依赖外部服务的传输或消息应用，更注重隐私、速度与稳定性。这个项目的核心能力是在附近设备间实现便捷的点对点传输与本地通信，适合家庭、办公室、教室、会议现场等网络可达但不方便使用公网服务的场景，也特别适用于临时快速分享照片、文档、安装包或文本信息的需求。", "source_period": "monthly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 15, "stars_in_period": 436 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 10, "stars_in_period": 1951 } }, "composite_score": 0.243981, "delta_since_last_run": 25, "previous_rank": 86 }, { "rank": 86, "repo_path": "mudler/LocalAI", "owner": "mudler", "name": "LocalAI", "repo_url": "https://github.com/mudler/LocalAI", "owner_url": "https://github.com/mudler", "description": "LocalAI is the open-source AI engine. Run any model - LLMs, vision, voice, image, video - on any hardware. No GPU required.", "language": "Go", "stars_total": 45223, "forks_total": 3909, "stars_in_period": 1916, "period_label": "最近30天", "description_zh": "LocalAI 是开源人工智能引擎。在任何硬件上运行任何模型 - 法学硕士、视觉、语音、图像、视频。无需 GPU。", "readme_summary_zh": "LocalAI 是开源人工智能引擎。在任何硬件上运行任何模型 - 法学硕士、视觉、语音、图像、视频。无需 GPU。由 Ettore Di Giacinto 创建和维护。", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 13, "stars_in_period": 76 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 4, "stars_in_period": 1916 } }, "composite_score": 0.266957, "delta_since_last_run": 30, "previous_rank": 88 }, { "rank": 87, "repo_path": "xianyu110/awesome-openclaw-tutorial", "owner": "xianyu110", "name": "awesome-openclaw-tutorial", "repo_url": "https://github.com/xianyu110/awesome-openclaw-tutorial", "owner_url": "https://github.com/xianyu110", "description": "从零开始玩转OpenClaw：最全面的中文教程，涵盖安装、配置、实战案例和避坑指南（github版）", "language": "Shell", "stars_total": 4023, "forks_total": 589, "stars_in_period": 1913, "period_label": "最近30天", "description_zh": "从零开始玩转OpenClaw：最全面的中文教程，涵盖安装、配置、实战案例和避坑指南（github版）", "readme_summary_zh": "从零开始打造你的AI工作助手：最全面的中文教程，涵盖安装、配置、实战案例和避坑指南 📖 纸质书《OpenClaw超级个体实操手册》已上市！ 清华大学出版社出版，在开源教程基础上全面重写+逐条验证。🛒 京东专属购买链接（¥42，原价¥59.8）", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 8, "stars_in_period": 1913 } }, "composite_score": 0.06818, "delta_since_last_run": 6, "previous_rank": 83 }, { "rank": 88, "repo_path": "iflytek/astron-agent", "owner": "iflytek", "name": "astron-agent", "repo_url": "https://github.com/iflytek/astron-agent", "owner_url": "https://github.com/iflytek", "description": "Enterprise-grade, commercial-friendly agentic workflow platform for building next-generation SuperAgents.", "language": "Java", "stars_total": 9645, "forks_total": 963, "stars_in_period": 1813, "period_label": "最近30天", "description_zh": "企业级、商业友好的代理工作流程平台，用于构建下一代超级代理。", "readme_summary_zh": "English | 简体中文 Astron Agent is an enterprise-grade, commercial-friendly Agentic Workflow development platform that integrates AI workflow orchestration, model management, AI and MCP tool integration, RPA automation, and team collaboration features. The platform supports high-availability deployment, enabling organizations to rapidly build scalable, production-ready intelligent agent applications and establish their.", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 14, "stars_in_period": 62 }, "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 10, "stars_in_period": 441 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 17, "stars_in_period": 1813 } }, "composite_score": 0.442296, "delta_since_last_run": -747, "previous_rank": 90 }, { "rank": 89, "repo_path": "tmux/tmux", "owner": "tmux", "name": "tmux", "repo_url": "https://github.com/tmux/tmux", "owner_url": "https://github.com/tmux", "description": "tmux source code", "language": "C", "stars_total": 44354, "forks_total": 2576, "stars_in_period": 1795, "period_label": "最近30天", "description_zh": "tmux 源代码", "readme_summary_zh": "tmux 是一款面向命令行环境的终端复用器，用来在同一个终端窗口中创建、管理和切换多个终端会话、窗口与窗格，让用户能够在单屏内同时处理多项任务。它的核心能力在于会话可分离与重新附着，也就是说即使关闭当前终端连接，正在运行的程序仍可在后台持续执行，稍后再接回继续操作，因此非常适合远程登录服务器、长时间运行任务、软件开发、系统运维以及需要稳定保持工作现场的场景。该项目可运行在 OpenBSD、FreeBSD、NetBSD、Linux、macOS 和 Solaris 等多种类 Unix 系统上，通常依赖 libevent 与 ncurses 等基础库，并需要常见的 C 语言构建工具进行编译。", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 3, "stars_in_period": 81 }, "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 5, "stars_in_period": 431 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 7, "stars_in_period": 1795 } }, "composite_score": 0.453614, "delta_since_last_run": 20, "previous_rank": 89 }, { "rank": 90, "repo_path": "tailcallhq/forgecode", "owner": "tailcallhq", "name": "forgecode", "repo_url": "https://github.com/tailcallhq/forgecode", "owner_url": "https://github.com/tailcallhq", "description": "AI enabled pair programmer for Claude, GPT, O Series, Grok, Deepseek, Gemini and 300+ models", "language": "Rust", "stars_total": 6426, "forks_total": 1329, "stars_in_period": 1778, "period_label": "最近30天", "description_zh": "适用于 Claude、GPT、O Series、Grok、Deepseek、Gemini 和 300 多个型号的 AI 结对编程器", "readme_summary_zh": "将 AI 功能与您的开发环境集成的综合编码代理", "source_period": "monthly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 3, "stars_in_period": 676 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 10, "stars_in_period": 1778 } }, "composite_score": 0.252167, "delta_since_last_run": 13, "previous_rank": 93 }, { "rank": 91, "repo_path": "nelvko/clash-for-linux-install", "owner": "nelvko", "name": "clash-for-linux-install", "repo_url": "https://github.com/nelvko/clash-for-linux-install", "owner_url": "https://github.com/nelvko", "description": "😼 优雅地使用基于 clash/mihomo 的代理环境", "language": "Shell", "stars_total": 11710, "forks_total": 1336, "stars_in_period": 1746, "period_label": "最近30天", "description_zh": "😼 优雅地使用基于 clash/mihomo 的代理环境", "readme_summary_zh": "clash-for-linux-install 是一个面向 Linux 环境的 Clash/Mihomo 代理内核一键安装与管理脚本，主要用于快速在服务器、桌面系统及容器环境中部署可用的代理运行环境。它支持 root 和普通用户，适配主流 Linux 发行版以及 AutoDL 等场景，能够自动识别系统架构、初始化系统和端口占用情况，下载匹配的内核与依赖，并生成相应的服务管理配置；当订阅需要处理时，还可调用 subconverter 进行本地转换。项目不仅提供代理内核的启停与升级，还能同步设置系统代理，支持通过 Web 控制台可视化切换节点、查看日志，并通过 Mixin 对订阅配置进行深度合并、覆盖和扩展，适合希望低成本搭建、长期维护并灵活定制 Clash 代理环境的个人用户、开发者以及远程 Linux 主机使用场景。", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 4, "stars_in_period": 56 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 13, "stars_in_period": 1746 } }, "composite_score": 0.269683, "delta_since_last_run": 10, "previous_rank": 92 }, { "rank": 92, "repo_path": "starship/starship", "owner": "starship", "name": "starship", "repo_url": "https://github.com/starship/starship", "owner_url": "https://github.com/starship", "description": "☄🌌️ The minimal, blazing-fast, and infinitely customizable prompt for any shell!", "language": "Rust", "stars_total": 56174, "forks_total": 2460, "stars_in_period": 1744, "period_label": "最近30天", "description_zh": "☄🌌️ 适用于任何 shell 的最小、速度极快且可无限定制的提示！", "readme_summary_zh": "Starship 是一个面向命令行用户的跨平台 Shell 提示符工具，主打极简外观、极高性能和几乎无限的可定制性，适用于各种操作系统和不同类型的 Shell 环境。它的核心作用是把当前目录、版本控制状态、编程语言运行时、工具链环境等关键信息以清晰紧凑的方式展示在提示符中，让用户在输入命令前就能快速掌握上下文。相比传统提示符，Starship 更强调速度与统一体验，同时支持对样式、显示内容和行为进行细致配置，并兼容许多常用开发工具。它尤其适合需要频繁切换项目、依赖多种语言或工具、追求高效终端工作流的开发者、运维人员和高级命令行用户。", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 16, "stars_in_period": 80 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 14, "stars_in_period": 1744 } }, "composite_score": 0.263371, "delta_since_last_run": 21, "previous_rank": 91 }, { "rank": 93, "repo_path": "memstechtips/Winhance", "owner": "memstechtips", "name": "Winhance", "repo_url": "https://github.com/memstechtips/Winhance", "owner_url": "https://github.com/memstechtips", "description": "Application designed to optimize, customize and enhance your Windows experience.", "language": "C#", "stars_total": 8388, "forks_total": 409, "stars_in_period": 1716, "period_label": "最近30天", "description_zh": "应用程序旨在优化、自定义和增强您的 Windows 体验。", "readme_summary_zh": "Winhance 是一款 C# 应用程序，旨在消除、优化和自定义您的 Windows 体验。从软件管理到系统优化和定制，Winhance 提供增强 Windows 10 和 11 系统所需的一切。 Winhance 具有与 UnattendedWinstall 大部分相同的增强功能，无需全新安装 Windows。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 10, "stars_in_period": 1716 } }, "composite_score": 0.067616, "delta_since_last_run": 15, "previous_rank": 96 }, { "rank": 94, "repo_path": "Diolinux/PhotoGIMP", "owner": "Diolinux", "name": "PhotoGIMP", "repo_url": "https://github.com/Diolinux/PhotoGIMP", "owner_url": "https://github.com/Diolinux", "description": "A Patch for GIMP 3+ for Photoshop Users", "language": "CSS", "stars_total": 8350, "forks_total": 258, "stars_in_period": 1709, "period_label": "最近30天", "description_zh": "面向 Photoshop 用户的 GIMP 3+ 补丁", "readme_summary_zh": "PhotoGIMP 是一个面向 GIMP 的免费社区补丁，目标是把这款开源图像编辑器调整成更接近 Adobe Photoshop 的使用体验，特别适合从 Photoshop 转向 GIMP 的用户快速上手。它并不替代 GIMP 的核心功能，而是在界面与默认配置层面进行优化，例如重新排列工具面板以贴近 Photoshop 的布局习惯，提供定制启动画面，并通过调整默认设置尽可能扩大画布工作区域，让修图、图像合成与平面设计过程更顺手。同时，它还为系统菜单提供独立的图标和应用名称，增强整体一致性。对于希望以零成本获得接近 Photoshop 操作感、又想继续使用 GIMP 跨平台能力的个人创作者、设计学习者和轻量级专业用户来说，这个项目尤其合适。", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 1, "stars_in_period": 44 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 1, "stars_in_period": 1709 } }, "composite_score": 0.273162, "delta_since_last_run": 12, "previous_rank": 97 }, { "rank": 95, "repo_path": "zellij-org/zellij", "owner": "zellij-org", "name": "zellij", "repo_url": "https://github.com/zellij-org/zellij", "owner_url": "https://github.com/zellij-org", "description": "A terminal workspace with batteries included", "language": "Rust", "stars_total": 31289, "forks_total": 1099, "stars_in_period": 1698, "period_label": "最近30天", "description_zh": "包含电池的终端工作区", "readme_summary_zh": "Zellij 是一个面向开发者、运维人员以及重度终端用户的终端工作空间工具，本质上属于终端复用器，但它强调在强大能力与易用体验之间取得平衡。它既适合刚接触这类工具的用户快速上手，也能满足高级用户对深度定制的需求，提供可配置布局、个人自动化、真正的多人协作，以及浮动窗格、堆叠窗格等更灵活的交互方式。项目还具备插件系统，支持使用任何可编译为 WebAssembly 的语言扩展功能，并内置 Web 客户端，使用户在某些场景下甚至不必依赖传统终端即可使用。它特别适合需要长期在命令行中组织复杂会话、进行远程协作、搭建个性化工作流或提升终端操作效率的场景。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 5, "stars_in_period": 1698 } }, "composite_score": 0.068505, "delta_since_last_run": 16, "previous_rank": 94 }, { "rank": 96, "repo_path": "ollama/ollama", "owner": "ollama", "name": "ollama", "repo_url": "https://github.com/ollama/ollama", "owner_url": "https://github.com/ollama", "description": "Get up and running with Kimi-K2.5, GLM-5, MiniMax, DeepSeek, gpt-oss, Qwen, Gemma and other models.", "language": "Go", "stars_total": 168516, "forks_total": 15507, "stars_in_period": 1667, "period_label": "最近7天", "description_zh": "启动并运行 Kimi-K2.5、GLM-5、MiniMax、DeepSeek、gpt-oss、Qwen、Gemma 和其他型号。", "readme_summary_zh": "开始使用开放模型进行构建。或手动下载要点： - ollama-python； - Georgi Gerganov 创立的 llama.cpp 项目； - CLI 参考。", "source_period": "weekly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 1, "stars_in_period": 1667 } }, "composite_score": 0.193693, "delta_since_last_run": 60, "previous_rank": 95 }, { "rank": 97, "repo_path": "maboloshi/github-chinese", "owner": "maboloshi", "name": "github-chinese", "repo_url": "https://github.com/maboloshi/github-chinese", "owner_url": "https://github.com/maboloshi", "description": "GitHub 汉化插件，GitHub 中文化界面。 (GitHub Translation To Chinese)", "language": "JavaScript", "stars_total": 22946, "forks_total": 1290, "stars_in_period": 1623, "period_label": "最近30天", "description_zh": "GitHub 汉化插件，GitHub 中文化界面。 (GitHub Translation To Chinese)", "readme_summary_zh": "这是一个用于将 GitHub 网页界面全面中文化的用户脚本项目，目标是帮助中文用户更顺畅地浏览和使用 GitHub。它不仅能把菜单栏、标题、按钮等常见界面元素翻译成中文，还具备基于正则的智能匹配能力，可覆盖更多动态或复杂文本场景；同时支持对项目描述进行机器与人工结合的翻译处理，并能自动将时间等本地化信息转换为更符合中文习惯的展示方式。项目会持续更新词库，以提升翻译准确度和覆盖范围。它最适合经常使用 GitHub 但更习惯中文界面的开发者、开源贡献者、学生以及初学者，尤其适用于日常查看仓库、阅读项目页面、管理 Issue 或 Pull Request 等场景。", "source_period": "monthly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 3, "stars_in_period": 610 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 13, "stars_in_period": 1623 } }, "composite_score": 0.250922, "delta_since_last_run": 28, "previous_rank": 100 }, { "rank": 98, "repo_path": "iflytek/astron-rpa", "owner": "iflytek", "name": "astron-rpa", "repo_url": "https://github.com/iflytek/astron-rpa", "owner_url": "https://github.com/iflytek", "description": "Agent-ready RPA suite with out-of-the-box automation tools. Built for individuals and enterprises.", "language": "Java", "stars_total": 6190, "forks_total": 654, "stars_in_period": 1618, "period_label": "最近30天", "description_zh": "代理就绪的 RPA 套件，具有开箱即用的自动化工具。专为个人和企业打造。", "readme_summary_zh": "🤖 Pioneering Open-Source Enterprise RPA Desktop Application English | 简体中文", "source_period": "monthly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 9, "stars_in_period": 320 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 15, "stars_in_period": 1618 } }, "composite_score": 0.244977, "delta_since_last_run": -680, "previous_rank": 99 }, { "rank": 99, "repo_path": "Anil-matcha/Open-Higgsfield-AI", "owner": "Anil-matcha", "name": "Open-Higgsfield-AI", "repo_url": "https://github.com/Anil-matcha/Open-Higgsfield-AI", "owner_url": "https://github.com/Anil-matcha", "description": "Open-source alternative to Higgsfield AI — Free AI image generation & cinema studio with 20+ models (Flux, SDXL, Midjourney, Ideogram). Self-hosted, customizable, MIT licensed.", "language": "JavaScript", "stars_total": 4158, "forks_total": 749, "stars_in_period": 1598, "period_label": "最近30天", "description_zh": "Higgsfield AI 的开源替代方案 — 免费 AI 图像生成和电影工作室，拥有 20 多个模型（Flux、SDXL、Midjourney、Ideogram）。自托管、可定制、麻省理工学院许可。", "readme_summary_zh": "Higgsfield AI 的免费开源替代品。使用 200 多个最先进的模型生成 AI 图像和视频 - 无需封闭的生态系统或订阅费用。", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 1, "stars_in_period": 235 }, "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 2, "stars_in_period": 1381 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 12, "stars_in_period": 1598 } }, "composite_score": 0.468026, "delta_since_last_run": 81, "previous_rank": 111 }, { "rank": 100, "repo_path": "songquanpeng/one-api", "owner": "songquanpeng", "name": "one-api", "repo_url": "https://github.com/songquanpeng/one-api", "owner_url": "https://github.com/songquanpeng", "description": "LLM API 管理 & 分发系统，支持 OpenAI、Azure、Anthropic Claude、Google Gemini、DeepSeek、字节豆包、ChatGLM、文心一言、讯飞星火、通义千问、360 智脑、腾讯混元等主流模型，统一 API 适配，可用于 key 管理与二次分发。单可执行文件，提供 Docker 镜像，一键部署，开箱即用。LLM API management & key redistribution system, unifying multiple providers under a single API. Single binary, Docker-ready, with an English UI.", "language": "JavaScript", "stars_total": 31763, "forks_total": 6051, "stars_in_period": 1589, "period_label": "最近30天", "description_zh": "LLM API 管理 & 分发系统，支持 OpenAI、Azure、Anthropic Claude、Google Gemini、DeepSeek、字节豆包、ChatGLM、文心一言、讯飞星火、通义千问、360 智脑、腾讯混元等主流模型，统一 API 适配，可用于 key 管理与二次分发。单可执行文件，提供 Docker 镜像，一键部署，开箱即用。LLM API management & key redistribution system, unifying multiple providers under a single API. Single binary, Docker-ready, with an English UI.", "readme_summary_zh": "one-api 是一个将多家大模型服务统一封装为标准 OpenAI API 格式的开源网关与管理平台，目标是让开发者以尽量少的改造成本接入不同厂商的模型能力，并实现开箱即用的调用体验。它支持 OpenAI 及 Azure OpenAI、Anthropic Claude、Google Gemini、Mistral，以及百度文心、阿里通义千问、讯飞星火、智谱 ChatGLM、腾讯混元、豆包、百川等众多国内外模型，也兼容 Groq、Ollama、Cohere、DeepSeek、Cloudflare Workers AI 等生态。项目的核心价值在于统一接口、聚合多模型资源、简化切换与管理，适合需要同时对接多个大模型供应商、希望降低接入复杂度、快速搭建中转服务或企业内部 AI 接口层的团队与开发场景。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 8, "stars_in_period": 1589 } }, "composite_score": 0.067849, "delta_since_last_run": 17, "previous_rank": 98 }, { "rank": 101, "repo_path": "bol-van/zapret2", "owner": "bol-van", "name": "zapret2", "repo_url": "https://github.com/bol-van/zapret2", "owner_url": "https://github.com/bol-van", "description": "anti-dpi software", "language": "C", "stars_total": 3552, "forks_total": 127, "stars_in_period": 1557, "period_label": "最近30天", "description_zh": "抗dpi软件", "readme_summary_zh": "独立的 DPI 对策，不需要连接任何第三方服务器。 TCP 和 UDP 协议的签名分析（例如，阻止 VPN）可以帮助绕过 HTTP(S) 站点的阻止或减慢。可用于协议的部分透明混淆。该项目主要针对低功耗嵌入式设备——运行 OpenWrt 的路由器。支持传统Linux系统、FreeBSD、OpenBSD、Windows。在某些情况下，可以将解决方案独立地附加到各种固件。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 1, "stars_in_period": 1557 } }, "composite_score": 0.069087, "delta_since_last_run": null, "previous_rank": null }, { "rank": 102, "repo_path": "wavetermdev/waveterm", "owner": "wavetermdev", "name": "waveterm", "repo_url": "https://github.com/wavetermdev/waveterm", "owner_url": "https://github.com/wavetermdev", "description": "An open-source, AI-integrated, cross-platform terminal for seamless workflows", "language": "Go", "stars_total": 19408, "forks_total": 895, "stars_in_period": 1536, "period_label": "最近30天", "description_zh": "开源、人工智能集成、跨平台终端，可实现无缝工作流程", "readme_summary_zh": "Wave 是一款面向 macOS、Linux 和 Windows 的开源 AI 终端工具，目标是把传统命令行、远程运维与智能助手整合到同一工作空间中。它支持接入多种 AI 模型，可使用 OpenAI、Claude、Gemini、Azure、Perplexity 等服务，也能通过 Ollama 或 LM Studio 运行本地模型，无需注册账号。项目的核心能力包括具备上下文感知的终端 AI 助手，可读取终端输出、分析当前工作区并执行文件读写与编辑；具备可持久化的 SSH 会话，在断网、切换网络或程序重启后仍能自动重连；还内置远程文件图形编辑器、丰富的文件预览、命令块隔离监控、拖放式多面板布局，以及通过 wsh 在本地和远程会话间管理工作区与同步文件。它尤其适合开发者、运维工程师和经常处理远程服务器、日志分析、跨主机协作与 AI 辅助命令行工作的场景。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 11, "stars_in_period": 1536 } }, "composite_score": 0.067251, "delta_since_last_run": 11, "previous_rank": 101 }, { "rank": 103, "repo_path": "CoplayDev/unity-mcp", "owner": "CoplayDev", "name": "unity-mcp", "repo_url": "https://github.com/CoplayDev/unity-mcp", "owner_url": "https://github.com/CoplayDev", "description": "Unity MCP acts as a bridge, allowing AI assistants (like Claude, Cursor) to interact directly with your Unity Editor via a local MCP (Model Context Protocol) Client. Give your LLM tools to manage assets, control scenes, edit scripts, and automate tasks within Unity.", "language": "C#", "stars_total": 8321, "forks_total": 955, "stars_in_period": 1510, "period_label": "最近30天", "description_zh": "Unity MCP 充当桥梁，允许 AI 助手（如 Claude、Cursor）通过本地 MCP（模型上下文协议）客户端直接与 Unity 编辑器交互。为您提供 LLM 工具来管理资源、控制场景、编辑脚本以及在 Unity 中自动执行任务。", "readme_summary_zh": "unity-mcp 是一个把 Unity 编辑器通过 Model Context Protocol 与各类大语言模型助手连接起来的桥接项目，适合让 Claude、Cursor、VS Code 等 AI 直接参与 Unity 开发流程。它的核心价值是把编辑器能力封装成可调用工具，使模型能够管理资源与包、控制场景和多场景编辑、修改脚本、操作相机与动画、处理图形渲染与后处理、进行 ProBuilder 网格编辑，并执行构建、平台切换、批量打包与异步任务跟踪。项目还提供对 Unity API 的反射检查和官方文档获取能力，帮助模型校验接口用法；新版进一步加入完整 2D/3D 物理管理，包括碰撞层矩阵、材质、关节、射线与重叠查询、刚体配置、施力、场景级验证和编辑模式模拟。它最适合用于 AI 辅助原型开发、自动化编辑器操作、批量内容处理、技术美术流程整合，以及需要让语言模型在真实 Unity 工程中安全、高效执行复杂任务的场景。", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 11, "stars_in_period": 44 }, "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 16, "stars_in_period": 283 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 6, "stars_in_period": 1510 } }, "composite_score": 0.440635, "delta_since_last_run": 11, "previous_rank": 102 }, { "rank": 104, "repo_path": "hiddify/hiddify-app", "owner": "hiddify", "name": "hiddify-app", "repo_url": "https://github.com/hiddify/hiddify-app", "owner_url": "https://github.com/hiddify", "description": "Multi-platform auto-proxy client, supporting Sing-box, X-ray, TUIC, Hysteria, Reality, Trojan, SSH etc. It’s an open-source, secure and ad-free.", "language": "Dart", "stars_total": 28469, "forks_total": 2530, "stars_in_period": 1509, "period_label": "最近30天", "description_zh": "多平台自动代理客户端，支持Sing-box、X-ray、TUIC、Hysteria、Reality、Trojan、SSH等。开源、安全、无广告。", "readme_summary_zh": "Hiddify 是一款基于 Sing-box 通用代理工具链打造的多平台代理客户端，面向希望更便捷管理和使用代理配置的用户。它支持 Android、iOS、Windows、macOS 和 Linux，提供直观易用的界面，并集成自动选节点、基于延迟的线路切换、TUN 模式、远程配置与订阅自动更新等核心能力。项目兼容 Vless、Vmess、Reality、TUIC、Hysteria、WireGuard、SSH 等多种协议，也能导入 Sing-box、V2Ray、Clash 及 Clash Meta 等常见配置格式，同时可显示套餐剩余天数与流量信息。作为无广告、开源且强调安全与隐私的客户端，它尤其适合需要跨设备统一使用代理、管理订阅节点，或在伊朗、中国、俄罗斯等网络环境复杂地区追求稳定访问体验的场景。", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 12, "stars_in_period": 42 }, "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 3, "stars_in_period": 336 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 3, "stars_in_period": 1509 } }, "composite_score": 0.447897, "delta_since_last_run": 10, "previous_rank": 103 }, { "rank": 105, "repo_path": "coollabsio/coolify", "owner": "coollabsio", "name": "coolify", "repo_url": "https://github.com/coollabsio/coolify", "owner_url": "https://github.com/coollabsio", "description": "An open-source, self-hostable PaaS alternative to Vercel, Heroku & Netlify that lets you easily deploy static sites, databases, full-stack applications and 280+ one-click services on your own servers.", "language": "PHP", "stars_total": 52895, "forks_total": 4029, "stars_in_period": 1501, "period_label": "最近30天", "description_zh": "Vercel、Heroku 和 Netlify 的开源、自托管 PaaS 替代品，让您可以在自己的服务器上轻松部署静态站点、数据库、全栈应用程序和 280 多个一键式服务。", "readme_summary_zh": "Coolify 是一个开源且可自托管的应用部署与基础设施管理平台，可视为 Heroku、Netlify、Vercel 等云平台的替代方案。它的核心价值在于让用户只通过 SSH 连接，就能在自己的服务器上统一管理应用、数据库和运行环境，适用于 VPS、裸金属服务器、树莓派等多种硬件。这个项目强调把云平台的易用性带回到自有基础设施中，既方便部署和运维，又避免厂商锁定，因为相关配置都保存在用户自己的服务器上。即使未来不再使用 Coolify，现有资源仍可继续自行管理，只是会失去平台提供的自动化能力。它尤其适合希望掌控数据与部署流程、追求灵活性和成本可控的个人开发者、小团队以及自建服务器场景。", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 17, "stars_in_period": 48 }, "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 13, "stars_in_period": 383 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 2, "stars_in_period": 1501 } }, "composite_score": 0.440448, "delta_since_last_run": 14, "previous_rank": 104 }, { "rank": 106, "repo_path": "memvid/memvid", "owner": "memvid", "name": "memvid", "repo_url": "https://github.com/memvid/memvid", "owner_url": "https://github.com/memvid", "description": "Memory layer for AI Agents. Replace complex RAG pipelines with a serverless, single-file memory layer. Give your agents instant retrieval and long-term memory.", "language": "Rust", "stars_total": 14819, "forks_total": 1285, "stars_in_period": 1486, "period_label": "最近30天", "description_zh": "AI 代理的内存层。使用无服务器、单文件内存层替换复杂的 RAG 管道。为您的代理提供即时检索和长期记忆。", "readme_summary_zh": "Memvid 是一个用于 AI 代理的单文件存储层，具有即时检索和长期记忆功能。持久性、版本化和便携式内存，无需数据库。", "source_period": "monthly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 4, "stars_in_period": 1142 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 4, "stars_in_period": 1486 } }, "composite_score": 0.256253, "delta_since_last_run": 23, "previous_rank": 105 }, { "rank": 107, "repo_path": "decolua/9router", "owner": "decolua", "name": "9router", "repo_url": "https://github.com/decolua/9router", "owner_url": "https://github.com/decolua", "description": "Connect All AI Code Tools (Claude Code, Cursor, Antigravity, Copilot, Codex, Gemini, OpenCode, Cline, OpenClaw...) to 40+ AI Providers & 100+ Models", "language": "JavaScript", "stars_total": 2233, "forks_total": 562, "stars_in_period": 1456, "period_label": "最近30天", "description_zh": "将所有 AI 代码工具（Claude Code、Cursor、Antigravity、Copilot、Codex、Gemini、OpenCode、Cline、OpenClaw...）连接到 40 多个 AI 提供商和 100 多个模型", "readme_summary_zh": "9Router 是一个面向 AI 编程工作流的智能路由工具，目标是把 Claude Code、Cursor、Copilot、Codex、Gemini、Cline 等各类代码助手统一接入到 40 多家 AI 提供商和 100 多种模型之上，帮助开发者持续稳定地编码而不被配额、限流或高昂费用打断。它的核心能力在于自动路由与智能降级：可以优先使用已有订阅额度，在额度将尽或触发限制时自动切换到更便宜甚至免费的模型，实现近乎无感的连续使用；同时支持多账号轮询，进一步提升单一提供商下的可用性与吞吐能力。这个项目尤其适合同时使用多种 AI 编码工具、希望充分榨干订阅价值、降低 API 成本、避免手动切换模型与账号的个人开发者、小团队，以及依赖命令行或编辑器内 AI 助手进行高频开发的场景。", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 2, "stars_in_period": 82 }, "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 1, "stars_in_period": 543 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 6, "stars_in_period": 1456 } }, "composite_score": 0.457108, "delta_since_last_run": 17, "previous_rank": 109 }, { "rank": 108, "repo_path": "YishenTu/claudian", "owner": "YishenTu", "name": "claudian", "repo_url": "https://github.com/YishenTu/claudian", "owner_url": "https://github.com/YishenTu", "description": "An Obsidian plugin that embeds Claude Code as an AI collaborator in your vault", "language": "TypeScript", "stars_total": 7248, "forks_total": 428, "stars_in_period": 1454, "period_label": "最近7天", "description_zh": "一个 Obsidian 插件，可将 Claude Code 作为 AI 协作者嵌入到您的保管库中", "readme_summary_zh": "一个 Obsidian 插件，可将 AI 编码代理（Claude Code、Codex 等）嵌入到您的保管库中。您的保管库成为代理的工作目录 - 文件读/写、搜索、bash 和多步骤工作流程全部开箱即用。", "source_period": "weekly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 5, "stars_in_period": 570 }, "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 4, "stars_in_period": 1454 } }, "composite_score": 0.408935, "delta_since_last_run": 66, "previous_rank": 112 }, { "rank": 109, "repo_path": "utkarshdalal/GameNative", "owner": "utkarshdalal", "name": "GameNative", "repo_url": "https://github.com/utkarshdalal/GameNative", "owner_url": "https://github.com/utkarshdalal", "description": "Native PC gaming with Steam, Epic, GOG and Amazon integrations on Android", "language": "Kotlin", "stars_total": 5105, "forks_total": 197, "stars_in_period": 1443, "period_label": "最近30天", "description_zh": "Android 上集成 Steam、Epic、GOG 和 Amazon 的原生 PC 游戏", "readme_summary_zh": "GameNative 是一个面向安卓设备的游戏启动与运行工具，目标是让用户直接在 Android 上游玩自己已拥有的 Steam、Epic 和 GOG 游戏，并支持云存档同步，方便在不同设备间延续进度。它基于一个安卓端 Steam 客户端项目分支演进而来，目前仍处于早期开发阶段，因此并非所有游戏都能稳定运行，部分作品可能需要额外调整或兼容性优化。这个项目的核心价值在于把 PC 数字游戏平台中的个人游戏库尽可能带到移动端，适合希望在手机或平板上体验已购游戏、重视跨设备存档衔接，且愿意接受早期版本限制并进行一定调试的用户使用。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 13, "stars_in_period": 1443 } }, "composite_score": 0.066793, "delta_since_last_run": 18, "previous_rank": 106 }, { "rank": 110, "repo_path": "dolthub/dolt", "owner": "dolthub", "name": "dolt", "repo_url": "https://github.com/dolthub/dolt", "owner_url": "https://github.com/dolthub", "description": "Dolt – Git for Data", "language": "Go", "stars_total": 22070, "forks_total": 735, "stars_in_period": 1423, "period_label": "最近30天", "description_zh": "隐藏 - Git 数据", "readme_summary_zh": "Dolt 是一个把 Git 式版本控制带入关系型数据的 SQL 数据库，兼具 MySQL 兼容访问方式与类似 Git 的分支协作模型。你可以像连接普通 MySQL 一样读写表结构和数据，同时对数据库执行 fork、clone、branch、merge、push、pull 等操作，并通过系统表、函数、存储过程或命令行直接管理版本历史、提交变更、导入 CSV、合并他人修改。它的核心价值在于让“表数据”像“代码文件”一样可追踪、可比较、可协作，特别适合需要审计数据变更、多人协同编辑数据集、维护可回溯的数据版本、共享公共数据，或希望在保留现有 MySQL 的前提下通过 binlog 复制获得版本控制能力的场景。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 12, "stars_in_period": 1423 } }, "composite_score": 0.066954, "delta_since_last_run": 6, "previous_rank": 110 }, { "rank": 111, "repo_path": "ohmyzsh/ohmyzsh", "owner": "ohmyzsh", "name": "ohmyzsh", "repo_url": "https://github.com/ohmyzsh/ohmyzsh", "owner_url": "https://github.com/ohmyzsh", "description": "🙃 A delightful community-driven (with 2,400+ contributors) framework for managing your zsh configuration. Includes 300+ optional plugins (rails, git, macOS, hub, docker, homebrew, node, php, python, etc), 140+ themes to spice up your morning, and an auto-update tool that makes it easy to keep up with the latest updates from the community.", "language": "Shell", "stars_total": 186065, "forks_total": 26369, "stars_in_period": 1402, "period_label": "最近30天", "description_zh": "🙃 一个令人愉快的社区驱动（拥有 2,400 多个贡献者）框架，用于管理您的 zsh 配置。包括 300 多个可选插件（rails、git、macOS、hub、docker、homebrew、node、php、python 等）、140 多个主题让您的早晨更加有趣，还有一个自动更新工具，让您可以轻松跟上社区的最新更新。", "readme_summary_zh": "Oh My Zsh 是一个开源、社区驱动的框架，用于管理您的 zsh 配置。听起来很无聊。让我们再试一次。", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 3, "stars_in_period": 68 }, "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 7, "stars_in_period": 363 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 11, "stars_in_period": 1402 } }, "composite_score": 0.450444, "delta_since_last_run": -27, "previous_rank": 113 }, { "rank": 112, "repo_path": "immich-app/immich", "owner": "immich-app", "name": "immich", "repo_url": "https://github.com/immich-app/immich", "owner_url": "https://github.com/immich-app", "description": "High performance self-hosted photo and video management solution.", "language": "TypeScript", "stars_total": 97554, "forks_total": 5354, "stars_in_period": 1360, "period_label": "最近7天", "description_zh": "高性能自托管照片和视频管理解决方案。", "readme_summary_zh": "Català Español Français Italiano 日本語 한국어 Deutsch Nederlands Türkçe 简体中文 正體中文 Українська Русский Português Brasileiro Svenska العربية Tiếng Việt ภาษาไทย ⚠️ Always follow 3-2-1 backup plan for your precious photos and videos!", "source_period": "weekly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 8, "stars_in_period": 1360 } }, "composite_score": 0.187443, "delta_since_last_run": 39, "previous_rank": 117 }, { "rank": 113, "repo_path": "SagerNet/sing-box", "owner": "SagerNet", "name": "sing-box", "repo_url": "https://github.com/SagerNet/sing-box", "owner_url": "https://github.com/SagerNet", "description": "The universal proxy platform", "language": "Go", "stars_total": 32452, "forks_total": 3810, "stars_in_period": 1336, "period_label": "最近30天", "description_zh": "通用代理平台", "readme_summary_zh": "由 Warp 赞助，为使用多个 AI 代理进行编码而构建的通用代理平台。", "source_period": "monthly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 4, "stars_in_period": 405 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 9, "stars_in_period": 1336 } }, "composite_score": 0.249139, "delta_since_last_run": 17, "previous_rank": 115 }, { "rank": 114, "repo_path": "FreeCAD/FreeCAD", "owner": "FreeCAD", "name": "FreeCAD", "repo_url": "https://github.com/FreeCAD/FreeCAD", "owner_url": "https://github.com/FreeCAD", "description": "Official source code of FreeCAD, a free and opensource multiplatform 3D parametric modeler.", "language": "C++", "stars_total": 30238, "forks_total": 5401, "stars_in_period": 1328, "period_label": "最近30天", "description_zh": "FreeCAD 的官方源代码，一个免费的开源多平台 3D 参数化建模器。", "readme_summary_zh": "FreeCAD 是一款开源的参数化 3D 建模软件，主要用于设计各种真实世界中的物体，从小型零件到较大尺度的工程对象都能处理。它的核心优势在于参数化建模，用户可以通过回溯模型历史并修改参数，快速调整设计方案；同时支持先绘制带几何约束的 2D 草图，再据此构建 3D 模型，也能从 3D 模型中提取尺寸与细节，生成适合生产的工程图纸。项目基于 OpenCASCADE 几何内核，结合 Coin3D、Qt 图形界面和丰富的 Python API，兼顾建模能力、可扩展性与自动化开发需求。它适用于产品设计、机械工程、建筑设计以及教学学习等场景，尤其适合需要跨平台使用、重视可修改性和定制能力的爱好者、学生、教师、程序员和专业 CAD 用户。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 6, "stars_in_period": 1328 } }, "composite_score": 0.067946, "delta_since_last_run": 9, "previous_rank": 116 }, { "rank": 115, "repo_path": "junegunn/fzf", "owner": "junegunn", "name": "fzf", "repo_url": "https://github.com/junegunn/fzf", "owner_url": "https://github.com/junegunn", "description": "🌸 A command-line fuzzy finder", "language": "Go", "stars_total": 79357, "forks_total": 2761, "stars_in_period": 1320, "period_label": "最近30天", "description_zh": "⭐ 命令行模糊查找器", "readme_summary_zh": "fzf 是一个通用的命令行模糊查找器。它是一个交互式过滤程序，适用于任何类型的列表；文件、命令历史记录、进程、主机名、书签、git 提交等。它实现了“模糊”匹配算法，因此您可以快速输入省略字符的模式，但仍然可以获得您想要的结果。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 17, "stars_in_period": 1320 } }, "composite_score": 0.065942, "delta_since_last_run": -42, "previous_rank": 114 }, { "rank": 116, "repo_path": "xindoo/agentic-design-patterns", "owner": "xindoo", "name": "agentic-design-patterns", "repo_url": "https://github.com/xindoo/agentic-design-patterns", "owner_url": "https://github.com/xindoo", "description": "谷歌新书Agent设计模式(agentic design patterns)最佳中文版，持续优化。附：在线阅读、pdf和epub电子书下载。", "language": "HTML", "stars_total": 3417, "forks_total": 471, "stars_in_period": 1315, "period_label": "最近30天", "description_zh": "谷歌新书Agent设计模式(agentic design patterns)最佳中文版，持续优化。附：在线阅读、pdf和epub电子书下载。", "readme_summary_zh": "xindoo - 本翻译项目的发起者和主要维护者 本项目是《Agentic Design Patterns》一书的中文翻译项目。该书系统介绍了 AI Agent 系统的各种设计模式，涵盖从基础到高级的 21 个核心模式，以及多个附录章节。", "source_period": "monthly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 3, "stars_in_period": 217 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 5, "stars_in_period": 1315 } }, "composite_score": 0.248832, "delta_since_last_run": 11, "previous_rank": 118 }, { "rank": 117, "repo_path": "ReVanced/revanced-manager", "owner": "ReVanced", "name": "revanced-manager", "repo_url": "https://github.com/ReVanced/revanced-manager", "owner_url": "https://github.com/ReVanced", "description": "💊 Application to use ReVanced on Android", "language": "Kotlin", "stars_total": 27304, "forks_total": 1064, "stars_in_period": 1310, "period_label": "最近30天", "description_zh": "💊 在 Android 上使用 ReVanced 的应用程序", "readme_summary_zh": "ReVanced Manager 是一款面向 Android 的应用管理与补丁工具，延续了 Vanced 的使用思路，核心作用是借助 ReVanced Patcher 对安卓应用进行修补和定制，让用户能够更灵活地调整目标应用的功能与行为。它集成了应用下载、补丁选择与应用、签名处理以及主题、更新和其他相关设置的管理能力，帮助用户在一个统一界面中完成从获取应用到生成可用版本的流程。这个项目尤其适合希望对常用安卓应用进行个性化修改、统一管理补丁方案，或需要更方便地维护已修补应用的用户与进阶玩家使用。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 3, "stars_in_period": 1310 } }, "composite_score": 0.068472, "delta_since_last_run": 6, "previous_rank": 119 }, { "rank": 118, "repo_path": "fastfetch-cli/fastfetch", "owner": "fastfetch-cli", "name": "fastfetch", "repo_url": "https://github.com/fastfetch-cli/fastfetch", "owner_url": "https://github.com/fastfetch-cli", "description": "A maintained, feature-rich and performance oriented, neofetch like system information tool.", "language": "C", "stars_total": 21815, "forks_total": 735, "stars_in_period": 1308, "period_label": "最近30天", "description_zh": "一个维护的、功能丰富的、面向性能的、类似 neofetch 的系统信息工具。", "readme_summary_zh": "Fastfetch 是一款类似 neofetch 的系统信息展示工具，主要使用 C 语言编写，重点强调高性能与高度可定制性。它的用途是在终端中快速获取并以更美观的方式展示设备与系统环境信息，适合用来查看操作系统、硬件平台等基础状态，也便于用户根据配置文件自定义输出样式和内容。该项目支持 Linux、macOS、Windows 8.1 及以上、Android，以及 FreeBSD、OpenBSD、NetBSD、DragonFly、Haiku 和 SunOS（包括 illumos、Solaris）等多个系统，当前主要在 x86-64 与 aarch64 平台上进行积极测试。它尤其适合终端爱好者、开发者、运维人员和需要在多平台环境中快速确认系统信息的用户使用。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 19, "stars_in_period": 1308 } }, "composite_score": 0.065567, "delta_since_last_run": 17, "previous_rank": 122 }, { "rank": 119, "repo_path": "bggRGjQaUbCoE/PiliPlus", "owner": "bggRGjQaUbCoE", "name": "PiliPlus", "repo_url": "https://github.com/bggRGjQaUbCoE/PiliPlus", "owner_url": "https://github.com/bggRGjQaUbCoE", "description": "PiliPlus", "language": "Dart", "stars_total": 12618, "forks_total": 768, "stars_in_period": 1299, "period_label": "最近30天", "description_zh": "霹雳加", "readme_summary_zh": "PiliPlus 是一款使用 Flutter 开发的哔哩哔哩第三方客户端，面向移动端与桌面端提供统一且较完整的观看、互动与内容发布体验。它不仅覆盖视频、音频、课堂内容播放，还支持离线缓存、DLNA 投屏、跳过番剧片头片尾、视频截取动图、AI 原声翻译等增强播放能力，并在移动端提供弹幕悬停后的点赞、复制、举报等操作。在社区功能方面，它支持编辑和发布动态、评论富文本与表情显示、@用户、发起投票、查看用户图文，以及修改消息和聊天设置、展示折叠消息等。项目尤其适合希望在不同设备上获得更丰富功能、更灵活交互，以及兼顾观看、社交和内容管理体验的 B 站重度用户使用。", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 1, "stars_in_period": 54 }, "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 6, "stars_in_period": 467 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 15, "stars_in_period": 1299 } }, "composite_score": 0.451699, "delta_since_last_run": 19, "previous_rank": 123 }, { "rank": 120, "repo_path": "telegramdesktop/tdesktop", "owner": "telegramdesktop", "name": "tdesktop", "repo_url": "https://github.com/telegramdesktop/tdesktop", "owner_url": "https://github.com/telegramdesktop", "description": "Telegram Desktop messaging app", "language": "C++", "stars_total": 31349, "forks_total": 6580, "stars_in_period": 1289, "period_label": "最近30天", "description_zh": "Telegram 桌面消息应用程序", "readme_summary_zh": "这是官方 Telegram Messenger 桌面客户端的完整源代码和构建说明，基于 Telegram API 和 MTProto 安全协议。源代码在 GPLv3 下发布（OpenSSL 除外），许可证可在此处获取。", "source_period": "monthly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 3, "stars_in_period": 933 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 4, "stars_in_period": 1289 } }, "composite_score": 0.254868, "delta_since_last_run": 8, "previous_rank": 121 }, { "rank": 121, "repo_path": "zarazhangrui/frontend-slides", "owner": "zarazhangrui", "name": "frontend-slides", "repo_url": "https://github.com/zarazhangrui/frontend-slides", "owner_url": "https://github.com/zarazhangrui", "description": "Create beautiful slides on the web using Claude's frontend skills", "language": "Shell", "stars_total": 13762, "forks_total": 1042, "stars_in_period": 1275, "period_label": "最近7天", "description_zh": "使用 Claude 的前端技能在网络上创建精美的幻灯片", "readme_summary_zh": "Frontend Slides 是一个面向 Claude Code 的演示文稿生成技能，旨在帮助不会 CSS 或 JavaScript 的用户快速制作美观、富有动画效果的 HTML 幻灯片，也支持把现有 PowerPoint 转成网页演示。它的核心特点是以“先看效果再选择”为主的设计流程：系统会生成多种视觉预览，用户无需准确描述审美偏好，只需挑选喜欢的风格即可。同时，它生成的是零依赖的单个 HTML 文件，内联包含 CSS 和 JavaScript，不需要 npm、构建工具或前端框架，便于直接分发、部署和二次修改。项目还强调避免千篇一律的 AI 风格，提供更有辨识度的视觉方案，并尽量保留 PPT 中的图片与内容，输出具有可访问性、响应式和可维护性的代码。它尤其适合需要快速制作产品介绍、教学课件、路演展示或将传统 PPT 发布到 Web 上的个人与团队使用。", "source_period": "weekly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 1, "stars_in_period": 1275 } }, "composite_score": 0.190473, "delta_since_last_run": 42, "previous_rank": 107 }, { "rank": 122, "repo_path": "XTLS/Xray-core", "owner": "XTLS", "name": "Xray-core", "repo_url": "https://github.com/XTLS/Xray-core", "owner_url": "https://github.com/XTLS", "description": "Xray, Penetrates Everything. Also the best v2ray-core. Where the magic happens. An open platform for various uses.", "language": "Go", "stars_total": 36884, "forks_total": 5165, "stars_in_period": 1268, "period_label": "最近30天", "description_zh": "X射线，穿透一切。也是最好的 v2ray 核心。魔法发生的地方。适合多种用途的开放平台。", "readme_summary_zh": "Project X 源于 XTLS 协议，提供了 Xray-core 和 REALITY 等一套网络工具。自述文件已开放，因此请随时在此处提交您的项目。", "source_period": "monthly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 7, "stars_in_period": 360 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 16, "stars_in_period": 1268 } }, "composite_score": 0.245833, "delta_since_last_run": 13, "previous_rank": 124 }, { "rank": 123, "repo_path": "adongwanai/AgentGuide", "owner": "adongwanai", "name": "AgentGuide", "repo_url": "https://github.com/adongwanai/AgentGuide", "owner_url": "https://github.com/adongwanai", "description": "https://adongwanai.github.io/AgentGuide | AI Agent开发指南 | LangGraph实战 | 高级RAG | 转行大模型 | 大模型面试 | 算法工程师 | 面试题库 | 强化学习｜数据合成", "language": "HTML", "stars_total": 3480, "forks_total": 337, "stars_in_period": 1242, "period_label": "最近30天", "description_zh": "https://adongwanai.github.io/AgentGuide | AI Agent开发指南 | LangGraph实战 | 高级RAG | 转行大模型 | 大模型面试 | 算法工程师 | 面试题库 | 强化学习｜数据合成", "readme_summary_zh": "AgentGuide 是一个面向 AI Agent 学习与求职的系统化指南项目，定位并非简单罗列资料，而是围绕“从入门到拿 Offer”构建完整路径，帮助学习者高效理解 Agent 开发与大模型岗位所需能力。它整合了优质课程、教程、论文与开源项目，在此基础上提供清晰的六步学习路线、算法岗与开发岗的岗位选择建议、按周规划的学习路线图，以及围绕 LangGraph、RAG、上下文工程、监督微调、强化学习等主题的技术教程。同时，项目强调实战与求职导向，配套实战项目合集、Agent 项目案例，以及大量面试题、系统设计题和编程题，并标注知识点在面试和简历中的应用方式。它尤其适合想从零系统学习 AI Agent、从传统开发转向大模型方向、准备相关岗位面试，或希望用更明确路线避免碎片化学习的人使用。", "source_period": "monthly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 5, "stars_in_period": 253 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 4, "stars_in_period": 1242 } }, "composite_score": 0.24817, "delta_since_last_run": 13, "previous_rank": 126 }, { "rank": 124, "repo_path": "gedoor/legado", "owner": "gedoor", "name": "legado", "repo_url": "https://github.com/gedoor/legado", "owner_url": "https://github.com/gedoor", "description": "Legado 3.0 Book Reader with powerful controls & full functions❤️阅读3.0, 阅读是一款可以自定义来源阅读网络内容的工具，为广大网络文学爱好者提供一种方便、快捷舒适的试读体验。", "language": "Kotlin", "stars_total": 45799, "forks_total": 5258, "stars_in_period": 1238, "period_label": "最近30天", "description_zh": "Legado 3.0 Book Reader with powerful controls & full functions❤️阅读3.0, 阅读是一款可以自定义来源阅读网络内容的工具，为广大网络文学爱好者提供一种方便、快捷舒适的试读体验。", "readme_summary_zh": "Legado（开源阅读）是一款面向 Android 的免费开源小说阅读器，定位为不直接提供内容、而是通过用户自行导入书源、订阅源和各类规则来构建个人阅读环境的工具。它的核心能力在于高度灵活的内容接入与阅读配置，例如支持书源规则、RSS 订阅、替换规则、本地 TXT 目录规则、在线朗读引擎、主题和排版配置等导入，还提供基于 URL 的一键唤起导入，以及 Web 方式和 Content Provider 方式的 API，便于与其他应用或服务联动。项目适合希望自定义小说、资讯等文本内容聚合与阅读体验的用户，也适合需要二次开发、自动化接入或构建个性化阅读工作流的开发者与进阶读者使用。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 6, "stars_in_period": 1238 } }, "composite_score": 0.067855, "delta_since_last_run": 10, "previous_rank": 125 }, { "rank": 125, "repo_path": "Dimillian/Skills", "owner": "Dimillian", "name": "Skills", "repo_url": "https://github.com/Dimillian/Skills", "owner_url": "https://github.com/Dimillian", "description": "My Codex Skills", "language": "Shell", "stars_total": 3251, "forks_total": 167, "stars_in_period": 1234, "period_label": "最近30天", "description_zh": "我的法典技能", "readme_summary_zh": "这是一个面向工程团队的可复用“技能”集合仓库，主要服务于 Apple 平台开发以及常见的软件交付流程，把一些高频、可重复的工程任务整理成独立、聚焦、可直接复用的能力模块。它覆盖的核心能力包括生成 App Store 发布说明、排查和调试 iOS 应用问题、优化 SwiftUI 与 React 代码和性能、打包 macOS 应用、执行多代理协作的差异审查与缺陷排查、开展代码评审与变更简化、组织更大规模的重构工作，以及审计项目真正缺少哪些新技能。这个项目最适合需要沉淀工程方法、提升研发效率与代码质量的个人开发者、Apple 生态应用团队，以及希望将评审、调试、重构和工作流标准化的项目环境。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 7, "stars_in_period": 1234 } }, "composite_score": 0.067669, "delta_since_last_run": 12, "previous_rank": 127 }, { "rank": 126, "repo_path": "Lordog/dive-into-llms", "owner": "Lordog", "name": "dive-into-llms", "repo_url": "https://github.com/Lordog/dive-into-llms", "owner_url": "https://github.com/Lordog", "description": "《动手学大模型Dive into LLMs》系列编程实践教程", "language": "Jupyter Notebook", "stars_total": 26594, "forks_total": 3185, "stars_in_period": 1224, "period_label": "最近7天", "description_zh": "《动手学大模型Dive into LLMs》系列编程实践教程", "readme_summary_zh": "项目动机 / 教程目录 / 贡献者列表 2025/06/06 感谢各位朋友们的关注和积极反馈！我们从以下两个方面对本教程进行了更新： Key points: - 上线国产化《大模型开发全流程》公益教程（含PPT、实验手册和视频），此处特别感谢华为昇腾社区的支持！; - 在原系列编程实践教程的基础上进行内容更新，并增加了新的主题（数学推理、GUI Agent、大模型对齐、隐写术等）！; - 🚀 前往昇腾社区探索《大模型开发全流程》系列课程 ： 👉《 大模型开发学习专区 》@ 昇腾社区 👈.", "source_period": "weekly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 1, "stars_in_period": 168 }, "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 12, "stars_in_period": 1224 } }, "composite_score": 0.392117, "delta_since_last_run": 33, "previous_rank": 128 }, { "rank": 127, "repo_path": "AppFlowy-IO/AppFlowy", "owner": "AppFlowy-IO", "name": "AppFlowy", "repo_url": "https://github.com/AppFlowy-IO/AppFlowy", "owner_url": "https://github.com/AppFlowy-IO", "description": "Bring projects, wikis, and teams together with AI. AppFlowy is the AI collaborative workspace where you achieve more without losing control of your data. The leading open source Notion alternative.", "language": "Dart", "stars_total": 69528, "forks_total": 5085, "stars_in_period": 1223, "period_label": "最近30天", "description_zh": "通过 AI 将项目、wiki 和团队结合在一起。 AppFlowy 是人工智能协作工作空间，您可以在不失去对数据控制的情况下实现更多目标。领先的开源 Notion 替代方案。", "readme_summary_zh": "AppFlowy 是一款以数据自主可控为核心的 AI 工作空间，旨在帮助个人与团队在不牺牲隐私和所有权的前提下，更高效地组织知识、协作办公与管理项目。它将文档、笔记、任务、数据库等能力整合到同一平台中，并结合 AI 功能提升内容创作、信息整理和工作流执行效率，同时强调开放、可定制与可自托管，让用户能够根据自身需求灵活部署和扩展。它尤其适合重视数据安全、希望替代传统封闭式协作工具的团队，以及需要在知识管理、项目推进和日常协作之间建立统一工作空间的个人和组织。", "source_period": "monthly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 1, "stars_in_period": 589 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 1, "stars_in_period": 1223 } }, "composite_score": 0.253585, "delta_since_last_run": 19, "previous_rank": 129 }, { "rank": 128, "repo_path": "Facepunch/sbox-public", "owner": "Facepunch", "name": "sbox-public", "repo_url": "https://github.com/Facepunch/sbox-public", "owner_url": "https://github.com/Facepunch", "description": "s&box is a modern game engine, built on Valve's Source 2 and the latest .NET technology, it provides a modern intuitive editor for creating games", "language": "C#", "stars_total": 4989, "forks_total": 409, "stars_in_period": 1188, "period_label": "最近30天", "description_zh": "s&box 是一款现代游戏引擎，基于 Valve 的 Source 2 和最新的 .NET 技术构建，它为创建游戏提供了现代直观的编辑器", "readme_summary_zh": "s&box 是一款现代游戏引擎，基于 Valve 的 Source 2 和最新的 .NET 技术构建，它为创建游戏提供了现代直观的编辑器。如果您的目标是使用 s&box 创建游戏，请从入门指南开始。该存储库用于为那些想要为引擎开发做出贡献的人从源代码构建引擎。", "source_period": "monthly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 1, "stars_in_period": 574 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 1, "stars_in_period": 1188 } }, "composite_score": 0.253426, "delta_since_last_run": 26, "previous_rank": 131 }, { "rank": 129, "repo_path": "anthropics/courses", "owner": "anthropics", "name": "courses", "repo_url": "https://github.com/anthropics/courses", "owner_url": "https://github.com/anthropics", "description": "Anthropic's educational courses", "language": "Jupyter Notebook", "stars_total": 20440, "forks_total": 2072, "stars_in_period": 1186, "period_label": "最近30天", "description_zh": "Anthropic的教育课程", "readme_summary_zh": "这个仓库是 Anthropic 提供的 Claude 学习课程集合，面向希望系统掌握其 API 与提示词实践的开发者和产品团队。目前内容涵盖五门课程，并建议按顺序学习：先通过 API 基础课程了解获取密钥、模型参数设置、多模态提示与流式响应等核心用法，再通过交互式提示工程教程掌握关键提示技巧，随后进入真实世界提示课程，把这些方法应用到更复杂的业务场景中；此外还包括提示词评测课程，用于构建生产环境中的质量衡量方法，以及工具使用课程，帮助将 Claude 与外部工具和工作流稳定集成。整体上，这个项目最适合用于 Claude 入门培训、团队能力建设，以及在实际产品开发中建立从提示设计到评估与工具调用的完整实践路径。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 2, "stars_in_period": 1186 } }, "composite_score": 0.068527, "delta_since_last_run": 21, "previous_rank": 130 }, { "rank": 130, "repo_path": "HKUDS/LightRAG", "owner": "HKUDS", "name": "LightRAG", "repo_url": "https://github.com/HKUDS/LightRAG", "owner_url": "https://github.com/HKUDS", "description": "[EMNLP2025] \"LightRAG: Simple and Fast Retrieval-Augmented Generation\"", "language": "Python", "stars_total": 32882, "forks_total": 4683, "stars_in_period": 1136, "period_label": "最近7天", "description_zh": "[EMNLP2025]“LightRAG：简单快速的检索增强生成”", "readme_summary_zh": "- [2025.11]🎯[新功能]：集成RAGAS用于评估和Langfuse用于跟踪。更新了 API 以返回检索到的上下文以及查询结果，以支持上下文精度指标。 - [2025.06]🎯[新功能]LightRAG 现在通过 RAG-Anything 集成支持全面的多模式数据处理，实现跨多种格式（包括 PDF、图像、Office 文档、表格和公式）的无缝文档解析和 RAG 功能。详情请参阅新的多式联运部分。", "source_period": "weekly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 15, "stars_in_period": 1136 } }, "composite_score": 0.181876, "delta_since_last_run": 37, "previous_rank": 120 }, { "rank": 131, "repo_path": "unchihugo/FluentFlyout", "owner": "unchihugo", "name": "FluentFlyout", "repo_url": "https://github.com/unchihugo/FluentFlyout", "owner_url": "https://github.com/unchihugo", "description": "The modern Flyout app for Windows 11, built with Fluent 2 Design principles. Media Flyouts, Taskbar Widgets and more.", "language": "C#", "stars_total": 2257, "forks_total": 89, "stars_in_period": 1067, "period_label": "最近30天", "description_zh": "适用于 Windows 11 的现代 Flyout 应用程序，采用 Fluent 2 设计原则构建。媒体弹出窗口、任务栏小部件等。", "readme_summary_zh": "FluentFlyout 是一款面向 Windows 的现代化弹出控制工具，遵循 Fluent 2 设计语言打造，整体视觉与 Windows 11 高度融合，能以更原生、干净且不中断操作的方式管理媒体播放与系统状态。它的核心能力包括音频弹窗中显示封面、歌曲标题、艺术家及播放控制，提供“下一首”信息预览，快速显示大小写锁定等按键状态，并可通过任务栏小组件直接查看媒体信息；同时支持浅色与深色模式、跟随系统配色、Mica 模糊效果、流畅动画以及弹窗位置等个性化设置，还包含单曲循环、列表循环和随机播放等常用功能。这个项目尤其适合希望替换默认媒体与状态提示、追求更统一 Windows 11 视觉体验，或经常在听歌、办公和多任务场景中快速查看与控制信息的用户使用。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 4, "stars_in_period": 1067 } }, "composite_score": 0.068043, "delta_since_last_run": 6, "previous_rank": 134 }, { "rank": 132, "repo_path": "venera-app/venera", "owner": "venera-app", "name": "venera", "repo_url": "https://github.com/venera-app/venera", "owner_url": "https://github.com/venera-app", "description": "A comic app", "language": "Dart", "stars_total": 9195, "forks_total": 303, "stars_in_period": 1063, "period_label": "最近30天", "description_zh": "一款漫画应用", "readme_summary_zh": "Venera 是一款面向漫画阅读的应用，既能阅读本地漫画，也能通过网络漫画源在线获取内容。它的核心特点是支持使用 JavaScript 自定义和扩展漫画源，因此不仅可以适配不同网站或服务，还能让阅读来源更灵活。除了基本的浏览与阅读功能外，它还提供收藏管理、漫画下载，以及在数据源支持的情况下查看评论、标签等作品信息；若对应来源允许登录，用户还可以进行评论、评分等互动操作。整体来看，这个项目特别适合希望把本地与在线漫画统一管理、需要可扩展内容源，并重视离线阅读与社区互动体验的用户。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 4, "stars_in_period": 1063 } }, "composite_score": 0.068039, "delta_since_last_run": 8, "previous_rank": 133 }, { "rank": 133, "repo_path": "tensorflow/tensorflow", "owner": "tensorflow", "name": "tensorflow", "repo_url": "https://github.com/tensorflow/tensorflow", "owner_url": "https://github.com/tensorflow", "description": "An Open Source Machine Learning Framework for Everyone", "language": "C++", "stars_total": 194648, "forks_total": 75278, "stars_in_period": 1047, "period_label": "最近30天", "description_zh": "适合所有人的开源机器学习框架", "readme_summary_zh": "TensorFlow 是一个面向机器学习的端到端开源平台，旨在为研究人员与开发者提供从模型设计、训练到部署的完整能力。它拥有较为完善且灵活的生态体系，涵盖多种工具、库和社区资源，既能支持前沿机器学习与神经网络研究，也适合构建和落地实际的智能应用。项目最初由 Google Brain 的研究人员与工程师开发，但框架本身通用性很强，并不局限于特定研究方向。TensorFlow 提供稳定的 Python 和 C++ API，同时也支持其他语言接口，因此适用于学术研究、原型验证、生产环境部署以及需要将机器学习能力集成到应用或服务中的场景。", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 10, "stars_in_period": 43 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 17, "stars_in_period": 1047 } }, "composite_score": 0.264374, "delta_since_last_run": 22, "previous_rank": 135 }, { "rank": 134, "repo_path": "microsoft/ML-For-Beginners", "owner": "microsoft", "name": "ML-For-Beginners", "repo_url": "https://github.com/microsoft/ML-For-Beginners", "owner_url": "https://github.com/microsoft", "description": "12 weeks, 26 lessons, 52 quizzes, classic Machine Learning for all", "language": "Jupyter Notebook", "stars_total": 85104, "forks_total": 20526, "stars_in_period": 1038, "period_label": "最近30天", "description_zh": "12 周、26 节课程、52 个测验，适合所有人的经典机器学习", "readme_summary_zh": "阿拉伯语 |孟加拉语 |保加利亚语 |缅甸语（缅甸） |中文（简体）|中文（繁体，香港）|中文（繁体，澳门）|中文（繁体，台湾）|克罗地亚语 |捷克语 |丹麦语 |荷兰语 |爱沙尼亚语 |芬兰语 |法语 |德语 |希腊语 |希伯来语 |印地语 |匈牙利语 |印度尼西亚语 |意大利语 |日语 |卡纳达语 |韩语 |立陶宛语 |马来语 |马拉雅拉姆语 |马拉地语 |尼泊尔语 |尼日利亚洋泾浜语 |挪威语 |波斯语（波斯语）|波兰语 |葡萄牙语（巴西）|葡萄牙语（葡萄牙）|旁遮普语（古尔穆奇语）|罗马尼亚语 |俄语 |塞尔维亚语（西里尔文）|斯洛伐克语 |斯洛文尼亚语 |西班牙语 |斯瓦希里语 |瑞典语 |他加禄语（菲律宾语）|泰米尔语 |泰卢固语 |泰语 |土耳其语 |乌克兰语 |乌尔都语 |越南人更喜欢本地克隆？", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 10, "stars_in_period": 1038 } }, "composite_score": 0.066925, "delta_since_last_run": 5, "previous_rank": 137 }, { "rank": 135, "repo_path": "screenpipe/screenpipe", "owner": "screenpipe", "name": "screenpipe", "repo_url": "https://github.com/screenpipe/screenpipe", "owner_url": "https://github.com/screenpipe", "description": "Run agents that work for you based on what you do. AI finally knows what you are doing", "language": "Rust", "stars_total": 18112, "forks_total": 1609, "stars_in_period": 1031, "period_label": "最近30天", "description_zh": "根据您的工作运行适合您的代理。 AI终于知道你在做什么", "readme_summary_zh": "Screenpipe 是一个把电脑使用过程转化为“AI 记忆”的本地化开源项目，旨在让你的设备成为了解你工作与操作上下文的个人 AI。它会在后台围绕你的屏幕活动进行记录、索引与检索，帮助你记住看过、听过和做过的内容，并通过自然语言搜索快速找到相关信息。在此基础上，项目还支持运行基于工作行为触发的 agents（称为 pipes），让自动化流程随着你的实际操作在后台执行，适合需要持续记录数字工作轨迹、提升个人知识回溯能力，以及希望在严格保护隐私前提下进行本地 AI 自动化的人群与场景。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 11, "stars_in_period": 1031 } }, "composite_score": 0.066736, "delta_since_last_run": 6, "previous_rank": 138 }, { "rank": 136, "repo_path": "farouqaldori/claude-island", "owner": "farouqaldori", "name": "claude-island", "repo_url": "https://github.com/farouqaldori/claude-island", "owner_url": "https://github.com/farouqaldori", "description": "Claude Code notifications without the context switch. A minimal, always-present session manager for macOS.", "language": "Swift", "stars_total": 1921, "forks_total": 276, "stars_in_period": 1015, "period_label": "最近30天", "description_zh": "无需上下文切换的克劳德代码通知。适用于 macOS 的最小且始终存在的会话管理器。", "readme_summary_zh": "macOS 菜单栏应用程序，可为 Claude Code CLI 会话带来动态岛式通知。", "source_period": "monthly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 2, "stars_in_period": 623 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 6, "stars_in_period": 1015 } }, "composite_score": 0.252212, "delta_since_last_run": 12, "previous_rank": 140 }, { "rank": 137, "repo_path": "ml-explore/mlx", "owner": "ml-explore", "name": "mlx", "repo_url": "https://github.com/ml-explore/mlx", "owner_url": "https://github.com/ml-explore", "description": "MLX: An array framework for Apple silicon", "language": "C++", "stars_total": 25295, "forks_total": 1673, "stars_in_period": 1002, "period_label": "最近30天", "description_zh": "MLX：Apple 芯片的阵列框架", "readme_summary_zh": "MLX 是苹果面向 Apple silicon 推出的机器学习数组框架，主要用于在 Mac 等设备上高效地研究、训练与部署模型。它提供接近 NumPy 的 Python 接口，并同时支持 C++、C 和 Swift，另外还通过 mlx.nn、mlx.optimizers 等高层模块提供类似 PyTorch 的建模体验，便于快速搭建复杂网络。MLX 的核心能力包括可组合的函数变换，如自动求导、自动向量化和计算图优化，以及惰性计算和动态图机制，使调试更直观、输入形状变化时也不会引发缓慢编译。其另一大特点是支持 CPU 与 GPU 的多设备运行，并利用 Apple silicon 的统一内存模型，让数组可在共享内存中直接被不同设备访问，减少数据搬运开销。这个项目尤其适合在苹果芯片平台上进行模型原型验证、算法研究、快速实验，以及语言模型训练、LoRA 微调和 Stable Diffusion 图像生成等场景。", "source_period": "monthly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 12, "stars_in_period": 271 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 10, "stars_in_period": 1002 } }, "composite_score": 0.243256, "delta_since_last_run": 11, "previous_rank": 139 }, { "rank": 138, "repo_path": "systemd/systemd", "owner": "systemd", "name": "systemd", "repo_url": "https://github.com/systemd/systemd", "owner_url": "https://github.com/systemd", "description": "The systemd System and Service Manager", "language": "C", "stars_total": 16195, "forks_total": 4436, "stars_in_period": 997, "period_label": "最近30天", "description_zh": "systemd 系统和服务管理器", "readme_summary_zh": "systemd 是一个面向 Linux 的系统与服务管理套件，主要用于在系统启动后统一接管初始化流程、后台服务、设备与会话等基础运行环节，是现代 Linux 发行版中最核心的底层组件之一。它的核心能力在于提供高效的服务启动与依赖管理机制，支持按需激活、并行启动、服务监控、日志整合以及系统状态控制，从而帮助管理员和开发者更稳定地管理系统生命周期与各类守护进程。该项目也为开发、调试和扩展提供了较完整的代码结构说明、构建要求、测试与贡献规范，适合操作系统发行版维护者、基础设施工程师、服务器运维人员以及需要深度集成系统服务管理能力的 Linux 开发场景使用。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 4, "stars_in_period": 997 } }, "composite_score": 0.067972, "delta_since_last_run": -1, "previous_rank": 136 }, { "rank": 139, "repo_path": "exelban/stats", "owner": "exelban", "name": "stats", "repo_url": "https://github.com/exelban/stats", "owner_url": "https://github.com/exelban", "description": "macOS system monitor in your menu bar", "language": "Swift", "stars_total": 37788, "forks_total": 1249, "stars_in_period": 973, "period_label": "最近30天", "description_zh": "菜单栏中的 macOS 系统监视器", "readme_summary_zh": "Stats 是一款面向 macOS 的轻量级系统监控工具，主要用于在菜单栏中实时展示电脑运行状态，让用户无需打开“活动监视器”就能快速掌握关键硬件与系统信息。它的核心能力通常包括对 CPU、内存、磁盘、网络、电池以及传感器等指标的持续监测，并以紧凑直观的菜单栏组件形式呈现，便于随时查看更详细的数据变化。这个项目尤其适合希望提升系统状态可见性的人群，例如开发者、运维人员、内容创作者，或经常关注设备性能、功耗与资源占用情况的普通 macOS 用户，用于日常性能观察、排查卡顿发热问题，以及优化工作过程中的资源使用体验。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 7, "stars_in_period": 973 } }, "composite_score": 0.067403, "delta_since_last_run": 8, "previous_rank": 141 }, { "rank": 140, "repo_path": "zhukunpenglinyutong/jetbrains-cc-gui", "owner": "zhukunpenglinyutong", "name": "jetbrains-cc-gui", "repo_url": "https://github.com/zhukunpenglinyutong/jetbrains-cc-gui", "owner_url": "https://github.com/zhukunpenglinyutong", "description": "Jetbrains Claude Code and Codex GUI Plugin", "language": "TypeScript", "stars_total": 2504, "forks_total": 312, "stars_in_period": 955, "period_label": "最近30天", "description_zh": "Jetbrains Claude Code 和 Codex GUI 插件", "readme_summary_zh": "Originally Claude Code GUI English · 简体中文", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 6, "stars_in_period": 41 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 11, "stars_in_period": 955 } }, "composite_score": 0.267605, "delta_since_last_run": 0, "previous_rank": 145 }, { "rank": 141, "repo_path": "OpenBMB/VoxCPM", "owner": "OpenBMB", "name": "VoxCPM", "repo_url": "https://github.com/OpenBMB/VoxCPM", "owner_url": "https://github.com/OpenBMB", "description": "VoxCPM2: Tokenizer-Free TTS for Multilingual Speech Generation, Creative Voice Design, and True-to-Life Cloning", "language": "Python", "stars_total": 8983, "forks_total": 1058, "stars_in_period": 953, "period_label": "最近24小时", "description_zh": "VoxCPM2：用于多语言语音生成、创意语音设计和逼真克隆的无分词器 TTS", "readme_summary_zh": "English | 中文 VoxCPM is a tokenizer-free Text-to-Speech system that directly generates continuous speech representations via an end-to-end diffusion autoregressive architecture , bypassing discrete tokenization to achieve highly natural and expressive synthesis.", "source_period": "daily", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 9, "stars_in_period": 953 } }, "composite_score": 0.228646, "delta_since_last_run": 582, "previous_rank": 144 }, { "rank": 142, "repo_path": "osaurus-ai/osaurus", "owner": "osaurus-ai", "name": "osaurus", "repo_url": "https://github.com/osaurus-ai/osaurus", "owner_url": "https://github.com/osaurus-ai", "description": "Own your AI. The native macOS harness for AI agents -- any model, persistent memory, autonomous execution, cryptographic identity. Built in Swift. Fully offline. Open source.", "language": "Swift", "stars_total": 4905, "forks_total": 237, "stars_in_period": 943, "period_label": "最近30天", "description_zh": "拥有你的人工智能。用于 AI 代理的本机 macOS 工具 - 任何模型、持久内存、自主执行、加密身份。内置于斯威夫特。完全离线。开源。", "readme_summary_zh": "Osaurus 是一个面向 macOS 的本地 AI 基础层，目标是把真正属于用户的上下文、记忆、工具和身份保留在自己的 Mac 上，而不是托管在他人的服务器中。它以纯 Swift 为 Apple Silicon 原生构建，可作为你与各类 AI 模型之间的“控制中枢”：既能在完全离线状态下配合本地模型运行，也能按需接入云端模型以获得更强能力。项目的核心在于提供可持续积累的 AI 体验，包括具备记忆的智能体、自主执行任务、运行真实代码，以及保持随时可访问的能力，让模型可替换而个人化能力不断沉淀。它特别适合重视隐私与数据掌控、希望在本机建立长期 AI 工作流的开发者、研究者和高频个人生产力用户使用。", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 1, "stars_in_period": 47 }, "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 1, "stars_in_period": 394 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 3, "stars_in_period": 943 } }, "composite_score": 0.455351, "delta_since_last_run": 8, "previous_rank": 148 }, { "rank": 143, "repo_path": "harness/harness", "owner": "harness", "name": "harness", "repo_url": "https://github.com/harness/harness", "owner_url": "https://github.com/harness", "description": "Harness Open Source is an end-to-end developer platform with Source Control Management, CI/CD Pipelines, Hosted Developer Environments, and Artifact Registries.", "language": "Go", "stars_total": 34498, "forks_total": 2971, "stars_in_period": 941, "period_label": "最近30天", "description_zh": "Harness Open Source 是一个端到端开发人员平台，具有源代码控制管理、CI/CD 管道、托管开发人员环境和 Artifact 注册表。", "readme_summary_zh": "Harness Open Source 是一个面向团队协作的软件开发与 DevOps 开源平台，旨在把代码托管、自动化流水线、托管式开发环境 Gitspaces 以及制品仓库整合到同一套系统中，为研发流程提供从代码管理到构建交付的端到端能力。它是在 Drone 持续集成基础上的进一步演进，不再只关注 CI，而是扩展到更完整的开发平台形态，并以逐步实现与 Drone 流水线能力对齐、支持平滑迁移为目标。该项目特别适合希望自建研发基础设施、统一开发与交付工具链、提升协作效率的团队使用，也适用于重视开源可控性、需要集中管理代码、构建任务、开发环境和制品的组织。", "source_period": "monthly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 12, "stars_in_period": 324 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 7, "stars_in_period": 941 } }, "composite_score": 0.244174, "delta_since_last_run": -81, "previous_rank": 156 }, { "rank": 144, "repo_path": "Snailclimb/JavaGuide", "owner": "Snailclimb", "name": "JavaGuide", "repo_url": "https://github.com/Snailclimb/JavaGuide", "owner_url": "https://github.com/Snailclimb", "description": "Java 面试 & 后端通用面试指南，覆盖计算机基础、数据库、分布式、高并发、系统设计与 AI 应用开发", "language": "Java", "stars_total": 154822, "forks_total": 46152, "stars_in_period": 934, "period_label": "最近30天", "description_zh": "Java 面试 & 后端通用面试指南，覆盖计算机基础、数据库、分布式、高并发、系统设计与 AI 应用开发", "readme_summary_zh": "JavaGuide 是一个面向 Java 后端开发学习与求职面试的开源知识库，核心目标是帮助开发者系统梳理后端基础、构建完整知识体系，并高效准备面试。项目内容覆盖 Java 基础、后端常见知识点与高频面试题总结，同时提供学习路线、面试重点、简历编写、项目经验整理以及校招求职等实用指导；此外还持续补充 AI 应用开发相关面试内容，涉及大模型基础、Agent、RAG、MCP 协议等当前热门方向。它尤其适合准备 Java 后端面试的学生、初中级开发者，以及希望查漏补缺、快速建立知识框架的人群，在自学成长、求职冲刺和面试复盘等场景中都很有价值。", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 20, "stars_in_period": 32 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 7, "stars_in_period": 934 } }, "composite_score": 0.259965, "delta_since_last_run": 8, "previous_rank": 143 }, { "rank": 145, "repo_path": "ventoy/Ventoy", "owner": "ventoy", "name": "Ventoy", "repo_url": "https://github.com/ventoy/Ventoy", "owner_url": "https://github.com/ventoy", "description": "A new bootable USB solution.", "language": "C", "stars_total": 75755, "forks_total": 4750, "stars_in_period": 929, "period_label": "最近30天", "description_zh": "一种新的可启动 USB 解决方案。", "readme_summary_zh": "Windows Windows 7、Windows 8、Windows 8.1、Windows 10、Windows 11、Windows Server 2012、Windows Server 2012 R2、Windows Server 2016、Windows Server 2019、Windows Server 2022、Windows Server 2025、WinPE Linux Debian、Ubuntu、CentOS(6/7/8/9/10)、RHEL(6/7/8/9/10)、深度操作系统、 Fedora、Rocky Linux、AlmaLinux、EuroLinux(6/7/8/9)、openEuler、OpenAnolis、SLES、openSUSE、MX Linux、Manjaro、Linux Mint、Endless OS、Elementary OS、Solus、Linx、Zorin、antiX、PClinuxOS、Arch、ArcoLinux、ArchLabs、BlackArch、Obarun、Artix Linux、Puppy Linux、Tails、Slax、Kali、Mageia、 Slackware、Q4OS、Archman、Gentoo、Pentoo、NixOS、Kylin、openKylin、Ubuntu Kylin、KylinSec、Lubuntu、Xubuntu、Kubuntu、Ubuntu MATE、Ubuntu Budgie、Ubuntu Studio、Bluestar、OpenMandriva、ExTiX、Netrunner、ALT Linux、Nitrux、Peppermint、KDE neon、Linux Lite、Parrot OS、Qubes、Pop OS、ROSA、 Void Linux、Star Linux、EndeavourOS、MakuluLinux、Voyager、Feren、ArchBang、LXLE、Knoppix、Calculate Linux、Clear Linux、Pure OS、Oracle Linux、Trident、Septor、Porteus、Devuan、GoboLinux、4MLinux、Simplicity Linux、Zeroshell、Android-x86、netboot。", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 15, "stars_in_period": 40 }, "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 1, "stars_in_period": 314 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 10, "stars_in_period": 929 } }, "composite_score": 0.44513, "delta_since_last_run": 9, "previous_rank": 146 }, { "rank": 146, "repo_path": "danielmiessler/SecLists", "owner": "danielmiessler", "name": "SecLists", "repo_url": "https://github.com/danielmiessler/SecLists", "owner_url": "https://github.com/danielmiessler", "description": "SecLists is the security tester's companion. It's a collection of multiple types of lists used during security assessments, collected in one place. List types include usernames, passwords, URLs, sensitive data patterns, fuzzing payloads, web shells, and many more.", "language": "PHP", "stars_total": 70167, "forks_total": 24963, "stars_in_period": 927, "period_label": "最近30天", "description_zh": "SecLists 是安全测试人员的伴侣。它是安全评估期间使用的多种类型列表的集合，并收集在一个位置。列表类型包括用户名、密码、URL、敏感数据模式、模糊测试负载、Web shell 等等。", "readme_summary_zh": "SecLists 是安全测试人员的伴侣。它是安全评估期间使用的多种类型列表的集合，并收集在一个位置。列表类型包括用户名、密码、URL、敏感数据模式、模糊测试负载、Web shell 等等。目标是使安全测试人员能够将此存储库拉到新的测试箱上，并有权访问可能需要的每种类型的列表。该项目由 Daniel Miessler、Jason Haddix、Ignacio Portal 和 g0tmi1k 维护。", "source_period": "monthly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 2, "stars_in_period": 306 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 12, "stars_in_period": 927 } }, "composite_score": 0.24843, "delta_since_last_run": 11, "previous_rank": 147 }, { "rank": 147, "repo_path": "aishwaryanr/awesome-generative-ai-guide", "owner": "aishwaryanr", "name": "awesome-generative-ai-guide", "repo_url": "https://github.com/aishwaryanr/awesome-generative-ai-guide", "owner_url": "https://github.com/aishwaryanr", "description": "A one stop repository for generative AI research updates, interview resources, notebooks and much more!", "language": "HTML", "stars_total": 26072, "forks_total": 5510, "stars_in_period": 913, "period_label": "最近30天", "description_zh": "生成式人工智能研究更新、访谈资源、笔记本等的一站式存储库！", "readme_summary_zh": "生成式人工智能正在快速发展，这个存储库是生成式人工智能研究、采访材料、笔记本等更新的综合中心！探索以下资源：", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 4, "stars_in_period": 31 }, "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 12, "stars_in_period": 193 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 12, "stars_in_period": 913 } }, "composite_score": 0.443936, "delta_since_last_run": 22, "previous_rank": 149 }, { "rank": 148, "repo_path": "nageoffer/ragent", "owner": "nageoffer", "name": "ragent", "repo_url": "https://github.com/nageoffer/ragent", "owner_url": "https://github.com/nageoffer", "description": "企业级 Agentic RAG 智能体 - 全链路覆盖文档解析、多路检索、意图识别、问题重写、会话记忆、MCP 工具调用与深度思考。面向真实业务场景，从 0 到 1 完整工程实现。", "language": "Java", "stars_total": 1405, "forks_total": 287, "stars_in_period": 903, "period_label": "最近30天", "description_zh": "企业级 Agentic RAG 智能体 - 全链路覆盖文档解析、多路检索、意图识别、问题重写、会话记忆、MCP 工具调用与深度思考。面向真实业务场景，从 0 到 1 完整工程实现。", "readme_summary_zh": "从 0 到 1 纯手工打造，一套真正能落地、经得起推敲的 Agentic RAG 系统。深入底层设计细节，覆盖全链路核心能力。配套完整的 RAG 学习路线、项目实战、部署教程与简历写法。 新年气象，2026 年春节假期最后一天，企业级 AI RAG 正式发布！", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 16, "stars_in_period": 19 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 14, "stars_in_period": 903 } }, "composite_score": 0.260545, "delta_since_last_run": 11, "previous_rank": 150 }, { "rank": 149, "repo_path": "9seconds/mtg", "owner": "9seconds", "name": "mtg", "repo_url": "https://github.com/9seconds/mtg", "owner_url": "https://github.com/9seconds", "description": "Highly opinionated MTPROTO proxy for Telegram", "language": "Go", "stars_total": 3238, "forks_total": 341, "stars_in_period": 901, "period_label": "最近30天", "description_zh": "备受好评的 Telegram MTPROTO 代理", "readme_summary_zh": "mtg 是一个面向 Telegram 的 MTPROTO 代理项目，定位非常明确：做一个精简、无冗余、易恢复部署的代理，而不是追求复杂功能堆叠。它不提供用户管理，也不强调广告标签这类依赖中间代理和公网暴露的能力，而是专注于在受限或审查环境中稳定提供 Telegram 连接。项目的核心能力包括支持域名前置：当请求校验失败、误访问或出现重放防护等情况时，会自动回退到真实网站并原样转发响应，避免连接行为显得异常；同时它还会模拟所伪装网站的 TLS 流量统计特征与分块模式，让代理流量更像普通站点访问，从而提高隐蔽性。此外，mtg 注重资源利用效率，目标是在合理规模下承载约一到两万并发连接。它最适合需要低维护、易部署、注重抗审查与伪装能力的 Telegram 代理场景。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 8, "stars_in_period": 901 } }, "composite_score": 0.067148, "delta_since_last_run": 3, "previous_rank": 151 }, { "rank": 150, "repo_path": "amnezia-vpn/amnezia-client", "owner": "amnezia-vpn", "name": "amnezia-client", "repo_url": "https://github.com/amnezia-vpn/amnezia-client", "owner_url": "https://github.com/amnezia-vpn", "description": "Amnezia VPN Client (Desktop+Mobile)", "language": "C++", "stars_total": 11281, "forks_total": 805, "stars_in_period": 898, "period_label": "最近30天", "description_zh": "Amnesia VPN 客户端（桌面+移动）", "readme_summary_zh": "Amnezia 是一款开源 VPN 客户端，主打“自建 VPN”能力，适合希望自己掌控服务器与连接方式的个人用户、技术爱好者以及对隐私和可用性有较高要求的场景。它的核心特点是可通过输入服务器 IP、SSH 账号和密码，自动在远程服务器上部署 VPN 所需的 Docker 容器并完成连接，显著降低了自建门槛。项目同时支持 OpenVPN、WireGuard、IKEv2 等经典协议，也提供基于流量伪装与混淆的方案，如 OpenVPN over Cloak、Shadowsocks、AmneziaWG 和 XRay，以增强在复杂网络环境下的可用性。此外，它具备分流功能，可按网站或应用选择是否走 VPN，并覆盖 Windows、macOS、Linux、Android 和 iOS 多个平台，因此特别适合需要跨设备、安全访问、自主管理网络出口，或在受限网络环境中提升连接成功率的用户。", "source_period": "monthly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 10, "stars_in_period": 236 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 9, "stars_in_period": 898 } }, "composite_score": 0.244109, "delta_since_last_run": 6, "previous_rank": 154 }, { "rank": 151, "repo_path": "SakanaAI/AI-Scientist", "owner": "SakanaAI", "name": "AI-Scientist", "repo_url": "https://github.com/SakanaAI/AI-Scientist", "owner_url": "https://github.com/SakanaAI", "description": "The AI Scientist: Towards Fully Automated Open-Ended Scientific Discovery 🧑‍🔬", "language": "Jupyter Notebook", "stars_total": 13165, "forks_total": 1882, "stars_in_period": 881, "period_label": "最近30天", "description_zh": "人工智能科学家：迈向全自动开放式科学发现🧑‍🔬", "readme_summary_zh": "AI Scientist 是一个面向“自动化科学发现”的研究系统，目标是让大模型不仅辅助人类做头脑风暴或写代码，而是能够较为独立地完成完整科研流程。它围绕基础模型尤其是大语言模型构建，支持在不同研究模板中自动提出想法、撰写与运行代码、执行实验、分析结果并生成论文式成果，从而形成端到端的科研闭环。项目同时提供论文实验中的运行结果与数据，并展示了在扩散模型、字符级语言建模、Transformer 学习率与 grokking 现象等方向生成的示例论文。它最适合用于探索“AI 能否自主做研究”这一前沿课题，以及用于快速验证开放式科研想法和搭建自动科研代理原型，但由于系统会执行模型生成的代码，存在安全与资源管理风险，更适合在受限容器和严格网络控制环境中由具备工程与研究经验的用户使用。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 3, "stars_in_period": 881 } }, "composite_score": 0.068035, "delta_since_last_run": 4, "previous_rank": 152 }, { "rank": 152, "repo_path": "Atmosphere-NX/Atmosphere", "owner": "Atmosphere-NX", "name": "Atmosphere", "repo_url": "https://github.com/Atmosphere-NX/Atmosphere", "owner_url": "https://github.com/Atmosphere-NX", "description": "Atmosphère is a work-in-progress customized firmware for the Nintendo Switch.", "language": "C++", "stars_total": 18854, "forks_total": 1438, "stars_in_period": 881, "period_label": "最近30天", "description_zh": "Atmosphère 是一款正在开发中的 Nintendo Switch 定制固件。", "readme_summary_zh": "Atmosphère 是一个面向 Nintendo Switch 的定制固件项目，目前仍在持续开发中，目标是以模块化方式替换或扩展原版系统的关键启动与运行组件，从而为主机提供更灵活的底层控制与功能增强。它由多个核心部分组成：Fusée 负责首阶段引导、加载并校验后续启动内容及内核相关模块；Exosphère 提供定制化的 TrustZone 与安全监控能力；Thermosphère 实现基于 EL2 的 EmuNAND 支持，用于备份并虚拟化或重定向 NAND 映像；Stratosphère 通过自定义系统模块扩展内核能力、重实现加载器并挂钩重要系统行为；Troposphère 则在 Horizon OS 的应用层提供补丁，以实现常见的 CFW 特性。这个项目最适合希望研究 Switch 启动链、安全机制、系统模块扩展，或需要使用虚拟 NAND、底层补丁与高级自定义功能的开发者与进阶用户。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 8, "stars_in_period": 881 } }, "composite_score": 0.067128, "delta_since_last_run": -1, "previous_rank": 157 }, { "rank": 153, "repo_path": "agentscope-ai/agentscope-java", "owner": "agentscope-ai", "name": "agentscope-java", "repo_url": "https://github.com/agentscope-ai/agentscope-java", "owner_url": "https://github.com/agentscope-ai", "description": "AgentScope Java: Agent-Oriented Programming for Building LLM Applications", "language": "Java", "stars_total": 2538, "forks_total": 528, "stars_in_period": 880, "period_label": "最近30天", "description_zh": "AgentScope Java：用于构建 LLM 应用程序的面向代理编程", "readme_summary_zh": "AgentScope Java 是一个面向智能体编程的 Java 框架，专门用于构建由大语言模型驱动的应用。它以 ReAct 推理—行动范式为核心，让智能体能够在执行复杂任务时自主规划、选择工具并根据环境变化动态调整，而不受僵化流程编排限制。项目同时强调生产可控性，提供安全中断、优雅取消和基于 Hook 的人类介入能力，便于在运行中暂停、纠偏或终止任务而不破坏上下文状态。它还内置 PlanNotebook 任务分解与跟踪、可自纠错的结构化输出、支持语义检索的长期记忆，以及对企业知识库友好的 RAG 能力，并可通过 MCP 协议扩展外部工具、通过 A2A 协议实现分布式多智能体协作。这个项目尤其适合企业级智能助手、复杂业务自动化、知识问答与多步骤决策系统等需要自治能力、可观测性和系统集成能力并重的场景。", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 19, "stars_in_period": 21 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 4, "stars_in_period": 880 } }, "composite_score": 0.260675, "delta_since_last_run": 1, "previous_rank": 153 }, { "rank": 154, "repo_path": "unslothai/notebooks", "owner": "unslothai", "name": "notebooks", "repo_url": "https://github.com/unslothai/notebooks", "owner_url": "https://github.com/unslothai", "description": "250+ Fine-tuning & RL Notebooks for text, vision, audio, embedding, TTS models.", "language": "Jupyter Notebook", "stars_total": 5202, "forks_total": 837, "stars_in_period": 878, "period_label": "最近30天", "description_zh": "250 多个用于文本、视觉、音频、嵌入、TTS 模型的微调和 RL 笔记本。", "readme_summary_zh": "这个项目是 Unsloth 提供的一组按模型分类整理的 Notebook 示例，主要用于帮助用户在本地或 Colab 环境中快速完成大模型微调全流程，包括数据准备、训练以及推理验证。它的核心价值在于把常见模型的实践步骤做成可直接运行的笔记本，适合想要上手监督微调、测试训练配置、验证推理效果的开发者、研究者和工程团队。README 还特别强调了运行环境中的若干关键兼容性问题，例如 NumPy 2.x 与 soxr 的冲突会导致 Unsloth 无法导入，soxr 重装可能再次拉回不兼容版本，过旧的 typing_extensions 会影响 PyTorch 导入，以及在 ROCm/AMD GPU 上使用 torch.compile 时可能出现训练崩溃。这些说明表明该项目尤其适合需要参考稳定实践、排查环境依赖问题并快速复现实验流程的用户。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 5, "stars_in_period": 878 } }, "composite_score": 0.067669, "delta_since_last_run": 4, "previous_rank": 132 }, { "rank": 155, "repo_path": "jurplel/InstantSpaceSwitcher", "owner": "jurplel", "name": "InstantSpaceSwitcher", "repo_url": "https://github.com/jurplel/InstantSpaceSwitcher", "owner_url": "https://github.com/jurplel", "description": "Native space switching on macOS with no animation", "language": "Swift", "stars_total": 1090, "forks_total": 31, "stars_in_period": 875, "period_label": "最近7天", "description_zh": "macOS 上的原生空间切换没有动画", "readme_summary_zh": "macOS 上的本机即时工作区切换。不再等待动画。提供了一个简单的 CLI ( InstantSpaceSwitcher.app/Contents/MacOS/ISSCli --help )", "source_period": "weekly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 12, "stars_in_period": 875 } }, "composite_score": 0.181329, "delta_since_last_run": 110, "previous_rank": 196 }, { "rank": 156, "repo_path": "maximhq/bifrost", "owner": "maximhq", "name": "bifrost", "repo_url": "https://github.com/maximhq/bifrost", "owner_url": "https://github.com/maximhq", "description": "Fastest enterprise AI gateway (50x faster than LiteLLM) with adaptive load balancer, cluster mode, guardrails, 1000+ models support & <100 µs overhead at 5k RPS.", "language": "Go", "stars_total": 3681, "forks_total": 413, "stars_in_period": 874, "period_label": "最近30天", "description_zh": "最快的企业 AI 网关（比 LiteLLM 快 50 倍），具有自适应负载均衡器、集群模式、护栏、1000 多个模型支持以及 5k RPS 时的开销 <100 µs。", "readme_summary_zh": "Bifrost 是一种高性能 AI 网关，可通过单个 OpenAI 兼容 API 统一对 15 个以上提供商（OpenAI、Anthropic、AWS Bedrock、Google Vertex 等）的访问。通过零配置在几秒钟内完成部署，并获得自动故障转移、负载平衡、语义缓存和企业级功能。", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 9, "stars_in_period": 31 }, "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 13, "stars_in_period": 195 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 5, "stars_in_period": 874 } }, "composite_score": 0.441774, "delta_since_last_run": 22, "previous_rank": 158 }, { "rank": 157, "repo_path": "rivet-dev/agent-os", "owner": "rivet-dev", "name": "agent-os", "repo_url": "https://github.com/rivet-dev/agent-os", "owner_url": "https://github.com/rivet-dev", "description": "A portable open-source operating system for agents. ~6 ms coldstarts, 32x cheaper than sandboxes. Powered by WebAssembly and V8 isolates.", "language": "Rust", "stars_total": 2632, "forks_total": 106, "stars_in_period": 867, "period_label": "最近7天", "description_zh": "面向代理的便携式开源操作系统。冷启动约 6 毫秒，比沙箱便宜 32 倍。由 WebAssembly 和 V8 隔离提供支持。", "readme_summary_zh": "用于人工智能代理的便携式开源操作系统。接近零的冷启动（约 6 毫秒），比沙箱便宜 32 倍。由 WebAssembly 和 V8 隔离提供支持。支持 Pi、Claude Code*、Codex*、Amp* 和 OpenCode* * 即将推出 agentOS 是一个在您的进程内运行的轻量级虚拟机。沙箱是完整的 Linux 环境。 agentOS 通过主机工具和精细权限将代理集成到您的后端。沙盒为您提供了适用于浏览器、本机二进制文件和开发服务器的完整操作系统。", "source_period": "weekly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 14, "stars_in_period": 867 } }, "composite_score": 0.180199, "delta_since_last_run": 6, "previous_rank": 160 }, { "rank": 158, "repo_path": "jordanbaird/Ice", "owner": "jordanbaird", "name": "Ice", "repo_url": "https://github.com/jordanbaird/Ice", "owner_url": "https://github.com/jordanbaird", "description": "Powerful menu bar manager for macOS", "language": "Swift", "stars_total": 27173, "forks_total": 658, "stars_in_period": 866, "period_label": "最近30天", "description_zh": "适用于 macOS 的强大菜单栏管理器", "readme_summary_zh": "Ice 是一款面向 macOS 的菜单栏管理工具，核心用途是帮助用户更灵活地隐藏、显示和整理菜单栏图标，从而让顶部菜单栏更加整洁、高效。除了最基础的图标收纳功能外，它的目标是逐步扩展为功能全面的菜单栏增强方案，覆盖更多与菜单栏管理、可见性控制和使用体验优化相关的能力，力求成为同类工具中更通用、更多才的选择。该项目目前仍处于积极开发阶段，部分规划中的功能尚未完成，因此尤其适合希望改善菜单栏杂乱问题、愿意尝试持续演进工具的 macOS 用户，以及对菜单栏效率和桌面整洁度有较高要求的日常办公与重度使用场景。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 8, "stars_in_period": 866 } }, "composite_score": 0.067113, "delta_since_last_run": 15, "previous_rank": 159 }, { "rank": 159, "repo_path": "ente-io/ente", "owner": "ente-io", "name": "ente", "repo_url": "https://github.com/ente-io/ente", "owner_url": "https://github.com/ente-io", "description": "💚 End-to-end encrypted cloud for everything.", "language": "Dart", "stars_total": 25867, "forks_total": 1529, "stars_in_period": 863, "period_label": "最近30天", "description_zh": "💚 适用于一切的端到端加密云。", "readme_summary_zh": "Ente 是一个完全开源、端到端加密的云端数据平台，目标是在不必信任服务提供商的前提下，让用户安全保存照片、身份验证信息以及重要文档。这个 monorepo 汇集了其完整源码，包括面向 iOS、Android、Web、Linux、macOS、Windows 等多平台的客户端，以及支撑服务运行的服务端。基于同一加密基础设施，Ente 目前提供三款核心应用：Ente Photos，用于替代 Apple Photos 和 Google Photos，支持三重数据复制、人脸检测、语义搜索、私密分享、协作相册、家庭方案、后台上传以及便捷导入导出；Ente Auth，是面向双重身份验证场景的开源 2FA 应用；Ente Locker，则用于加密保存证件、凭据、笔记等敏感资料，并支持私密共享与紧急访问设置。项目代码与密码学实现还经过多家外部安全机构审计，因此特别适合重视隐私、希望跨平台统一管理敏感数据，或有自托管需求的个人与家庭用户。", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 21, "stars_in_period": 23 }, "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 17, "stars_in_period": 162 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 13, "stars_in_period": 863 } }, "composite_score": 0.430749, "delta_since_last_run": -1, "previous_rank": 155 }, { "rank": 160, "repo_path": "chatwoot/chatwoot", "owner": "chatwoot", "name": "chatwoot", "repo_url": "https://github.com/chatwoot/chatwoot", "owner_url": "https://github.com/chatwoot", "description": "Open-source live-chat, email support, omni-channel desk. An alternative to Intercom, Zendesk, Salesforce Service Cloud etc. 🔥💬", "language": "Ruby", "stars_total": 28441, "forks_total": 6842, "stars_in_period": 859, "period_label": "最近30天", "description_zh": "开源实时聊天、电子邮件支持、全渠道服务台。 Intercom、Zendesk、Salesforce Service Cloud 等的替代品。🔥💬", "readme_summary_zh": "Chatwoot 是一款现代化、开源且可自托管的客户支持平台，可作为 Intercom、Zendesk 等产品的替代方案，适合希望掌控客户数据、同时统一管理多渠道沟通的企业使用。它将网站在线聊天、电子邮件以及 Facebook、Instagram、Twitter、WhatsApp、Telegram、Line、短信等渠道的会话集中到一个收件箱中，帮助客服团队在同一界面处理咨询。平台内置帮助中心，可发布知识库、FAQ 和指南，鼓励用户自助解决常见问题；同时提供私有备注、@提及、标签、快捷回复、自动分配、多语言支持、自定义视图、工作时间与自动回复、团队协作、联系人管理、客户分群、活动触达和预聊天表单等能力。其 AI 助手 Captain 还能自动回复常见问题、减轻坐席负担，因此尤其适合需要规模化客服运营、跨渠道服务和流程自动化的 SaaS、互联网公司、电商及成长型企业。", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 1, "stars_in_period": 96 }, "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 1, "stars_in_period": 253 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 1, "stars_in_period": 859 } }, "composite_score": 0.456051, "delta_since_last_run": 19, "previous_rank": 163 }, { "rank": 161, "repo_path": "MatsuriDayo/NekoBoxForAndroid", "owner": "MatsuriDayo", "name": "NekoBoxForAndroid", "repo_url": "https://github.com/MatsuriDayo/NekoBoxForAndroid", "owner_url": "https://github.com/MatsuriDayo", "description": "NekoBox for Android / sing-box / universal proxy toolchain for Android", "language": "Kotlin", "stars_total": 19916, "forks_total": 1558, "stars_in_period": 849, "period_label": "最近30天", "description_zh": "NekoBox for Android / single-box / Android 通用代理工具链", "readme_summary_zh": "NekoBox for Android 是一款基于 sing-box 的 Android 通用代理工具，面向需要在手机上灵活管理和使用网络代理的用户。它的核心作用是将 sing-box 提供的通用代理能力整合到安卓设备中，帮助用户在移动端进行代理连接、配置与使用，适合需要多场景网络访问、测试代理配置或日常使用通用代理方案的人群。项目强调开源版本的获取渠道应以 GitHub Releases 为准，并特别提醒用户不要下载自 2024 年 5 月起已被第三方控制的 Google Play 版本，因为该版本已不再是开源版本。", "source_period": "monthly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 4, "stars_in_period": 257 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 8, "stars_in_period": 849 } }, "composite_score": 0.247609, "delta_since_last_run": 14, "previous_rank": 162 }, { "rank": 162, "repo_path": "microsoft/agent-framework", "owner": "microsoft", "name": "agent-framework", "repo_url": "https://github.com/microsoft/agent-framework", "owner_url": "https://github.com/microsoft", "description": "A framework for building, orchestrating and deploying AI agents and multi-agent workflows with support for Python and .NET.", "language": "Python", "stars_total": 9294, "forks_total": 1517, "stars_in_period": 845, "period_label": "最近7天", "description_zh": "用于构建、编排和部署 AI 代理和多代理工作流程的框架，支持 Python 和 .NET。", "readme_summary_zh": "欢迎使用 Microsoft 的综合多语言框架，用于构建、编排和部署 AI 代理，并支持 .NET 和 Python 实现。该框架提供了从简单的聊天代理到具有基于图形编排的复杂多代理工作流程的一切。观看完整的代理框架介绍（30 分钟）", "source_period": "weekly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 12, "stars_in_period": 845 } }, "composite_score": 0.181083, "delta_since_last_run": 22, "previous_rank": 161 }, { "rank": 163, "repo_path": "KaringX/karing", "owner": "KaringX", "name": "karing", "repo_url": "https://github.com/KaringX/karing", "owner_url": "https://github.com/KaringX", "description": "Simple & Powerful proxy utility, Support routing rules for clash/sing-box", "language": "Dart", "stars_total": 10804, "forks_total": 930, "stars_in_period": 826, "period_label": "最近30天", "description_zh": "简单而强大的代理实用程序，支持冲突/单盒的路由规则", "readme_summary_zh": "Karing 是一款面向代理与网络分流场景的跨平台配置管理工具，主要用于统一接入和管理 Clash、V2ray/V2fly、Sing-box、Shadowsocks 以及订阅和 GitHub 订阅等多种来源，支持完整 Clash 配置与部分 clash.meta 配置。它的核心能力在于将多订阅源与一套路由规则结合，自动挑选更高效的节点，并提供自定义路由规则组、节点组以及内置 geo-IP、geo-site、ACL 等规则集，同时也为新手准备了开箱即用的默认分组和更简化的初学者模式。项目还内置修改版 sing-box 内核，并提供配置备份与多设备同步能力，可通过 iCloud、局域网、WebDAV 以及 ZIP 导入导出实现统一配置管理，因此特别适合需要在手机、电脑等多设备间同步代理配置、希望降低复杂配置门槛，或希望灵活整合多种订阅与分流策略的用户使用。", "source_period": "monthly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 5, "stars_in_period": 212 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 8, "stars_in_period": 826 } }, "composite_score": 0.246683, "delta_since_last_run": 18, "previous_rank": 164 }, { "rank": 164, "repo_path": "dotnet/skills", "owner": "dotnet", "name": "skills", "repo_url": "https://github.com/dotnet/skills", "owner_url": "https://github.com/dotnet", "description": "Repository for skills to assist AI coding agents with .NET and C#", "language": "C#", "stars_total": 1022, "forks_total": 77, "stars_in_period": 822, "period_label": "最近30天", "description_zh": "使用 .NET 和 C# 协助 AI 编码代理的技能存储库", "readme_summary_zh": "该存储库包含 .NET 团队精选的一组核心技能和用于编码代理的自定义代理。有关代理技能标准的信息，请参阅agentskills.io。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 2, "stars_in_period": 822 } }, "composite_score": 0.068156, "delta_since_last_run": 14, "previous_rank": 142 }, { "rank": 165, "repo_path": "IvanMurzak/Unity-MCP", "owner": "IvanMurzak", "name": "Unity-MCP", "repo_url": "https://github.com/IvanMurzak/Unity-MCP", "owner_url": "https://github.com/IvanMurzak", "description": "AI Skills, MCP Tools, and CLI for Unity Engine. Full AI develop and test loop. Use cli for quick setup. Efficient token usage, advanced tools. Any C# method may be turned into a tool by a single line. Works with Claude Code, Gemini, Copilot, Cursor and any other absolutely for free.", "language": "C#", "stars_total": 2044, "forks_total": 188, "stars_in_period": 810, "period_label": "最近30天", "description_zh": "适用于 Unity 引擎的 AI 技能、MCP 工具和 CLI。完整的人工智能开发和测试循环。使用 cli 进行快速设置。高效的代币使用，先进的工具。任何 C# 方法都可以通过一行代码变成一个工具。与 Claude Code、Gemini、Copilot、Cursor 和任何其他完全免费的工具一起使用。", "readme_summary_zh": "Unity MCP 是一个面向 Unity 编辑器与运行时的 AI 游戏开发助手，通过 MCP 将 Claude、Cursor、Windsurf 等智能体接入 Unity，帮助开发者自动化工作流、生成代码、编写与执行测试，并以自然对话方式完成项目协作。它的核心价值在于既能在编辑器内调用大量工具操作 Unity、结合项目环境自动生成适配的 AI 技能、辅助排查和修复问题，也能直接运行在已编译的游戏中，让大语言模型参与实时调试、玩家与 AI 交互以及动态 NPC 行为控制。项目还支持本地或远程部署，并允许开发者扩展自定义工具，因此特别适合需要提升开发效率、构建 AI 驱动玩法、进行运行时调试，或希望在同一套体系中整合多家模型服务而避免厂商绑定的 Unity 游戏开发场景。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 5, "stars_in_period": 810 } }, "composite_score": 0.0676, "delta_since_last_run": 4, "previous_rank": 165 }, { "rank": 166, "repo_path": "cat3399/blbl", "owner": "cat3399", "name": "blbl", "repo_url": "https://github.com/cat3399/blbl", "owner_url": "https://github.com/cat3399", "description": "一个使用原生Android开发的哔哩哔哩安卓客户端，支持触摸、遥控，以及安卓5", "language": "Kotlin", "stars_total": 1282, "forks_total": 86, "stars_in_period": 803, "period_label": "最近30天", "description_zh": "一个使用原生Android开发的哔哩哔哩安卓客户端，支持触摸、遥控，以及安卓5", "readme_summary_zh": "blbl 是一款面向安卓设备的第三方哔哩哔哩客户端，兼顾触摸操作与遥控器交互，并支持 Android 5，特别适合平板、电视和车机等非手机场景使用。项目围绕大屏与多终端体验构建，提供推荐、分类、动态、直播、我的、搜索、追番等常用页面，并通过侧边栏导航提升浏览效率，同时包含扫码登录入口，方便账号接入。在播放能力上，它基于 Media3（ExoPlayer）实现视频播放，支持分辨率、编码、倍速、字幕、弹幕等多种设置，也提供播放与弹幕相关偏好配置。整体采用 Kotlin、AndroidX、ViewBinding 等技术栈开发，适合希望在大屏设备上获得更完整、更易操作的哔哩哔哩内容浏览与观看体验的用户与开发者参考。", "source_period": "monthly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 6, "stars_in_period": 247 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 12, "stars_in_period": 803 } }, "composite_score": 0.245689, "delta_since_last_run": 6, "previous_rank": 167 }, { "rank": 167, "repo_path": "mihonapp/mihon", "owner": "mihonapp", "name": "mihon", "repo_url": "https://github.com/mihonapp/mihon", "owner_url": "https://github.com/mihonapp", "description": "Free and open source manga reader for Android", "language": "Kotlin", "stars_total": 19776, "forks_total": 1071, "stars_in_period": 800, "period_label": "最近30天", "description_zh": "适用于 Android 的免费开源漫画阅读器", "readme_summary_zh": "Mihon 是一款面向 Android 8.0 及以上设备的漫画阅读应用，主要用于更轻松地发现和阅读漫画、条漫、欧美漫画等各类图像内容。它支持本地内容导入与离线阅读，内置可高度自定义的阅读器，能够根据不同作品和个人习惯调整阅读方向、显示方式及多种阅读设置，同时提供明暗主题切换，提升长时间阅读体验。项目还支持将书库按分类整理，并可定时检查新章节更新，方便持续追更；此外还能与 MyAnimeList、AniList、Kitsu、MangaUpdates、Shikimori 和 Bangumi 等追踪服务同步阅读进度，并支持本地或云端备份。它尤其适合希望集中管理个人漫画书库、重视阅读体验定制、需要离线使用或跨服务记录阅读状态的 Android 用户。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 19, "stars_in_period": 800 } }, "composite_score": 0.065049, "delta_since_last_run": 8, "previous_rank": 166 }, { "rank": 168, "repo_path": "JingMatrix/Vector", "owner": "JingMatrix", "name": "Vector", "repo_url": "https://github.com/JingMatrix/Vector", "owner_url": "https://github.com/JingMatrix", "description": "Modern Xposed Framework", "language": "Java", "stars_total": 10642, "forks_total": 677, "stars_in_period": 783, "period_label": "最近30天", "description_zh": "现代 Xpose 框架", "readme_summary_zh": "Vector 是一个面向现代 Android 的高性能 ART Hook 框架，以 Zygisk 模块形式运行，并尽量保持与原版 Xposed 一致的 API 体验，方便现有模块和开发者迁移使用。它构建在 LSPlant 之上，提供更稳定的原生级插桩能力，可在内存中动态修改系统与应用行为，而无需直接篡改 APK 文件，因此整体方式更非破坏性，出现问题时通常重启即可恢复，兼容性也更适合不同 ROM 与多个 Android 版本环境。项目支持 Android 8.1 至 Android 17 Beta 设备，依赖较新的 Magisk 或 KernelSU 并启用 Zygisk，特别适合需要进行系统功能扩展、应用行为定制、运行时调试与兼容性适配的高级安卓用户和模块开发者。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 13, "stars_in_period": 783 } }, "composite_score": 0.066121, "delta_since_last_run": 7, "previous_rank": 168 }, { "rank": 169, "repo_path": "PolymathicAI/the_well", "owner": "PolymathicAI", "name": "the_well", "repo_url": "https://github.com/PolymathicAI/the_well", "owner_url": "https://github.com/PolymathicAI", "description": "A 15TB Collection of Physics Simulation Datasets", "language": "Jupyter Notebook", "stars_total": 2809, "forks_total": 328, "stars_in_period": 775, "period_label": "最近30天", "description_zh": "15TB 物理模拟数据集", "readme_summary_zh": "欢迎来到 Well，这是一个大规模的机器学习数据集，其中包含各种时空物理系统的数值模拟。 The Well 吸引了领域科学家和数值软件开发人员的帮助，提供 16 个数据集的 15TB 数据，涵盖生物系统、流体动力学、声学散射以及河外流体或超新星爆炸的磁流体动力学模拟等不同领域。这些数据集可以单独使用，也可以作为更广泛的基准套件的一部分使用，以加速机器学习和计算科学的研究。安装 Well 包并下载数据后，您可以在训练管道中使用它们。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 6, "stars_in_period": 775 } }, "composite_score": 0.067383, "delta_since_last_run": 15, "previous_rank": 174 }, { "rank": 170, "repo_path": "dockur/macos", "owner": "dockur", "name": "macos", "repo_url": "https://github.com/dockur/macos", "owner_url": "https://github.com/dockur", "description": "MacOS inside a Docker container.", "language": "Shell", "stars_total": 19459, "forks_total": 963, "stars_in_period": 765, "period_label": "最近30天", "description_zh": "Docker 容器内的 MacOS。", "readme_summary_zh": "dockur/macos 是一个把 macOS 运行在 Docker 容器中的项目，目标是用尽量简单的方式快速启动可用的 macOS 虚拟环境。它结合 KVM 硬件加速提升运行性能，提供基于浏览器的图形界面查看器，方便用户直接通过网页访问和操作系统桌面，同时还能自动下载所需的 macOS 安装资源，减少手动准备镜像和复杂配置的步骤。这个项目特别适合希望在 Linux 服务器、实验环境或具备虚拟化支持的主机上快速部署 macOS 用于测试、兼容性验证、临时开发调试和远程访问的人群，也适用于想以容器化方式统一管理 macOS 运行实例的场景。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 14, "stars_in_period": 765 } }, "composite_score": 0.065921, "delta_since_last_run": 2, "previous_rank": 171 }, { "rank": 171, "repo_path": "coleam00/Archon", "owner": "coleam00", "name": "Archon", "repo_url": "https://github.com/coleam00/Archon", "owner_url": "https://github.com/coleam00", "description": "The first open-source harness builder for AI coding. Make AI coding deterministic and repeatable.", "language": "TypeScript", "stars_total": 15775, "forks_total": 2562, "stars_in_period": 756, "period_label": "最近24小时", "description_zh": "第一个用于 AI 编码的开源线束构建器。使人工智能编码具有确定性和可重复性。", "readme_summary_zh": "第一个用于 AI 编码的开源线束构建器。使人工智能编码具有确定性和可重复性。 Archon 是人工智能编码代理的工作流引擎。将您的开发流程定义为 YAML 工作流程 - 规划、实施、验证、代码审查、PR 创建 - 并在所有项目中可靠地运行它们。", "source_period": "daily", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 2, "stars_in_period": 756 } }, "composite_score": 0.226317, "delta_since_last_run": 661, "previous_rank": 172 }, { "rank": 172, "repo_path": "files-community/Files", "owner": "files-community", "name": "Files", "repo_url": "https://github.com/files-community/Files", "owner_url": "https://github.com/files-community", "description": "A modern file manager that helps users organize their files and folders.", "language": "C#", "stars_total": 42956, "forks_total": 2677, "stars_in_period": 756, "period_label": "最近30天", "description_zh": "一个现代文件管理器，可以帮助用户组织他们的文件和文件夹。", "readme_summary_zh": "Files 是一款面向 Windows 的现代化文件管理器，旨在帮助用户更高效地整理、浏览和管理文件与文件夹，并以开放协作的方式持续改进。这个项目由开源社区共同构建和维护，重视用户反馈与问题报告，以不断提升功能完善度和整体稳定性。它的核心能力包括强大的多任务处理体验、文件标签管理、与系统及相关功能的深度集成，以及直观易用的界面设计，使日常文件操作更加流畅清晰。对于希望替代或增强传统 Windows 资源管理体验的个人用户、重视效率的办公场景，以及需要更灵活组织大量文件内容的用户来说，这个项目尤其适合。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 21, "stars_in_period": 756 } }, "composite_score": 0.064642, "delta_since_last_run": 6, "previous_rank": 173 }, { "rank": 173, "repo_path": "tiann/KernelSU", "owner": "tiann", "name": "KernelSU", "repo_url": "https://github.com/tiann/KernelSU", "owner_url": "https://github.com/tiann", "description": "A Kernel based root solution for Android", "language": "Kotlin", "stars_total": 15963, "forks_total": 3334, "stars_in_period": 754, "period_label": "最近30天", "description_zh": "基于内核的 Android 根解决方案", "readme_summary_zh": "KernelSU 是一个面向 Android 设备的内核级 Root 方案，核心目标是在内核层提供 su 与 Root 权限管理，相比传统用户空间实现更贴近系统底层。它同时提供基于 metamodules 的模块体系，可用于进行可插拔的无系统分区修改，并通过 App Profile 机制对应用的 Root 能力进行细粒度约束，降低权限滥用风险。该项目官方重点支持 Android GKI 2.0 设备与 5.10 及以上内核，也兼容部分 4.14 以上旧内核，但通常需要手动编译内核；此外还适用于 WSA、ChromeOS 以及基于容器的 Android 环境。它目前支持 arm64-v8a 和 x86_64 架构，尤其适合需要高可控 Root、系统级定制、模块扩展和跨 Android 运行环境统一管理权限的开发者与高级用户使用。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 7, "stars_in_period": 754 } }, "composite_score": 0.06718, "delta_since_last_run": 4, "previous_rank": 169 }, { "rank": 174, "repo_path": "mithun50/openclaw-termux", "owner": "mithun50", "name": "openclaw-termux", "repo_url": "https://github.com/mithun50/openclaw-termux", "owner_url": "https://github.com/mithun50", "description": "Run OpenClaw AI Gateway on Android — standalone Flutter app with built-in terminal, web dashboard, and one-tap setup. Also available as a Termux CLI package.", "language": "Dart", "stars_total": 1293, "forks_total": 207, "stars_in_period": 749, "period_label": "最近30天", "description_zh": "在 Android 上运行 OpenClaw AI Gateway — 具有内置终端、Web 仪表板和一键设置的独立 Flutter 应用程序。也可作为 Termux CLI 包使用。", "readme_summary_zh": "在 Android 上运行 OpenClaw AI Gateway — 具有内置终端、Web 仪表板、可选开发工具和一键设置的独立 Flutter 应用程序。也可作为 Termux CLI 包使用。 OpenClaw 将 OpenClaw AI 网关引入 Android。它通过 proot 设置完整的 Ubuntu 环境，安装 Node.js 和 OpenClaw，并提供原生 Flutter UI 来管理一切 - 无需 root。", "source_period": "monthly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 12, "stars_in_period": 176 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 6, "stars_in_period": 749 } }, "composite_score": 0.242944, "delta_since_last_run": 7, "previous_rank": 170 }, { "rank": 175, "repo_path": "punitarani/fli", "owner": "punitarani", "name": "fli", "repo_url": "https://github.com/punitarani/fli", "owner_url": "https://github.com/punitarani", "description": "Google Flights MCP and Python Library", "language": "Python", "stars_total": 1739, "forks_total": 202, "stars_in_period": 745, "period_label": "最近7天", "description_zh": "Google Flights MCP 和 Python 库", "readme_summary_zh": "一个强大的 Python 库，通过优雅的 CLI 界面提供对 Google Flights 数据的编程访问。轻松搜索航班、找到最优惠的价格并筛选结果。 🚀 是什么让 fli 如此特别？与其他依赖网络抓取的航班搜索库不同，Fli 通过逆向工程直接与 Google Flights 的 API 交互。这意味着：", "source_period": "weekly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 10, "stars_in_period": 745 } }, "composite_score": 0.181327, "delta_since_last_run": 10, "previous_rank": 175 }, { "rank": 176, "repo_path": "jgraph/drawio-desktop", "owner": "jgraph", "name": "drawio-desktop", "repo_url": "https://github.com/jgraph/drawio-desktop", "owner_url": "https://github.com/jgraph", "description": "Official electron build of draw.io", "language": "JavaScript", "stars_total": 60355, "forks_total": 5649, "stars_in_period": 743, "period_label": "最近30天", "description_zh": "Draw.io 的官方电子版本", "readme_summary_zh": "drawio-desktop 是一款基于 Electron 的桌面制图应用，本质上是对 draw.io 核心编辑器的桌面封装，适合在本地环境中创建和编辑流程图、架构图、UML 图等各类图表。项目以安全隔离和离线使用为首要目标，除启动时检查更新外，应用默认不依赖互联网运行，所有 JavaScript 资源均内置，并通过严格的内容安全策略禁止加载远程脚本，也不会将图表数据或使用分析发送到外部。它采用 Apache 2.0 许可，未修改代码的前提下可免费用于任何用途。该项目尤其适合对数据隐私、内网部署、离线办公和受控运行环境有较高要求的个人与企业用户，也适合希望基于现有 draw.io 编辑能力进行本地开发、调试和桌面分发的开发者使用。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 16, "stars_in_period": 743 } }, "composite_score": 0.065536, "delta_since_last_run": 8, "previous_rank": 176 }, { "rank": 177, "repo_path": "YunaiV/ruoyi-vue-pro", "owner": "YunaiV", "name": "ruoyi-vue-pro", "repo_url": "https://github.com/YunaiV/ruoyi-vue-pro", "owner_url": "https://github.com/YunaiV", "description": "🔥 官方推荐 🔥 RuoYi-Vue 全新 Pro 版本，优化重构所有功能。基于 Spring Boot + MyBatis Plus + Vue & Element 实现的后台管理系统 + 微信小程序，支持 RBAC 动态权限、数据权限、SaaS 多租户、Flowable 工作流、三方登录、支付、短信、商城、CRM、ERP、AI 大模型、IoT 物联网等功能。你的 ⭐️ Star ⭐️，是作者生发的动力！", "language": "Java", "stars_total": 36382, "forks_total": 7854, "stars_in_period": 737, "period_label": "最近30天", "description_zh": "🔥 官方推荐 🔥 RuoYi-Vue 全新 Pro 版本，优化重构所有功能。基于 Spring Boot + MyBatis Plus + Vue & Element 实现的后台管理系统 + 微信小程序，支持 RBAC 动态权限、数据权限、SaaS 多租户、Flowable 工作流、三方登录、支付、短信、商城、CRM、ERP、AI 大模型、IoT 物联网等功能。你的 ⭐️ Star ⭐️，是作者生发的动力！", "readme_summary_zh": "RuoYi-Vue-Pro（芋道）是一个面向开发者的全开源 Java 快速开发平台，适合个人和企业用来搭建中后台管理系统及各类业务平台。项目以后端 Spring Boot 多模块架构为核心，支持 JDK 8 以及 JDK 17/21，并结合 MyBatis Plus、Redis、Redisson 等常用技术栈，数据库适配也较广，可用于 MySQL、PostgreSQL、Oracle、SQL Server、MariaDB、达梦、TiDB 等环境。前端提供多种管理后台方案，包括基于 Vue3 的 element-plus 和 vben，以及 Vue2 的 element-ui，同时还支持 uni-app 移动端，一套代码覆盖 APP、小程序和 H5。项目既有只含系统与基础设施能力的精简版，也有包含会员中心、数据报表、工作流、商城、公众号、CRM、ERP 等完整业务模块的完整版，因此特别适合需要快速落地企业级管理系统、业务中台或多端一体化应用的场景。", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 3, "stars_in_period": 52 }, "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 6, "stars_in_period": 223 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 6, "stars_in_period": 737 } }, "composite_score": 0.44954, "delta_since_last_run": 7, "previous_rank": 187 }, { "rank": 178, "repo_path": "flutter/flutter", "owner": "flutter", "name": "flutter", "repo_url": "https://github.com/flutter/flutter", "owner_url": "https://github.com/flutter", "description": "Flutter makes it easy and fast to build beautiful apps for mobile and beyond", "language": "Dart", "stars_total": 175923, "forks_total": 30243, "stars_in_period": 736, "period_label": "最近30天", "description_zh": "Flutter 让您可以轻松快速地为移动设备及其他领域构建精美的应用程序", "readme_summary_zh": "Flutter 是 Google 的 SDK，用于通过单个代码库为移动、Web 和桌面打造美观、快速的用户体验。 Flutter 可与现有代码配合使用，由世界各地的开发人员和组织使用，并且免费且开源。如需发布和其他公告，请加入 flutter-announce 邮件列表。我们的文档还跟踪各个版本之间的重大更改。", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 8, "stars_in_period": 32 }, "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 4, "stars_in_period": 163 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 19, "stars_in_period": 736 } }, "composite_score": 0.444231, "delta_since_last_run": 8, "previous_rank": 177 }, { "rank": 179, "repo_path": "mayocream/koharu", "owner": "mayocream", "name": "koharu", "repo_url": "https://github.com/mayocream/koharu", "owner_url": "https://github.com/mayocream", "description": "ML-powered manga translator, written in Rust.", "language": "Rust", "stars_total": 2443, "forks_total": 129, "stars_in_period": 726, "period_label": "最近7天", "description_zh": "由 ML 驱动的漫画翻译器，用 Rust 编写。", "readme_summary_zh": "由 ML 驱动的漫画翻译器，用 Rust 编写。 Koharu 引入了一种新的漫画翻译工作流程，利用机器学习的力量来实现流程自动化。它结合了对象检测、OCR、修复和法学硕士的功能，打造无缝翻译体验。", "source_period": "weekly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 3, "stars_in_period": 114 }, "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 5, "stars_in_period": 726 } }, "composite_score": 0.388861, "delta_since_last_run": 37, "previous_rank": 194 }, { "rank": 180, "repo_path": "jgraph/drawio", "owner": "jgraph", "name": "drawio", "repo_url": "https://github.com/jgraph/drawio", "owner_url": "https://github.com/jgraph", "description": "draw.io is a JavaScript, client-side editor for general diagramming.", "language": "JavaScript", "stars_total": 4715, "forks_total": 779, "stars_in_period": 726, "period_label": "最近30天", "description_zh": "draw.io 是一个用于一般图表绘制的 JavaScript 客户端编辑器。", "readme_summary_zh": "draw.io 是一款可配置的图表绘制与在线白板应用，由 draw.io Ltd 与 draw.io AG 共同开发和维护，同时也提供了成熟的线上生产部署服务。它主要用于帮助个人与团队高效创建各类可视化内容，如流程图、架构图、组织结构图、UML 图以及头脑风暴白板等，并可根据不同使用环境进行灵活配置。该项目的核心价值在于同时兼顾通用绘图、协作表达与可部署性，适合需要将图形化沟通融入产品设计、软件开发、业务分析、教学培训或远程协作场景的用户，用来整理思路、说明系统结构和沉淀团队知识。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 14, "stars_in_period": 726 } }, "composite_score": 0.065881, "delta_since_last_run": 4, "previous_rank": 180 }, { "rank": 181, "repo_path": "microsoft/terminal", "owner": "microsoft", "name": "terminal", "repo_url": "https://github.com/microsoft/terminal", "owner_url": "https://github.com/microsoft", "description": "The new Windows Terminal and the original Windows console host, all in the same place!", "language": "C++", "stars_total": 102635, "forks_total": 9200, "stars_in_period": 724, "period_label": "最近30天", "description_zh": "新的 Windows 终端和原来的 Windows 控制台主机都在同一个地方！", "readme_summary_zh": "microsoft/terminal 是微软开发的现代化 Windows 终端项目，用于为命令行用户提供比传统控制台更强大、更一致的使用体验。它的核心包括 Windows Terminal、经典的 Windows Console Host 以及两者共享的底层组件，其中 Windows Terminal 侧重于多标签页、多窗格、可配置外观与更现代的交互方式，能够更方便地承载 PowerShell、Cmd、WSL 等多种命令行环境；同时项目也说明了新旧终端架构的关系，方便开发者理解控制台生态并参与构建。这个项目最适合经常进行开发、运维、系统管理和跨环境命令行工作的 Windows 用户，也适合希望研究、编译和贡献 Windows 终端基础设施的开发者。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 19, "stars_in_period": 724 } }, "composite_score": 0.064972, "delta_since_last_run": 15, "previous_rank": 178 }, { "rank": 182, "repo_path": "nibzard/awesome-agentic-patterns", "owner": "nibzard", "name": "awesome-agentic-patterns", "repo_url": "https://github.com/nibzard/awesome-agentic-patterns", "owner_url": "https://github.com/nibzard", "description": "A curated catalogue of awesome agentic AI patterns", "language": "HTML", "stars_total": 4270, "forks_total": 352, "stars_in_period": 722, "period_label": "最近30天", "description_zh": "很棒的代理 AI 模式的精选目录", "readme_summary_zh": "代理人工智能模式的精选目录——现实世界的技巧、工作流程和微型架构，可帮助自主或半自主人工智能代理在生产中完成有用的工作。为什么？教程展示玩具演示。真正的产品隐藏着杂乱的部分。此列表列出了弥合差距的可重复模式，以便我们能够提供更智能、更快速的代理。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 8, "stars_in_period": 722 } }, "composite_score": 0.066966, "delta_since_last_run": 2, "previous_rank": 182 }, { "rank": 183, "repo_path": "PowerShell/PowerShell", "owner": "PowerShell", "name": "PowerShell", "repo_url": "https://github.com/PowerShell/PowerShell", "owner_url": "https://github.com/PowerShell", "description": "PowerShell for every system!", "language": "C#", "stars_total": 52364, "forks_total": 8249, "stars_in_period": 720, "period_label": "最近30天", "description_zh": "适用于每个系统的 PowerShell！", "readme_summary_zh": "PowerShell 是一个面向 Windows、Linux 和 macOS 的跨平台自动化与配置管理工具框架，适合把命令行操作、脚本编写和系统集成统一到同一套工作流中。它核心包含命令行 shell、配套脚本语言以及用于处理 cmdlet 的执行框架，尤其擅长处理 JSON、CSV、XML 等结构化数据，调用 REST API，并围绕对象模型进行操作，因此比传统以纯文本为中心的 shell 更适合现代运维与开发场景。该仓库主要对应 PowerShell 7.x 及以上版本，而非回传到 Windows PowerShell 5.1。这个项目最适合用于跨平台运维自动化、批量配置管理、云资源管理、数据处理、CI/CD 流程编排，以及需要将现有工具和服务高效串联起来的场景。", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 3, "stars_in_period": 15 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 7, "stars_in_period": 720 } }, "composite_score": 0.268979, "delta_since_last_run": 2, "previous_rank": 179 }, { "rank": 184, "repo_path": "psmux/psmux", "owner": "psmux", "name": "psmux", "repo_url": "https://github.com/psmux/psmux", "owner_url": "https://github.com/psmux", "description": "Tmux on Windows Powershell - tmux for PowerShell, Windows Terminal, cmd.exe. Includes psmux, pmux, and tmux commands. This is native Powershell Tmux designed for Windows in Rust 🦀", "language": "Rust", "stars_total": 1141, "forks_total": 79, "stars_in_period": 717, "period_label": "最近30天", "description_zh": "Windows Powershell 上的 Tmux - 用于 PowerShell、Windows 终端、cmd.exe 的 tmux。包括 psmux、pmux 和 tmux 命令。这是用 Rust 为 Windows 设计的原生 Powershell Tmux 🦀", "readme_summary_zh": "本机 Windows tmux。诞生于 PowerShell，用 Rust 制造。完整的鼠标支持 · tmux 主题 · tmux 配置 · 76 个命令 · 速度极快 安装 · 使用 · 克劳德代码 · 功能 · 兼容性 · 性能 · 插件 · 按键 · 脚本 · 配置 · 鼠标/SSH · 常见问题解答 · 相关项目", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 12, "stars_in_period": 717 } }, "composite_score": 0.066235, "delta_since_last_run": 13, "previous_rank": 183 }, { "rank": 185, "repo_path": "HKUDS/AI-Trader", "owner": "HKUDS", "name": "AI-Trader", "repo_url": "https://github.com/HKUDS/AI-Trader", "owner_url": "https://github.com/HKUDS", "description": "\"AI-Trader: 100% Fully-Automated Agent-Native Trading\"", "language": "Python", "stars_total": 12917, "forks_total": 2181, "stars_in_period": 716, "period_label": "最近7天", "description_zh": "“AI-Trader：100%全自动代理原生交易”", "readme_summary_zh": "就像人类有自己的交易平台一样，人工智能代理也需要自己的。 AI-Trader 是一个代理原生交易平台：通过人工智能代理交流想法并提高交易技能！", "source_period": "weekly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 16, "stars_in_period": 716 } }, "composite_score": 0.177893, "delta_since_last_run": 113, "previous_rank": 188 }, { "rank": 186, "repo_path": "bannedbook/fanqiang", "owner": "bannedbook", "name": "fanqiang", "repo_url": "https://github.com/bannedbook/fanqiang", "owner_url": "https://github.com/bannedbook", "description": "翻墙-科学上网", "language": "Kotlin", "stars_total": 43120, "forks_total": 7702, "stars_in_period": 716, "period_label": "最近30天", "description_zh": "翻墙-科学上网", "readme_summary_zh": "这是一个面向中文用户的网络访问与代理工具资料整合项目，主要用于汇总各平台“翻墙”相关的软件、教程和使用指引，帮助用户在安卓、Windows、macOS、iPhone 以及 Linux 等环境中快速找到合适的科学上网方案。项目核心内容包括安卓端 FQNews 等应用说明，Chrome、Edge、Firefox、Tor Browser 等浏览器的一键配置包，自建 V2Ray 和 Shadowsocks 服务器的简明教程，以及 Clash for Windows、V2rayN、SSTap、SSR 等常见客户端的入门指导，同时还整理了部分免费账号信息和中国大陆注册 ChatGPT 的相关教程。它最适合希望低门槛获取代理工具入口、快速完成跨平台配置，或需要按设备与场景选择不同方案的普通用户与新手参考使用。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 4, "stars_in_period": 716 } }, "composite_score": 0.067685, "delta_since_last_run": 5, "previous_rank": 184 }, { "rank": 187, "repo_path": "wilsonfreitas/awesome-quant", "owner": "wilsonfreitas", "name": "awesome-quant", "repo_url": "https://github.com/wilsonfreitas/awesome-quant", "owner_url": "https://github.com/wilsonfreitas", "description": "A curated list of insanely awesome libraries, packages and resources for Quants (Quantitative Finance)", "language": "HTML", "stars_total": 25426, "forks_total": 3369, "stars_in_period": 712, "period_label": "最近30天", "description_zh": "为 Quants（定量金融）提供的超棒库、软件包和资源的精选列表", "readme_summary_zh": "这是一个面向量化金融从业者与研究者的精选资源汇总项目，主要用于帮助用户快速发现和筛选量化分析、建模、交易与研究中常用的开源库、工具包和学习资料。其核心能力在于按主题系统整理量化领域的重要资源，覆盖数值计算与数据结构、金融产品定价、技术指标、交易与回测、组合优化与风险分析、因子研究、情绪分析与另类数据、时间序列分析、市场数据源、可视化、Excel 集成、研究环境、跨语言框架以及书籍与训练材料等内容，并收录了如 NumPy 这类科学计算基础工具。它最适合用于搭建量化研究技术栈、为策略开发选型、补充学习路径，或作为团队梳理量化生态与资源地图的参考依据。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 11, "stars_in_period": 712 } }, "composite_score": 0.066411, "delta_since_last_run": 5, "previous_rank": 181 }, { "rank": 188, "repo_path": "axios/axios", "owner": "axios", "name": "axios", "repo_url": "https://github.com/axios/axios", "owner_url": "https://github.com/axios", "description": "Promise based HTTP client for the browser and node.js", "language": "JavaScript", "stars_total": 108982, "forks_total": 11613, "stars_in_period": 698, "period_label": "最近30天", "description_zh": "适用于浏览器和 Node.js 的基于 Promise 的 HTTP 客户端", "readme_summary_zh": "我们热衷于使开源可持续发展。扫描您的依赖关系树，以更好地了解哪些开源项目需要资助... Hopper 提供了一个安全的开源注册表，其中每个组件都经过恶意软件验证，并持续修复任何版本的漏洞。简而言之，Hopper 完全消除了管理软件供应链风险的需要。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 15, "stars_in_period": 698 } }, "composite_score": 0.065671, "delta_since_last_run": -31, "previous_rank": 189 }, { "rank": 189, "repo_path": "smtg-ai/claude-squad", "owner": "smtg-ai", "name": "claude-squad", "repo_url": "https://github.com/smtg-ai/claude-squad", "owner_url": "https://github.com/smtg-ai", "description": "Manage multiple AI terminal agents like Claude Code, Codex, OpenCode, and Amp.", "language": "Go", "stars_total": 6941, "forks_total": 491, "stars_in_period": 693, "period_label": "最近30天", "description_zh": "管理多个 AI 终端代理，例如 Claude Code、Codex、OpenCode 和 Amp。", "readme_summary_zh": "Claude Squad 是一款面向终端的多代理任务管理工具，用来同时调度多个代码助手，例如 Claude Code、Codex、Gemini，以及 Aider 等本地代理，并将它们分别放在独立工作区中并行处理不同任务。它的核心能力是让多个开发任务在后台持续执行，支持自动接受这类高自动化工作流，同时在同一个终端窗口中统一管理各个实例与任务进度。每个任务都会拥有隔离的 Git 工作区，从而避免分支和文件冲突；在真正合并或提交前，用户还能先审查改动、检查结果，再决定是否应用或推送。这个项目尤其适合需要并行推进多项开发工作、频繁试验代码修改，或希望在保持终端工作习惯的同时安全管理 AI 编码代理输出的开发者与技术团队。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 10, "stars_in_period": 693 } }, "composite_score": 0.066573, "delta_since_last_run": 6, "previous_rank": 190 }, { "rank": 190, "repo_path": "jqlang/jq", "owner": "jqlang", "name": "jq", "repo_url": "https://github.com/jqlang/jq", "owner_url": "https://github.com/jqlang", "description": "Command-line JSON processor", "language": "C", "stars_total": 34329, "forks_total": 1760, "stars_in_period": 682, "period_label": "最近30天", "description_zh": "命令行 JSON 处理器", "readme_summary_zh": "jq 是一个轻量、灵活的命令行 JSON 处理工具，可以把它看作面向 JSON 数据的 sed、awk 或 grep。它使用可移植的 C 语言编写，运行时没有额外依赖，因此很适合在各种环境中快速部署和使用。这个项目的核心能力是对结构化数据进行切片、筛选、映射和转换，帮助用户从复杂的 JSON 中提取所需字段、重组数据结构，或对结果进行进一步加工。它尤其适合命令行数据处理、日志分析、接口返回结果检查、自动化脚本以及需要高效处理 JSON 数据流的开发与运维场景，是日常处理 JSON 时非常实用的基础工具。", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 11, "stars_in_period": 75 }, "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 3, "stars_in_period": 268 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 9, "stars_in_period": 682 } }, "composite_score": 0.447047, "delta_since_last_run": 136, "previous_rank": 199 }, { "rank": 191, "repo_path": "dandavison/delta", "owner": "dandavison", "name": "delta", "repo_url": "https://github.com/dandavison/delta", "owner_url": "https://github.com/dandavison", "description": "A syntax-highlighting pager for git, diff, grep, rg --json, and blame output", "language": "Rust", "stars_total": 30053, "forks_total": 516, "stars_in_period": 679, "period_label": "最近30天", "description_zh": "用于 git、diff、grep、rg --json 和 Blame 输出的语法高亮分页器", "readme_summary_zh": "安装它（在大多数包管理器中该包被称为“git-delta”，但可执行文件只是 delta ）并将其添加到您的 ~/.gitconfig ：Delta 有很多功能并且非常可定制；请参阅 delta -h （简短帮助）或 delta --help （完整手册），或在线用户手册。", "source_period": "monthly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 17, "stars_in_period": 174 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 16, "stars_in_period": 679 } }, "composite_score": 0.238379, "delta_since_last_run": 6, "previous_rank": 191 }, { "rank": 192, "repo_path": "woheller69/FreeDroidWarn", "owner": "woheller69", "name": "FreeDroidWarn", "repo_url": "https://github.com/woheller69/FreeDroidWarn", "owner_url": "https://github.com/woheller69", "description": "暂无简介", "language": "Java", "stars_total": 1440, "forks_total": 55, "stars_in_period": 667, "period_label": "最近30天", "description_zh": "暂无简介", "readme_summary_zh": "FreeDroidWarn 是一个面向 Android 应用开发者的轻量库，用于在应用中弹出弃用警告对话框，提前告知用户：由于 Google 计划从 2026/2027 年起在已认证 Android 设备上强制要求开发者完成身份验证，且项目作者不接受向 Google 提交个人身份资料，因此相关应用届时可能无法在多数带有 Google Play 服务的认证设备上继续运行。它的核心能力不是绕过限制，而是以统一、直接的方式向用户解释这一政策对开发者隐私、设备自主权、自由开源软件生态、开发者安全以及开源分叉与本地化改造带来的影响，并提示用户了解替代方案，如使用不预装 Google Play 服务的系统、移除 Google 服务或通过 ADB 安装应用。这个项目最适合重视隐私与数字权利、希望提前进行用户沟通的独立开发者、开源项目维护者及相关倡议型应用使用。", "source_period": "monthly", "signal_details": { "daily": { "period": "daily", "period_label": "最近24小时", "rank": 17, "stars_in_period": 28 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 5, "stars_in_period": 667 } }, "composite_score": 0.261653, "delta_since_last_run": 21, "previous_rank": 185 }, { "rank": 193, "repo_path": "vernesong/OpenClash", "owner": "vernesong", "name": "OpenClash", "repo_url": "https://github.com/vernesong/OpenClash", "owner_url": "https://github.com/vernesong", "description": "A Clash Client For OpenWrt", "language": "HTML", "stars_total": 25205, "forks_total": 3847, "stars_in_period": 667, "period_label": "最近30天", "description_zh": "OpenWrt 的 Clash 客户端", "readme_summary_zh": "本插件是一个可运行在 OpenWrt 上的 Mihomo(Clash) 客户端 兼容 Shadowsocks、ShadowsocksR、Vmess、Trojan、Snell 等协议，根据灵活的规则配置实现策略代理", "source_period": "monthly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 6, "stars_in_period": 158 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 13, "stars_in_period": 667 } }, "composite_score": 0.244638, "delta_since_last_run": 3, "previous_rank": 192 }, { "rank": 194, "repo_path": "electron/electron", "owner": "electron", "name": "electron", "repo_url": "https://github.com/electron/electron", "owner_url": "https://github.com/electron", "description": "Build cross-platform desktop apps with JavaScript, HTML, and CSS", "language": "C++", "stars_total": 120834, "forks_total": 17121, "stars_in_period": 661, "period_label": "最近30天", "description_zh": "使用 JavaScript、HTML 和 CSS 构建跨平台桌面应用程序", "readme_summary_zh": "Electron 是一个用于构建跨平台桌面应用的框架，开发者可以使用熟悉的 JavaScript、HTML 和 CSS 来开发同时运行在 Windows、macOS 和 Linux 上的应用。它以 Chromium 提供界面渲染能力，并结合 Node.js 提供系统与本地能力访问，因此既适合制作拥有现代 Web 界面的桌面软件，也能满足需要文件系统、进程管理等原生功能的应用需求。像 Visual Studio Code 这样的知名产品也基于它构建，说明它非常适合需要统一技术栈、提升前后端协作效率、快速将 Web 开发经验迁移到桌面端的团队和项目。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 14, "stars_in_period": 661 } }, "composite_score": 0.065815, "delta_since_last_run": 5, "previous_rank": 193 }, { "rank": 195, "repo_path": "HapeLee/legado-with-MD3", "owner": "HapeLee", "name": "legado-with-MD3", "repo_url": "https://github.com/HapeLee/legado-with-MD3", "owner_url": "https://github.com/HapeLee", "description": "使用 Material Design 3 全新设计的阅读 3.0", "language": "Kotlin", "stars_total": 3019, "forks_total": 150, "stars_in_period": 652, "period_label": "最近30天", "description_zh": "使用 Material Design 3 全新设计的阅读 3.0", "readme_summary_zh": "Legado with MD3 是在开源阅读应用“阅读（Legado）”基础上重构的 Material Design 3 风格版本，面向希望获得更现代界面与一致交互体验的电子书阅读用户。项目不仅重绘了整体 UI，还逐步从传统 View 迁移到 Jetpack Compose，并基于 Monet 主题系统带来更统一的视觉效果，同时加入预测性返回、共享元素动画等新式交互。相比原版，它强化了阅读场景的个性化配置，支持更细致的阅读记录统计、漫画与有声书体验优化、更丰富的书架布局以及平板适配，还提供书籍备注、已读未读智能分组、手柄翻页、本地 TXT 和 EPUB 阅读、文件扫描、广告净化、正文替换及多种翻页模式。这个项目尤其适合重视界面设计、阅读数据管理、本地书籍导入与高度自定义体验的安卓阅读爱好者使用。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 20, "stars_in_period": 652 } }, "composite_score": 0.064717, "delta_since_last_run": 10, "previous_rank": 200 }, { "rank": 196, "repo_path": "automazeio/vibeproxy", "owner": "automazeio", "name": "vibeproxy", "repo_url": "https://github.com/automazeio/vibeproxy", "owner_url": "https://github.com/automazeio", "description": "Native macOS menu bar app to use your Claude Code & ChatGPT subscriptions with AI coding tools - no API keys needed", "language": "Swift", "stars_total": 2320, "forks_total": 157, "stars_in_period": 650, "period_label": "最近30天", "description_zh": "本机 macOS 菜单栏应用程序可将您的 Claude Code 和 ChatGPT 订阅与 AI 编码工具结合使用 - 无需 API 密钥", "readme_summary_zh": "停止为人工智能支付两次费用。 VibeProxy 是一款漂亮的原生 macOS 菜单栏应用程序，可让您使用现有的 Claude Code、ChatGPT、Gemini、Qwen、Antigravity 和 Z.AI GLM 订阅以及 Factory Droids 等强大的 AI 编码工具 - 无需单独的 API 密钥。它基于 CLIProxyAPIPlus 构建，自动处理 OAuth 身份验证、令牌管理和 API 路由。一键验证，零摩擦编码。", "source_period": "monthly", "signal_details": { "weekly": { "period": "weekly", "period_label": "最近7天", "rank": 6, "stars_in_period": 210 }, "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 12, "stars_in_period": 650 } }, "composite_score": 0.245229, "delta_since_last_run": null, "previous_rank": null }, { "rank": 197, "repo_path": "google-ai-edge/LiteRT", "owner": "google-ai-edge", "name": "LiteRT", "repo_url": "https://github.com/google-ai-edge/LiteRT", "owner_url": "https://github.com/google-ai-edge", "description": "LiteRT, successor to TensorFlow Lite. is Google's On-device framework for high-performance ML & GenAI deployment on edge platforms, via efficient conversion, runtime, and optimization", "language": "C++", "stars_total": 2163, "forks_total": 270, "stars_in_period": 636, "period_label": "最近30天", "description_zh": "LiteRT是TensorFlow Lite的继任者。是谷歌的On-device框架，通过高效的转换、运行时间和优化，在边缘平台上进行高性能ML和GenAI部署", "readme_summary_zh": "LiteRT 是谷歌面向边缘设备推出的端侧机器学习与生成式 AI 部署框架，目标是在手机、嵌入式设备及各类受限硬件上，以更低延迟和更高效率运行模型。它围绕“模型转换、运行时执行与性能优化”提供完整能力，可将训练好的模型高效转换为适合端侧部署的格式，并借助轻量级运行时在不同边缘平台上稳定推理，同时结合硬件加速与优化机制提升速度、降低资源占用和功耗。该项目尤其适合需要离线推理、实时响应、隐私保护或网络条件有限的应用场景，例如移动端智能功能、IoT 设备上的感知任务，以及在本地运行小型生成式 AI 模型的产品。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 5, "stars_in_period": 636 } }, "composite_score": 0.067422, "delta_since_last_run": null, "previous_rank": null }, { "rank": 198, "repo_path": "cockpit-project/cockpit", "owner": "cockpit-project", "name": "cockpit", "repo_url": "https://github.com/cockpit-project/cockpit", "owner_url": "https://github.com/cockpit-project", "description": "Cockpit is a web-based graphical interface for servers.", "language": "JavaScript", "stars_total": 13865, "forks_total": 1265, "stars_in_period": 635, "period_label": "最近30天", "description_zh": "Cockpit 是一个基于 Web 的服务器图形界面。", "readme_summary_zh": "Web 浏览器中的系统管理员登录会话 cockpit-project.org", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 17, "stars_in_period": 635 } }, "composite_score": 0.065244, "delta_since_last_run": 0, "previous_rank": 195 }, { "rank": 199, "repo_path": "Snailclimb/interview-guide", "owner": "Snailclimb", "name": "interview-guide", "repo_url": "https://github.com/Snailclimb/interview-guide", "owner_url": "https://github.com/Snailclimb", "description": "基于 Spring Boot 4.0 + Java 21 + Spring AI + PostgreSQL + pgvector + RustFS + Redis，实现简历智能分析、AI模拟面试、知识库RAG检索等核心功能。非常适合作为学习和简历项目，学习门槛低。", "language": "Java", "stars_total": 1522, "forks_total": 332, "stars_in_period": 630, "period_label": "最近30天", "description_zh": "基于 Spring Boot 4.0 + Java 21 + Spring AI + PostgreSQL + pgvector + RustFS + Redis，实现简历智能分析、AI模拟面试、知识库RAG检索等核心功能。非常适合作为学习和简历项目，学习门槛低。", "readme_summary_zh": "智能 AI 面试官平台 - 基于大语言模型的简历分析和模拟面试系统 InterviewGuide 是一个集成了简历分析、模拟面试和知识库管理的智能面试辅助平台。系统利用大语言模型（LLM）和向量数据库技术，为求职者和 HR 提供智能化的简历评估和面试练习服务。", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 18, "stars_in_period": 630 } }, "composite_score": 0.065057, "delta_since_last_run": 4, "previous_rank": 197 }, { "rank": 200, "repo_path": "STranslate/STranslate", "owner": "STranslate", "name": "STranslate", "repo_url": "https://github.com/STranslate/STranslate", "owner_url": "https://github.com/STranslate", "description": "A ready-to-go translation ocr tool developed with WPF/WPF 开发的一款即用即走的翻译、OCR工具", "language": "C#", "stars_total": 6355, "forks_total": 331, "stars_in_period": 627, "period_label": "最近30天", "description_zh": "A ready-to-go translation ocr tool developed with WPF/WPF 开发的一款即用即走的翻译、OCR工具", "readme_summary_zh": "English | 简体中文 A ready-to-go translation and OCR tool developed with WPF.", "source_period": "monthly", "signal_details": { "monthly": { "period": "monthly", "period_label": "最近30天", "rank": 15, "stars_in_period": 627 } }, "composite_score": 0.065599, "delta_since_last_run": null, "previous_rank": null } ] }