近 30 天新建、star 最多的 AI 开源项目 Top 30（数据来自 GitHub 官方 API，每日快照）。
今日导读
本周最大看点是xAI官方开源了grok-build，作为Grok的编码Agent终端工具一举收获逾万star，与LangChain推出的openwiki几乎同档量级，两者共同印证一个信号：头部实验室正在加速把Agent基础设施标准化并向外输出。更值得留意的是列表中悄然浮现的"技能插件化"趋势——Cloudflare安全审计skill、公众号排版skill、个人skills接连上榜，意味着Agent生态正在朝乐高式模块拼装演进，谁能建立高质量skill分发渠道，谁就可能成为下一个Agent Store。此外，红队自动化平台T3MP3ST与Codex破甲提示词包同日冲榜绝非偶然，攻击侧工具的活跃度与Agent能力同步走高，安全债务正在悄悄积累，这个方向的关注度恐怕还会持续上升。
| # | 项目 | Star | 语言 | 创建 |
|---|---|---|---|---|
| 1 | xai-org/grok-build SpaceXAI's coding agent harness and TUI. Fullscreen, mouse interactive, extensible. | 13,067 | Rust | 2026-07-14 |
| 2 | langchain-ai/openwiki OpenWiki is a CLI that writes and maintains agent documentation for your codebase. | 11,940 | TypeScript | 2026-06-22 |
| 3 | elder-plinius/T3MP3ST autonomous red teaming platform; multi-agent offensive-security meta-harness | 4,849 | TypeScript | 2026-07-02 |
| 4 | oomol-lab/open-connector Open-source auth gateway connecting 1000+ SaaS providers to AI agents through SDK, CLI, MCP, HTTP, and OpenAPI. | 2,774 | TypeScript | 2026-06-29 |
| 5 | davidondrej/skills access to david ondrej's personal agent skills | 2,567 | Shell | 2026-06-24 |
| 6 | cloudflare/security-audit-skill A coding-agent skill for multi-phase security audits with independently verified, machine-readable findings | 2,546 | JavaScript | 2026-06-18 |
| 7 | synthetic-sciences/openscience The open-source AI workbench for scientific research | 2,524 | TypeScript | 2026-07-03 |
| 8 | isjiamu/gzh-design-skill 把 Markdown 一键排成可直接粘进公众号编辑器的精致 HTML —— 6 套精选主题 + 主题生成器 + 双关卡校验。An AI-agent skill that turns Markdown into paste-ready WeChat article HTML. | 2,259 | HTML | 2026-07-01 |
| 9 | shy3130/tickflow-stock-panel 自托管、零运维的 A 股「选股 + 监控 + 回测」量化工作台 | 基于 TickFlow 数据源 | LLM能力驱使策略定制+个股分析+复盘 | 自由接入第三方数据源与个性化扩展数据 | 个人开源 ,非TickFlow官方项目 | 2,223 | Python | 2026-06-18 |
| 10 | MDX-Tom/gpt-5.6-instruct A Codex CLI jailbreak prompt and test pack for gpt-5.6-sol. 针对 gpt-5.6 系列的 Codex CLI 破甲提示词与测试包。 | 1,720 | Python | 2026-07-11 |
| 11 | HUANGCHIHHUNGLeo/claude-real-video Let Claude (or any LLM) actually watch a video — scene-aware, deduplicated frames + transcript, from a URL or local file. Runs locally, MIT. | 1,693 | Python | 2026-06-30 |
| 12 | jamesob/local-llm Everything I know about running LLMs locally | 1,536 | Shell | 2026-07-03 |
| 13 | shepherd-agents/shepherd A runtime substrate that turns an agent's execution into a reversible, Git-like trace, so meta-agents can observe, fork, replay, and revert any run. Couples agent and environments in a copy-on-write fork ~5x faster than docker commit, with ~95% KV-cache reuse on replay. Framework built for meta-agents to supervise, optimize, and train other agents | 1,434 | Python | 2026-06-24 |
| 14 | amplifthq/opentag Open-source @agent mentions for Slack and GitHub. OpenTag routes tagged requests to Codex, Claude Code, then returns results in thread. | 1,363 | TypeScript | 2026-06-24 |
| 15 | Forsy-AI/agent-apprenticeship The living ecosystem where AI agents complete tasks through workflow loops, improve through iterative execution, are evaluated by mentor agents or humans in the loop, and turn completed work into reusable work experience and data to improve future agents. | 1,317 | Python | 2026-06-19 |
| 16 | Sahir619/fable-method The Fable Workflow: how Claude Fable 5 worked, distilled into skills any model can run, with the eval that keeps it honest. Think / act / prove. | 1,310 | Python | 2026-07-06 |
| 17 | lyra81604/zhengxi-views 可溯源的郑希(易方达基金经理)投研 Agent Skill——基于他全部公开观点原文 + 有原话佐证的投资方法 + 全市场基金真实数据，能溯源问答、按他框架给基金打分，绝不杜撰。⚠️仅研究学习辅助，不构成投资建议‼️website是郑希主页！ | 1,305 | Python | 2026-06-20 |
| 18 | sums001/Windows-Copilot-API Reverse engineered Windows Copilot into an OpenAI-compatible API. Access GPT-4 and GPT-5 models through a simple REST interface without API keys or billing. | 1,133 | Python | 2026-06-19 |
| 19 | fancydirty/mediary-scout Agent-driven media library for your cloud drives (115 / Quark / 光鸭 GuangYaPan). | 1,065 | TypeScript | 2026-06-21 |
| 20 | abundantbeing/hermes-browser-extension Browser-native side panel for Hermes Agent — connect web context to your local Hermes runtime. | 1,029 | JavaScript | 2026-06-24 |
| 21 | AlephAITech/WorkBuddyGuide A practical, open-source guide to mastering WorkBuddy through real-world workflows.开源的 WorkBuddy 实战蓝皮书：教程、真实工作流、Skills、MCP、自动化与多智能体实践。 | 1,012 | Python | 2026-07-10 |
| 22 | jmerelnyc/Talos GPU worker client for the Talos network. Pairs with your Talos account, serves open-model inference jobs over a WebSocket, and reports uptime for payouts. | 992 | Python | 2026-07-02 |
| 23 | modiqo/skillspec SkillSpec makes agent skills followable, testable, and provable with Doctor risk reports, guided imports, structured contracts, and alignment proof. | 974 | Rust | 2026-06-19 |
| 24 | lycorp-jp/sim-use Give your AI agent eyes and hands on iOS Simulator and Android emulator/devices. | 928 | Swift | 2026-06-26 |
| 25 | deer-flow/llm-space A desktop app to prototype agent ideas, inspect every harness step, replay failures, and evaluate performance, all in one place. Local-first. | 902 | TypeScript | 2026-06-28 |
| 26 | Kulaxyz/self-learning-skills A self-improving skill for AI coding agents (Claude Code, Cursor, AGENTS.md): recognize a hard-won golden path in a session and harvest it into a reusable skill/rule for next time. | 882 | — | 2026-06-28 |
| 27 | simonlin1212/Vibe-Research Vibe-Research: Your Personal Trading Research Agent · A股/美股/港股 的个人投研 Agent：每日复盘、资讯雷达、个股数据、板块中心、我的持仓、研究记录。Vibe-Research 把数据和功能配齐，由你自己的 AI 驱动投资研究。 | 869 | TypeScript | 2026-07-05 |
| 28 | HKUDS/OpenOPC OpenOPC: Build Your Personal AI-Native Company — Self-Built, Self-Run, Self-Grown | 854 | Python | 2026-07-01 |
| 29 | uzairansaruzi/hermex Native iPhone app for your Hermes agent | 831 | Swift | 2026-07-02 |
| 30 | QwenLM/Qwen-AgentWorld Qwen-AgentWorld: Language World Models for General Agents | 830 | Python | 2026-06-22 |
历史日报
数据来源 GitHub Search API；「近 30 天新建 + star 排序」反映新项目热度，与 GitHub Trending 口径略有不同。项目描述为原文，导读为本站编辑内容。