仓库/Skill 库

DeepSeek v4 Flash EXL3 运行于单台 DGX SparkDeepSeek v4 Flash EXL3 on one DGX Spark

A self-evolved, sovereign desktop AI agent built for real-world action — beyond model wrappers and coding copilots. / 真实实践中自我演化的独立桌面 AI Agent。

MCP 连接器，让你的 AI Agent 直接访问 typesafe ai 的 jev 模型mcp connector to give your AI agent direct access to typesafe ai's jev model

Model Context Protocol (MCP) 服务器的 Node.js 实现，支持从多源（包括 **Web of Science**、arXiv 等）搜索和下载学术论文。A Node.js implementation of the Model Context Protocol (MCP) server for searching and downloading academic papers from multiple sources, including **Web of Science**, arXiv, and more.

GLM-5.3 Flash EXL3，针对 2x DGX Sparks。GLM-5.3 Flash EXL3 for 2x DGX Sparks

AI 驱动的 CLI 编码 Agent，内置 20+ 工具，支持 MCP 与多模型提供商。AI-powered CLI coding agent with 20+ built-in tools, MCP support, and multi-model providers

mu (μ)：一个先思考后行动的 coding agent。小型快速的 judge 处理日常调用，大模型完成实际工作。基于 pi 和 AionUi 构建。mu (μ): a coding agent that thinks before it acts. A small, fast judge makes the routine calls, the big model does the work. Built on pi and AionUi.

最省钱的自托管 AI Agent 工作空间，集成 tool calling、MCP、多模型路由、沙箱执行、多 Agent 工作流，以及由 LLM 编写并通过 Three.js 渲染的 3D 角色动画。Cheapest Money-Saving Self-hosted AI agent workspace with tool calling, MCP, multi-model routing, sandboxed execution, multi-agent workflows, and LLM-authored 3D character animation rendered with Three.js.

AI 大模型世界，把 556 个大模型拟人化成像素小人的可视化站点。进来就能看到此刻谁最聪明、谁最会写代码、谁最便宜、谁刚发布，往下是国内与国外分区的厂商广场、完整的发布时间线和多维排行榜。搜索认模型名、厂商和能力，输入「多模态」会直接列出全部多模态模型。数据取自 Epoch AI、models.dev、LiveBench 与 Hugging Face，每小时自动同步，所有文案由真实数据生成，不调用任何 LLM。Next.js 静态导出，零后端。

一个 Model Context Protocol (MCP) 服务器，使 AI 助手能够与 AKS 集群交互，作为 AI 工具（如 Claude、Cursor 和 GitHub Copilot）与 AKS 之间的桥梁A Model Context Protocol (MCP) server that enables AI assistants to interact with AKS clusters. It serves as a bridge between AI tools (like Claude, Cursor, and GitHub Copilot) and AKS.

🦆🧠 给你小巧的机器人装上大脑。用自然语言告诉 Microduck、自制的 Open Duck Mini、Reachy Mini、LeRobot 机械臂或任意 ROS base 你想要什么。LLM（Claude、OpenAI、Gemini、Grok，或通过 Ollama、vLLM、llama.cpp 接入的本地模型）调用它已有的 Skill。包含模拟器、.duck 任务文件、MCP、flocks。Apache 2.0。🦆🧠 Give your small robot a brain. Tell a Microduck, an Open Duck Mini you built yourself, a Reachy Mini, a LeRobot arm or any ROS base what you want in plain language. An LLM (Claude, OpenAI, Gemini, Grok, or a local model via Ollama, vLLM or llama.cpp) uses the skills it already has. Simulator, .duck task files, MCP, flocks. Apache 2.0.

Code World Model 官方仓库。Code World Model Official Repo

面向 agentic 界面的 React design system。提供每个免费 BoardUI 组件的源代码，并以使用你自己 model key 的可运行 AI chat 应用作为主页。React design system for agentic interfaces. Every free BoardUI component as source, with a working AI chat app on your own model key as the homepage.

将任意 LLM 转变为 Jev 风格决策模型：类型化决策、真实概率、无需训练。（持续更新）Turn any LLM into a Jev-style decision model: typed decisions, real probabilities, no training. (continue updating)

Apeireth —— 一个纯 Safe Rust 的 AGI 操作系统与认知微内核。包含 16 个 crate：连续拓扑记忆、因果世界模型、认知调度器、Ember HUD、三层 onion 安全机制以及可移植 USB Agent。Apeireth — A Pure Safe Rust AGI Operating System & Cognitive Microkernel. 16 crates: continuous topological memory, causal world model, cognitive scheduler, Ember HUD, triple-onion security & portable USB agent.

用于解析日期和时间的小型神经模型A small neural model for parsing date & time

JevK5：TypeSafe Jev 的开源权重替代方案。单次前向即可输出带概率的类型化决策；权重与代码均采用 Apache-2.0 协议。JevK5: open-weight alternative to TypeSafe Jev. Typed decisions with probabilities in one forward pass; Apache-2.0 weights and code.

AI 数据交换的十字路口。一个统一的、可自托管的 Web 界面，用于发现、配置和管理 Model Context Protocol (MCP) 服务器——将来自多个 MCP 来源（Claude、Cursor 等）的 AI 工具、工作区、提示词与日志汇聚一处。The Crossroads for AI Data Exchanges. A unified, self-hostable web interface for discovering, configuring, and managing Model Context Protocol (MCP) servers—bringing together AI tools, workspaces, prompts, and logs from multiple MCP sources (Claude, Cursor, etc.) under one roof.

开源 multi-agent 研究助手：约 2 分钟生成带引用的网络报告，或由 25 个 agent 流水线配合人工检查点产出完整学术论文。欢迎贡献。Open-source multi-agent research assistant: cited web reports in ~2 min, or a full academic paper from a 25-agent pipeline with human checkpoints. Contributors welcome.

将 Minimax-H3 蒸馏为 4 步。Distill Minimax-H3 into 4 steps

开源核心的 AI 工作台——覆盖任何模型的 notebooks、Agent、RAG、语音和图像：OpenAI、Anthropic、Google、xAI，或通过 Ollama/vLLM 的本地模型。BSL 1.1，两年后自动转换为 Apache-2.0。当别人的 AI 停止运行时，你的 AI 仍在运行。The open-core AI workbench — notebooks, agents, RAG, voice, and images across any model: OpenAI, Anthropic, Google, xAI, or local via Ollama/vLLM. BSL 1.1, auto-converting to Apache-2.0 on a two-year clock. Your AI keeps running when theirs doesn't.

一个小型开放决策模型：state + 类型化问题 → 校准概率。在 Qwen3.5 上的 Jev / System One 复现。A small open decision model: state + typed questions -> calibrated probabilities. A Jev / System One re-creation on Qwen3.5.

通过你的 ZCode Desktop 账号路由，从你自己的 agent harness 调用模型——在你选择的 harness 中使用每日免费 Tokens 和奖励 TokensModel access from your own agent harness through routed frim your ZCode Desktop account — use your free daily Tokens and bonus Tokens in the harness of your choice

DeepSeek Harness 插件，支持跨预设会话迁移的可预览；固定 schema 的交接保留状态、源模型意图与未处理图像，原始会话保持不变DeepSeek Harness plugin for previewable cross-preset session migration. Fixed-schema handoffs preserve state, source-model intent, and unresolved images; the original session stays untouched.

基于 PyTorch 的韩语 LLM 实现与训练项目PyTorch 기반 한국어 LLM 구현 및 학습 프로젝트

🧠💻 将互联网重新构想为自组织思维导图 🤖🔎 STREAM：基于顶部结果提取与答案模型的搜索 📈📝 REASON 文档撰写 Agent 🚜📜 Tractor 文本提取器 🔤📊 SEEKTOPIC🧠💻 Reimagine the Internet as Self-Organizing Mind Map 🤖🔎 STREAM: Search with Top Result Extraction & Answer Model 📈📝 REASON Docs Writing Agent 🚜📜 Tractor the Text Extractor 🔤📊 SEEKTOPIC

小到 AI 可以整体容纳的 App。ObjectStack 将整个 App——数据模型、UI、工作流、权限——转化为类型化元数据：一个完整 CRM 压缩至 150k token 以内，单个上下文窗口即可装下。Agent 可整体读取、整体推理、整体重构。Apps small enough for AI to hold whole. ObjectStack turns the whole app — data model, UI, workflows, permissions — into typed metadata: a complete CRM in under 150k tokens, one context window. Agents read it whole, reason it whole, refactor it whole.

在 220k GB200 上训练前沿模型的完整脚本。"普罗米修斯从众神那里盗取火种，将其赐予人类。"Full script for training a frontier model on 220k GB200s. "Prometheus stole fire from the gods and gave it to man."

一个 Model Context Protocol (MCP) 服务器，使 AI 助手能够与 Kubernetes 集群交互，作为 AI 工具（如 Claude、Cursor 和 GitHub Copilot）与 Kubernetes 之间的桥梁A Model Context Protocol (MCP) server that enables AI assistants to interact with Kubernetes clusters. It serves as a bridge between AI tools (like Claude, Cursor, and GitHub Copilot) and Kubernetes

3D 打印骑行装备 2026：轮挡、平板支架与 AirTag 挂架3D Printable Cycling Gear 2026: Wheel Blocks, Tablet Stands & AirTag Holders

基于 Jev API 的开放决策模型：通过前向传递给出带概率的类型化答案（yes/no、choice、score、multi），无需生成。基于 Qwen3.5-4B + LoRA，单卡 GPU。Open decisions model with Jev's API: typed answers (yes/no, choice, score, multi) with probabilities from forward passes, no generation. Qwen3.5-4B + LoRA, one GPU.

Agentic RuntimeAgentic Runtime

发布 AI Agent 制品，而非 demo 表演 — 验证器把关、跨模型族运行，自动学习哪个模型最优。Ship AI-agent artifacts, not demo theater — verifier-gated, cross-family runs that learn which model wins.

面向 Claude、Cursor 及 AI agent 的 GeneXus 18 MCP server，通过 Model Context Protocol 对 KB 对象（transactions、web panels、procedures、SDTs）进行读取、编辑与分析。GeneXus 18 MCP server for Claude, Cursor, and AI agents — read, edit, analyze KB objects (transactions, web panels, procedures, SDTs) over the Model Context Protocol.

通过 Ollama 使用本地 LLM 自动生成文献综述：检索学术数据库、获取全文论文并撰写 APA 第 7 版格式综述；模型在结构上被禁止编造引文，每条引文均逐字核验并按 APA 第 7 版排版。Automated literature reviews using local LLMs via Ollama. Searches academic databases, acquires full-text papers, and writes APA 7th reviews where the model is structurally prevented from fabricating quotes — every quote is verified verbatim and is formated in APA7th.

[CoLM 2026] MoANT 官方代码：面向多任务大语言模型微调的语义感知秩一专家混合模型[CoLM 2026] Official code for MoANT: Mixture-of-Rank-One-Experts with semantic-aware Intuition for Multi-task Large Language Model Finetuning