仓库/Skill 库

面向 OCR 历史文献的小型可复现预处理流水线，保留页码映射以实现透明搜索与导出。A small, reproducible preprocessing pipeline for OCR-based historical sources, retaining page mappings for transparent search and export.

AI 助手的安检守卫。检查 AI 尝试调用的每一个工具，防止数据泄露与未授权命令。A security guard for AI assistants. It checks every tool an AI tries to use, preventing data leaks and unauthorized commands.

🌐 通过 Geo-Llama 利用几何深度学习增强语言理解，结合 conformal manifolds 和递归等距变换提升 AI 模型性能。🌐 Enhance language understanding through geometric deep learning with Geo-Llama, leveraging conformal manifolds and recursive isometries for improved AI models.

本地 LLM 基准测试、RAG、语音交互与 AI 助手Local LLM benchmarking, RAG, voice interaction and AI assistant

NexaCrew（作者 Sin Chi Chiu / MAP Studio USA）—— AI-native、自托管的企业运营平台：ERP、POS、库存（IMS）、仓储（WMS）、HR、考勤、访客管理与门禁均由聊天 prompt 驱动。ISO 级记录、人脸识别 kiosk、station/kiosk 模式、自定义 SOP 构建器、车队更新。NexaCrew by Sin Chi Chiu / MAP Studio USA - AI-native, self-hosted Enterprise Operations Platform: ERP, POS, Inventory (IMS), Warehouse (WMS), HR, time-clock, visitor management and access control driven by chat prompts. ISO-grade records, face-recognition kiosks, station/kiosk modes, custom SOP builder, fleet updates.

你的 Mac 又慢又烫又吵。machogs 用朴实的英文告诉你罪魁祸首是谁，揪出那些吞噬 CPU 的 AI 工具辅助进程和卡住的进程。Your Mac is slow, hot and loud. machogs tells you what is doing it, in plain English. Finds the AI tool helpers and stuck processes eating your CPU.

用自然英语描述一个零件，即可获得完全可编辑的参数化 Fusion 360 模型。LLM 针对 28 个类型化 CAD 端点生成经过校验的 JSON feature 方案，通过 MCP server 与 Fusion 360 插件执行。附带 mock 后端，无需安装 CAD 即可运行。Describe a part in plain English, get a fully editable parametric Fusion 360 model. An LLM emits a validated JSON feature plan against 28 typed CAD endpoints, executed through an MCP server and a Fusion 360 add-in. Ships a mock backend so it runs with no CAD install.

2026 掌握 LLM 搜索：完整的语义 AI 手册Master LLM Search in 2026: The Complete Semantic AI Handbook

基于 MCP 的实时来源声明与引用验证，支持 FIGI 解析与签名投递回执。Live-source claim and citation verification over MCP, with FIGI resolution and signed delivery receipts.

面向 LLM Agent 的确定性写入时记忆整合与预取召回：采用本地模型，召回路径与整合决策中均不含 LLM。Deterministic write-time memory consolidation and pre-fetch recall for LLM agents. Local models, no LLM in the recall path or the consolidation decision.

使用纯 C99 MoE 推理引擎在 CPU 上原生运行 DeepSeek-V4-Flash-0731，无需 GPU、CUDA 或 PyTorch。Run native DeepSeek-V4-Flash-0731 on CPU with a pure C99 MoE inference engine — no GPU, CUDA, or PyTorch needed.

用于在 LLM 辅助工作中强制执行项目约定的 RETE 规则引擎。A RETE rules engine for durable enforcement of project conventions in LLM-assisted work.

🎨 利用 AI 实时生成高保真 UI 设计，对比多版本方案，并跨多个模型导出可直接使用的代码。🎨 Generate high-fidelity UI designs in real-time with AI, compare variations, and export ready-to-use code across multiple models.

基于 Torch2PC 的预测编码硕士论文可复现项目：在 Ubuntu/ROCm 上对 backpropagation 进行逐层与 compute-matched 对比，实现 PC-CATM/PC-TREF，对 state inference 进行机制诊断，并构建 QWake-PC 以实现自适应 exact inference —— 面向机制可解释的可复现预测编码研究。Воспроизводимый проект магистерской диссертации по predictive coding в Torch2PC: послойное и compute-matched сравнение с backpropagation, PC-CATM/PC-TREF, механизмная диагностика state inference и QWake-PC для адаптивного exact inference в Ubuntu/ROCm — reproducible research on mechanism-aware predictive coding.

AI-Core 2026：面向 OpenAI、Anthropic、Gemini 与 Grok API 管理的集中化 WordPress AI Provider 中枢。AI-Core 2026: Centralized WordPress AI Provider Hub for OpenAI, Anthropic, Gemini & Grok API Management

Agentic AI 平台，用于仓库分析、迁移规划、上下文驱动的代码转换及迁移后验证。Agentic AI platform for repository analysis, migration planning, context-grounded code transformation, and post-migration validation.

为 LLM 记忆而构建。Built for LLMs memory

使用这款专为架构设计与引擎特定任务优化的专用 LLM，提升游戏开发工作流效率。Optimize game development workflows with this specialized large language model designed for architecture design and engine-specific tasks.

拥有持久记忆的终端 AI 角色扮演客户端 —— 本地 SQLite、语义召回以及每次调用的成本统计。兼容 OpenAI，已在 OpenRouter 上测试。A terminal AI roleplay client with a memory that does not forget — local SQLite, semantic recall, and what every call cost. OpenAI-compatible, tested on OpenRouter.

Uniphore 是一家企业级 AI 公司——"The Business AI Company"——成立于 2008 年，总部位于加利福尼亚州帕罗奥图。其旗舰平台 Business AI Cloud（BAIC）是一个主权化、可组合且安全的 AI 平台，分为四层：用于对企业数据进行零拷贝访问的 Data Layer、用于检索的 Knowledge Layer……Uniphore is an enterprise AI company — "The Business AI Company" — founded in 2008 and headquartered in Palo Alto, California. Its flagship platform, the Business AI Cloud (BAIC), is a sovereign, composable and secure AI platform organised into four layers: a Data Layer for zero-copy access to enterprise data, a Knowledge Layer for retrieval and…

Fintool 是一款面向对冲基金、资管机构及买方/卖方分析师等机构投资者的 AI 驱动股票研究分析师与金融副驾驶。其助手可摄取 SEC 文件（10-K、10-Q、8-K）、财报电话会议记录与金融数据，并附带来源引用地回答研究问题，加速……Fintool is an AI-powered equity research analyst and financial copilot built for institutional investors such as hedge funds, asset managers, and buy-side and sell-side analysts. Its assistant ingests SEC filings (10-K, 10-Q, 8-K), earnings-call transcripts, and financial data to answer research questions with source citations, accelerating…

RIFT — Race-state Inference From Telemetry。一项可复现研究实现，仅基于位置遥测重建具备事件感知的关联性比赛状态，涵盖路线进度、共享事件标识、检查点通过、物理顺序、前车关系、间距与间隔。RIFT — Race-state Inference From Telemetry. A reproducible research implementation for reconstructing occurrence-aware relational race state from positional telemetry alone, including route progress, shared occurrence identity, checkpoint crossings, physical order, car-ahead relations, gaps and intervals.

通过动手实践的 Jupyter notebook 学习 agentic AI 概念，涵盖 LangGraph、CrewAI 与 OpenAI Agents 工作流。Learn agentic AI concepts through hands-on Jupyter notebooks featuring LangGraph, CrewAI, and OpenAI Agents workflows.

基于 Docker Compose 的 AI、LLM 与 RAG 平台基础设施蓝图，含路由、存储与可观测性说明。AI, LLM, and RAG platform infrastructure blueprint with Docker Compose, routing, storage, and observability notes.

通过分析 trace 诊断与调试 RAG 流水线，定位并修复影响答案质量与相关性的问题Diagnose and debug Retrieval Augmented Generation pipelines by analyzing traces to identify and fix issues affecting answer quality and relevance.

🌐 利用多跳图推理结合知识图谱和向量搜索，挖掘隐藏的全球供应链风险，获得更深层洞察。🌐 Map hidden global supply chain risks with multi-hop graph reasoning combining knowledge graphs and vector search for deeper insights.

AI 驱动的加密情绪日记——基于 RAG 的 LLM 反馈（TAIDE + ChromaDB）结合 Random Forest 情绪预测。Live demo：heartbox.tw（test1/test1）。AI-powered encrypted mood journal — RAG-grounded LLM feedback (TAIDE + ChromaDB) and Random Forest emotion forecasting. Live demo: heartbox.tw (test1/test1)

🌐 在数据受限条件下重新思考多模态大语言模型的设计与扩展，以 NaViL 通过 Native Training 提升效率与性能🌐 Rethink Multimodal Large Language Models design and scaling under data constraints with NaViL, enhancing efficiency and performance through Native Training.

一个 agentic、循证的研究系统：通过工具调用、RAG 与显式的论据—来源核验从研究问题生成结构化、带引用的报告，避免盲目的 LLM 摘要；能识别来源冲突并如实报告不确定性An agentic, evidence-grounded research system that generates a structured, citation-backed report from a research question — using tool-calling, RAG, and explicit claim-to-source verification instead of blind LLM summarization. Detects conflicting sources and reports uncertainty rather than hiding it.

LLM 驱动的科学论文推荐系统LLM-driven scientific paper recommendation system

一个基于 C# 与本地 LLM 的学术论文 RAG（Retrieval-Augmented Generation）系统。目标是摄入研究论文语料，构建可回答文献综述类问题（如"其他论文关于 X 发现了什么"）的系统，回答严格基于真实源材料。A RAG (Retrieval-Augmented Generation) system for academic papers using C# and a local LLM. The goal is to ingest a corpus of research papers and build a system that can answer literature review questions like "what have other papers found about X"; grounded in the actual source material.

SLM-PICO-Screener 是用于自动化系统综述筛选的轻量 NLP 流水线，基于 Phi-3 Mini 配合 LoRA 与 4-bit 量化，将文章分类为 8 个 PICO 标签。体积约 90 MB，可在消费级 GPU/CPU 上运行，加速生物医学研究中的证据综合。SLM-PICO-Screener is a lightweight NLP pipeline automating systematic review screening. Built on Phi-3 Mini with LoRA & 4-bit quantization, it classifies articles into 8 PICO labels. At ~90 MB, it runs on consumer GPUs/CPU, accelerating evidence synthesis in biomedical research.

232M token 的 LLM 训练语料库：140 个颠覆性编程范式（60 个基础 + 80 个神经科学启发）+ 10 个神经 AI 表格数据集。真实神经科学内容：STDP、预测编码、自由能、海马体、Hebb。包含 JSONL+JSON+CSV。142,800 条范式条目 + 100K 神经样本。232M-token LLM training corpus: 140 disruptive programming paradigms (60 foundational + 80 neuroscience-grounded) + 10 neuro AI tabular datasets. Real neuroscience: STDP, predictive coding, free energy, hippocampus, Hebb. JSONL+JSON+CSV. 142,800 paradigm entries + 100K neuro samples.

60 个神经科学启发的软件架构范式。真实研究（Hebb、Bi & Poo、Friston、Buzsaki、Moser、Hodgkin-Huxley）映射到软件模式。61,200 条 JSONL 条目，约 82M token。全英文。60 neuroscience-grounded paradigms for software architecture. Real research (Hebb, Bi & Poo, Friston, Buzsaki, Moser, Hodgkin-Huxley) mapped to software patterns. 61,200 JSONL entries, ~82M tokens. All in English.

Ultimate Rulesync AI Agent CLI Tool 2026 – Streamline Coding Workflows FastUltimate Rulesync AI Agent CLI Tool 2026 – Streamline Coding Workflows Fast