仓库/Skill 库

Cypress AI 演示Cypress AI Demo

专注后端的 fullstack 与 applied AI 开发。基于 LLMs 与 agents 构建有趣的项目，执着于 clean architecture 与零缺陷代码的生产部署。Backend-focused fullstack & applied ai dev. Building cool stuff with LLMs & agents. Obsessed with clean architecture & shipping no-bug code to prod.

🤖 仅需 10 行代码即可在 Android 上免 root 自动回复聊天消息，适合使用 Autojs 或 Hamibot 进行测试。🤖 Automate chat replies on Android without root using just 10 lines of code. Ideal for testing with Autojs or Hamibot.

基于 FastAPI + LangChain + RAG 的 AI 智能对话助手，支持多轮对话记忆、图片分析、流式回复、知识库 RAG 检索、上传定义知识库。

面向 AI 生成可执行方案的 verifier-first 运行时，支持独立验证、对比与检索。A verifier-first runtime for independently verifying, comparing, and searching AI-generated executable solutions.

使用 AI agent 自动化文献综述，基于结构化协议实现全面研究与高质量报告生成📚 Automate literature surveys with AI agents using a structured protocol for thorough research and insightful report generation.

基于 LLM 的科研 Wiki，支持 Zotero 文献导入、MinerU PDF 转换以及知识图谱辅助的科学写作A llm research wiki for Zotero-backed literature ingestion, MinerU PDF conversion, and knowledge-graph-assisted scientific writing.

AI 安全测试——发现 LLM 应用、聊天机器人、AI Agent、MCP 服务器与 RAG 系统中的安全漏洞。对齐 OWASP（LLM、Agentic、MCP）与 NIST AI RMFAI security testing — find security vulnerabilities in LLM apps, chatbots, AI agents, MCP servers, and RAG systems. Mapped to OWASP (LLM, Agentic, MCP) and NIST AI RMF.

插件化形态的 14-Agent 软件公司：在硬性预算上限内规划、构建并对每次变更进行对抗式评审。Markdown + POSIX shell，零运行时依赖。A 14-agent software company in a plugin: plan, build, and adversarially judge every change under hard budget caps. Markdown + POSIX shell, zero runtime deps.

面向可信 AI Agent 的开源基础设施，覆盖身份、权限、执行、证据、验证与 Agent 间组网。Open-source infrastructure for trusted AI agents — identity, permissions, execution, evidence, verification and agent-to-agent networking.

证据约束的代码诊断工作台：确定性静态分析拥有源码事实，LLM 只提假设，resolver/validator 决定能否绑定回真实操作——证据不足保持 unknown/abstained，从架构抑制幻觉。已在真实开源项目发现 bug（stunner #89 已被上游确认修复）。Python。

自托管、100% 本地的 AI 平台——在一个 Docker 栈中集成 LLM 推理、RAG 与知识图谱。无需 API key。Self-hosted, 100% local AI platform — LLM inference, RAG, and knowledge graphs in one Docker stack. No API keys.

自构建的 AI Agent 框架——多模型 ReAct + 分层缓存友好的上下文 + 异步多 Agent + XML 工作流。CLI 与 WebUI，零 LangChain。An AI agent framework that builds itself — multi-model ReAct + tiered cache-friendly context + async multi-agent + XML workflows. CLI & WebUI, zero LangChain.

Vantio Optics — 为 LLM/agent 出站流量提供免费的可观测能力。Sight Loop。设计即盲视。非代理。@vantio/cli 0.3.2 · vantio-agent-sdk 3.0.2Vantio Optics — free observe for LLM/agent egress. Sight Loop. Blind by design. Not a proxy. @vantio/cli 0.3.2 · vantio-agent-sdk 3.0.2

基于 RAG 的文档问答机器人，可对任意 PDF/文本文件提问。使用本地 embeddings + Groq API。学习要点：embeddings、向量搜索、分块、RAG 流水线。Document Q&A bot using RAG (Retrieval-Augmented Generation). Ask questions about any PDF/text file. Uses local embeddings + Groq API. Learns: embeddings, vector search, chunking, RAG pipeline.

全栈可控的开源 TypeScript SDK，专为 AI agent 设计 ——Apache-2.0 本地运行时、43 个 LLM provider（使用自有 key），以及可选云端。Theo 生态的 Harness 支柱。The open-source TypeScript SDK for AI agents you own end to end — an Apache-2.0 local runtime, 43 LLM providers on your own keys, and an opt-in cloud. The Harness pillar of the Theo ecosystem.

《大模型推理原理与优化》：面向系统/架构/后端研发工程师的模型原理入门课，目标是通俗易懂的解释推理过程，理解原理有助于系统开发/维护工作

一切留痕：需求即 issue，决策即审批评论，工作即 PR，理由即该 PR 的记录。协调器将对话中继到 GitHub，并为每个角色派生沙盒化会话，各自仅持有自己的规则集。Everything on the record: a requirement is an issue, a decision is an approval comment, work is a PR, rationale is that PR's record. The orchestrator relays conversation to GitHub and spawns sandboxed role sessions, each with only its own rulebook.

🌐 通过这款高效的网页抓取 MCP server，从任意 URL（含 Cloudflare 防护站点）提取干净、可直接用于 LLM 的 markdown🌐 Extract clean, LLM-ready markdown from any URL, including Cloudflare-protected sites, with this effective MCP server for web scraping.

Meta 分析研究软件：通过厂商无关的 LLM 适配器从学术 PDF 中半自动提取效应量，由首席研究员进行人机协同验证，并采用不可变的数据锁定流程，导出可用于三层 Meta 分析回归的可复现效应量数据集Research software for meta-analysis: semi-automated effect-size extraction from academic PDFs with a vendor-neutral large-language-model adapter, human-in-the-loop verification by the principal investigator, and an immutable data-lock workflow that exports a reproducible effect-size dataset for three-level meta-analytic regression.

💹 使用 Python 和 CoinGecko API 分析主要加密货币的价格、市值与趋势，辅助投资决策💹 Analyze major cryptocurrencies' prices, market caps, and trends using Python and the CoinGecko API for informed investment decisions.

DP-ES 官方实现：面向 Prompt 优化的差分隐私进化策略（EMNLP 2026）。Official implementation of DP-ES: Differentially Private Evolution Strategies for Prompt Optimization (EMNLP 2026)

一个可复用的 Codex Skill，用于证据对齐、AI 与人工审稿人就绪的学术写作。A reusable Codex skill for evidence-aligned, AI- and human-reviewer-ready academic writing.

面向生产环境的、有内存约束的跨模型 KV-cache 传输，具备受保护的回退机制与可复现研究工具链。Production-oriented, memory-bounded cross-model KV-cache transfer with guarded fallback and reproducible research tooling.

面向 AI Agent 工具调用的确定性 <0.1ms 不变量清算内核。零外泄 AST SQL 防火墙、数值速率限制与 Merkle GRC 审计追溯。Deterministic <0.1ms Invariant Clearance Kernel for AI Agent Tool Calls. Zero-egress AST SQL firewall, numeric velocity limits & Merkle GRC audit trails.

🧠 构建模拟人类记忆的 AI 认知架构，分离情景记忆与语义记忆，并保证多语言支持与数据完整性🧠 Build a cognitive architecture for AI that mimics human memory, separating episodic and semantic memory while ensuring multilingual support and data integrity.

用于微调的合成数据样本集合，旨在为更大规模合成数据集提供种子数据This is a collection of synthetically generated datapoints for fine tuning meant for seeding larger synthetic datasets.

基于 Microsoft Agent Framework (.NET) 的 LLM Agent 三层评估：运行时护栏、PR 闸口、校准过的评审者。该框架可自我评估。Three-tier evaluation for LLM agents on Microsoft Agent Framework (.NET): runtime guardrails, a PR gate, and calibrated judges. The framework evaluates itself.

一个 API 接入 2000+ AI 模型、2000+ 工具与托管 agents。One API for 2000+ AI models, 2,000+ tools, and managed agents.

2026 年顶级开源 AI Agent 🤖 | 最佳自主工具与框架Top Open-Source AI Agents 2026 🤖 | Best Autonomous Tools & Frameworks

使用 HTML 和 CSS 构建简洁、风格化的网站，配备响应式布局、导航与现代设计，提升用户参与度🌐 Build a simple, stylish website using HTML and CSS, featuring a responsive layout, navigation, and a modern design for easy user engagement.

GitHub 个人主页 README。GitHub profile README

面向 AI agent 与 agentic workflow 的开源 MLflow 插件：涵盖 prompt、tool、skill、MCP server、RAG 知识库、评估、部署、可观测性以及 Aria copilot。Open-source MLflow plugin for AI agents and agentic workflows: prompts, tools, skills, MCP servers, RAG knowledge bases, evaluation, deployment, observability, and Aria copilot.

软件工程作品集 | Web 开发 | 桌面应用开发Software Engineering Portfolio | Web Development | Desktop App Development

🛠️ 使用 Haystack 轻松构建强大搜索系统，该框架用于开发端到端问答与搜索应用。🛠️ Build powerful search systems effortlessly with Haystack, a framework for developing end-to-end question answering and search applications.