海外机会工作台

Healthy Diet AI Agent — Version-Aware RAG for Evolving Nutrition Guidelines English | 日本語 | 繁體中文 This repository contains a nutrition-advice AI agent and the research behind its retrieval layer. It is the follow-up to our first paper, whose system is the PU-Hub/healthy-diet project (a healthy-diet app with YOLO food recognition and LLM-based advice). The first work established the end-to-end application. This project focuses on one open problem we found while building it: how a retrieval-augmented generation (RAG) system should handle health guidelines that change over time. Research question. When a knowledge base holds several versions of an official guideline, which evidence should a RAG…

Genshin-TS Use TypeScript to develop Genshin UGC (Miliastra Wonderland) projects. Full type system, practical helpers, JS-native and Unity3D-style APIs, npm ecosystem support, and an AI-friendly workflow. English · 简体中文 Docs · Intro Video · Template Guide & Usage Details Introduction Genshin-TS (gsts) is a toolchain for building Miliastra Wonderland projects in TypeScript. It focuses on a code-first experience with controllable node-graph semantics, supporting compilation, injection, debugging, optimization, parallel builds, and incremental builds. https://github.com/user-attachments/assets/0f2ec8af-c753-4c54-940c-8147ec856635 Quick Start If the Windows execution policy prevents scripts fro…

obsidian-web-mcp A secure, remote-accessible MCP server that gives LLMs read/write access to your Obsidian vault from anywhere -- your desktop, your phone, a hotel Wi-Fi network. Unlike local-only Obsidian MCP servers, this one runs over HTTPS with real authentication, so Claude (or any MCP client) can reach your vault whether you're at your desk or not. It reads and writes markdown files on disk, parses YAML frontmatter, maintains an in-memory frontmatter index for fast queries, and handles full-text search -- all behind OAuth 2.0 authentication and a Cloudflare Tunnel that never exposes your machine directly to the internet. Why This Exists There are many Obsidian MCP servers. Most are lo…

Supriya Supriya is a Python API for SuperCollider. Supriya lets you: Boot and communicate with SuperCollider's synthesis engine in realtime. Explore nonrealtime composition with scores. Compile SuperCollider SynthDefs natively in Python code Build time-agnostic asyncio-aware applications with the context interface. Schedule patterns and callbacks with tempo- and meter-aware clocks Integrate with IPython, Sphinx and Graphviz Quickstart Get Supriya Install from PyPI: pip install supriya Or from source: git clone https://github.com/supriya-project/supriya.git cd supriya pip install -e . Get SuperCollider Get SuperCollider from http://supercollider.github.io/. Boot the server Start your Python…

Web3-Ethereum-Defi Web-Ethereum-DeFi (ethdefi) allows you to integrate EVM-compatible Web3 and DeFi protocols into your Python application. Use cases Supported protocols, chains and integrations Prerequisites Install Example code Uniswap swap example How to use the library in your Python project Documentation Development and contributing Version history Support Social media License Use cases Use cases for this package include Trading and bots Data research, extraction, transformation and loading Portfolio management and accounting System integrations and backends AI agent interaction for EVM chains Supported protocols, chains and integrations 👉 Read the full API documentation. This is a MIT…

English · 简体中文 · 繁體中文 · 한국어 · 日本語 · Русский · Français Codex Runway How much longer can your Codex keep running? Codex Runway is a native macOS menu bar app for checking Codex and Grok quota. It also provides Codex reset-today status, reset credits, API-equivalent cost, and local sessions, with separate multi-account management for each provider. Highlights Check remaining Codex quota from the menu bar. Switch from the always-visible Codex | Grok tabs; the menu bar and context menu follow the selected provider. View included Grok quota, product breakdown (Build / Imagine / Chat), weekly/monthly periods, prepaid balance, and on-demand usage via the official CLI chat-proxy billing API. On the…

brain-mcp v1.0 Transportable AI memory, now shipping as a distributable. Your AI conversation history, queryable from any MCP-aware LLM, stored entirely on your machine. What changed in v1.0 brain-mcp v0.x was a hosted SHELET reference implementation pointed at a Supabase substrate. v1.0 is a thin MCP server that wraps the Bob protocol. Storage moved from cloud Postgres to local parquets under /.bob/. The protocol is free; brain-mcp is one implementation; a paid bob managed tier (cloud sync, recipes) is opt-in and lives in a separate package. The 53 of you who starred v0.x: thank you. v1.0 is the version we were always going to build — memory belongs to the rememberer is the architecture th…