Backend / AI Engineer
Repositories
yushui2022/MathModel-Skill
Agent-native mathematical modeling workflow skills for Trae, Claude Code, and Codex, covering problem parsing, modeling, code generation, evidence checks, paper writing, and Word delivery.
yushui2022/EasySandbox
A deliberately small Linux process sandbox launcher for agent workloads: namespaces, read-only rootfs, Landlock/seccomp policy, host-side CONNECT proxy allowlist, JSONL records and verify.
yushui2022/ForkGate
yushui2022/Aftercare-Agent
yushui2022/jev-hypit-commerce
Open-source product-to-creator matching: BM25 / Qdrant retrieval, Jev decisions, human review and dual-brand Hypit video production.
yushui2022/dsh-novel
yushui2022/Ouroboros
yushui2022/mathmodel
基于langchain实现的自动化建模
yushui2022/EdgeHome-Harness
面向 1B 端侧小模型的 Rust Agent Harness：在 2GB RAM 约束下实现稳定 JSON/intent 输出、内存预算、安全策略、执行控制与失败回放。
yushui2022/Evidence-Gated-Memory
Evidence-gated graph memory for hard-anchor enterprise AI agents. Built on raw evidence refs, fact lineage, gated task state, freshness checks, audit trails, and DAG-style provenance.
yushui2022/AI-galgame
yushui2022/easy-coding-agents
One of the early China-built lightweight Claude Code-inspired terminal coding agents, with autonomous execution, tool calling, custom agents, task control, and evidence-gated memory.
yushui2022/AntiAIGC
Model Fingerprint-Based Anti-AIGC System for LLM Trace Analysis, Risk Detection and Revision Assistance
yushui2022/multigent
Multigent is an AI-native collaboration platform that brings humans and autonomous agents together. Manage teams, projects, workflows, tools, and shared context through a unified control plane designed for the future of intelligent work.
yushui2022/ORG2
ORG-2 turns AI agents into persistent, observable colleagues for local-first development.
yushui2022/Mini-Model-Harness
yushui2022/Agent-Group
yushui2022/openclaw_town
本项目叫openclaw_town,龙虾小镇:这是一个基于两年前开源项目斯坦福小镇的二次开发,可以让大家的openclaw进入里面游玩,且只有openclaw可以进去,openclaw拥有邀请码以后,就可以注册人物,通过json格式的命令控制人物活动,行走与聊天,后端会返回给openclaw周围的环境,所处坐标,附件有什么事物(:如厕所,花园,田地,或者人物坐标)具体细节已经归纳好文档,可以让openclaw阅读以后继续操作,人类智能在web端观看(http://120.26.179.165/)他们在干什么
yushui2022/research-social-public-data-skill
yushui2022/How-to-learn-agent
yushui2022/Brick-Vault
WIP - do not attempt to install for now.
yushui2022/rank
yushui2022/AI-SocialOps
yushui2022/defending-code-reference-harness
Skills for threat modeling, scanning, triage, patching, plus an autonomous scanning harness you can /customize
yushui2022/Cutix
yushui2022/Orlando-Agent-Notes
These are my study & interview notes.
yushui2022/one-api
LLM API 管理 & 分发系统，支持 OpenAI、Azure、Anthropic Claude、Google Gemini、DeepSeek、字节豆包、ChatGLM、文心一言、讯飞星火、通义千问、360 智脑、腾讯混元等主流模型，统一 API 适配，可用于 key 管理与二次分发。单可执行文件，提供 Docker 镜像，一键部署，开箱即用。LLM API management & key redistribution system, unifying multiple providers under a single API. Single binary, Docker-ready, with an English UI.
yushui2022/CLIProxyAPI
Wrap Gemini CLI, Antigravity, ChatGPT Codex, Claude Code, Grok Build as an OpenAI/Gemini/Claude/Codex compatible API service, allowing you to enjoy the free Gemini 3.1 Pro, GPT 5.5, Grok 4.3, Claude model through API
yushui2022/LLM-WAF
Zero-intrusion LLM firewall for OpenAI-compatible and Anthropic traffic, with streaming-safe scanning, PII redaction, and audit logs.