学习 learn-claude-code 后的 PR Review Agent 实践项目：构建 Debate Council 多 Agent 审查、Tool Calling、结构化报告与 AI Judge 评估。

Real-time, AI-judged competitive debate platform. Engage in structured 1v1 intellectual battles scored by Gemini AI on Logic, Facts, and Relevance.

An AI system that evaluates hackathon projects like a real judge.

AI Arbitration Platform with Python & LangChain

MIT UiPath AgentHack adapter/demo for AI Judge agentic test governance

The core automation engine for OpenSpark. This background worker continuously scrapes the open internet for hackathon/project data, cleans it, runs it through an AI evaluation pipeline, stores unique ideas, and posts them to the OpenSpark ecosystem.

Evaluate German translation answers on Sharplingo with AI, using your own model, API, and prompt template.

PromptForge — privacy-first prompt engineering studio. Run one prompt across many OpenAI-compatible models in parallel (Xiaomi MiMo, OpenAI, DeepSeek, Groq, Ollama), compare side-by-side, and use one model as a judge. Next.js 16 + React 19 + TypeScript + Tailwind CSS 4.

🥊 Decide between 2-4 ideas with a structured tournament of independent fresh-context AI judges. Auditable, evidence-based verdicts. Claude Code skill.

AI-powered Prompt Engineering Lab with Meta-Prompting and Auto-Evaluation (GPT-5.2 Judge)

Evaluation Pipeline for "Ideas Recall" telegram bot

Thor — autonomous LLM evaluation system (LLMOps). Compares prompts/models with reproducible methodology, custom code, no external frameworks. Built for Sistema Liz municipal chatbot.

Public AI Judge audit demo: evidence, dissent, human signoff, and archived verdict packet.

一个基于阿里云百炼qwen-plus模型的智能评分系统，专门用于评估学生是否适合参加算法竞赛团队。系统支持处理真实的申请表数据，提供深度AI分析和四维度综合评分。

Real-time multiplayer Mao card game with LLM-powered judge and dynamic rules engine