No description available

Agentic Loops Playground

toy agent-trading playground: simple market simulator, rule-based agents, and a minimal backtesting loop.

Interactive playground that teaches how Claude Code works under the hood - hooks, MCP servers, skills, agent loop, and more

Agent Playground — interactive AI agents built on FastAPI (Python) backend and React frontend, leveraging Pydantic for schema validation and OpenAI LLMs for agent reasoning and action execution. Demonstrates structured agent loops, tool integrations, and AI-driven workflows.

A modular playground for building and experimenting with AI Agents using LangChain, OpenAI, and tool integrations like Serper, DuckDuckGo, and math reasoning. Features include prompt design, reasoning chains, web-search agents, and autonomous decision loops.

A hands-on playground for **agentic AI** research: lightweight agents with tools, planning, memory, and evaluation loops — no heavy frameworks required. Everything is implemented in plain Python with a small set of dependencies, so you can read the code and tweak the ideas.

`infection-protocol-lab` is a full-stack TypeScript playground for simulating autonomous LLM agents under uncertainty. Agents talk in rounds, spread a hidden infection, form beliefs about one another, evolve their own strategies, and surface the entire loop through a live ReactFlow interface and a benchmark runner.

The project focuses on a tick-based battlefield where autonomous agents interact through an HTTP API, make decisions from game state, and compete over units, terrain, and resources. It is designed as a practical playground for experimenting with agent behavior, game rules, and API-driven strategy loops.

LoveTago is a Tinder-like matchmaking sandbox for AI bots. Agents can self-register, swipe, match, and chat via a simple web API. Built for fun and testing autonomous interaction loops under safe rate limits—basically a bot “reality show” playground.