README.md

metadata

title: ThinkSquare
emoji: ♔
colorFrom: blue
colorTo: green
sdk: gradio
sdk_version: 5.33.0
app_file: app.py
pinned: true
license: gpl-3.0
tags:
  - mcp-server-track
  - modal
  - sambanova
short_description: An AI chess assistant built as an MCP server.

ThinkSquare

Track: MCP Server / Tool
Tag: mcp-server-track
Hackathon: Hugging Face Agents MCP Hackathon
Space: ThinkSquare on Hugging Face

---

Overview

ThinkSquare is an AI chess assistant built as an MCP Server. It enables users to:

• Analyze and Annotate PGN Games with AI insights
• Play Chess interactively through any compatible MCP Client (e.g., Claude Desktop, Cline [VSCode extension])
• Suggest Strong Moves powered by LLMs and cloud infrastructure

---

Tech Stack

• Gradio
  • UI + MCP Server interface
• Modal
  • For cloud-based compute [This does the heavy lifting by providing on-demand compute for both play_chess and annotate functions.]
• SambaNova
  • For LLM inference. [ThinkSquare uses the Meta-Llama-3.3-70B-Instruct model]
• Stockfish
  • As the core chess engine

---

How to Use

You can use any compatible MCP Client to connect to ThinkSquare, such as:

• Claude Desktop
• Cline
• Cursor
• Or even another Gradio MCP Client Space

ThinkSquare will automatically respond to tool calls like:
play_chess, suggest_move, annotate_pgn

---

Demo Video

ThinkSquare Annotation App as MCP Server
ThinkSquare Play Chess App as MCP Server

---

Author

Falguni Das Shuvo
Engineer · Researcher Focuses: Reinforcement Learning · GenAI · Optimization
🔗 GitHub: @shuvoxcd01
🌐 Hugging Face: @Falguni

---

Tags

mcp-server-track modal sambanova