mcp_explorer / README.md
alihmaou's picture
Update README.md
3ef33c3 verified

A newer version of the Gradio SDK is available: 5.45.0

Upgrade
metadata
title: Mcp Explorer
emoji: 
colorFrom: yellow
colorTo: pink
sdk: gradio
sdk_version: 5.33.0
app_file: app.py
pinned: false
short_description: A little app with a smolagent to explore and test MCP Tools
tags:
  - agent-demo-track
  - Agents-MCP-Hackathon

🚀 MCP Explorer

Project submitted for the Agents-MCP-Hackathon – June 2025
Explore, test, and debug any MCP-compatible server — in one simple interface.


🎥 Demo video

Watch the demo


🧩 What is MCP Explorer?

MCP Explorer is a Gradio-based tool built to help participants of the MCP Hackathon (and beyond) interact with the Model Context Protocol ecosystem.

It allows you to:

  • 🔍 Discover and visualize tools exposed by any MCP server
  • 🧪 Test real-time prompts using a smolagent powered by Qwen 32B
  • 🔄 Automatically refresh the list of tools from any endpoint
  • 🧠 Browse available tools in a clean interactive table
  • 🌍 Quickly switch between hackathon candidates and custom endpoints

🎯 Features

Feature Description
🔧 Live MCP querying Send a prompt and let the agent reason using the selected MCP tools
📊 Structured tool viewer Display each tool's name, description, and input schema in a readable way
🔀 Endpoint selector Toggle between hackathon submissions or custom MCP server URLs
Auto-refresh on load Automatically pulls tools on app startup
🗃 Tool index builder (bonus) Aggregates all MCP tools from the hackathon into a single parquet file

🧪 How to use it

  1. Clone or open the Space.
  2. Select a known MCP server or paste a custom one.
  3. Click “🔄 Refresh MCP tools list” to load the available tools.
  4. Type a natural language prompt — the agent will use the tools if applicable.

📦 Bonus: MCP Tool Index

In the src/bonus.py module, you'll find extracthackathontools()
a utility that scans all mcp-server tagged Spaces from the hackathon and builds a global index of available tools in hackathon_mcp_tools.parquet.


🧠 Credits

Built by Ali Hmaou
With gratitude to the Hugging Face team for the amazing infrastructure, API access, and documentation.


🛠 Tech stack

  • Python, Gradio
  • smolagents, Qwen 32B, pandas
  • MCP Protocol (over gradio sse endpoints)