chat-with-mcp-tools
Hackathon project showing how to connect a gradio chatbot to a few different mcp servers
GitHubスター
0
ユーザー評価
未評価
お気に入り
0
閲覧数
12
フォーク
0
イシュー
0
title: Chat With Mcptools
emoji: 🤖
colorFrom: blue
colorTo: indigo
sdk: gradio
sdk_version: 5.33.0
app_file: chat.py
pinned: false
license: cc-by-nc-2.0
python_version : 3.12
tags:
- agent-demo-trackgit
This chatbot demonstrates how MCP servers can help with data science tasks. Currently, it allows for connecting to
- HuggingFace MCP to search across models, spaces, datasets, and papers etc
- Kaggle MCP space-- a more limited in scope MCP
that provides read access to Kaggle datasets, competitions and models. It relys on Kaggle's API
It is worth noting that these MCPs really do rely on the underlying services, e.g. the Huggingface MCP is much more
informative as it can handle semantic search, whereas the kaggle api is very strict in what inputs it accepts, and
often any sort of typos etc will result in no results being returned.
At the moment, this demo comes with a supplied Anthropic key, so to run it yourself you will have to set it through
env vars (see the .env.example file for what you need to set) You can do this via the space's settings, or if
running locally, through a .env file