mcp-local-assistant

Prakashbishal/mcp-local-assistant

3.2

If you are the rightful owner of mcp-local-assistant and would like to certify it and/or have it hosted online, please leave a comment on the right or send an email to dayong@mcphub.com.

MCP Local Assistant is a lightweight AI-powered local assistant designed to work with local files and datasets through the Model Context Protocol (MCP).

Tools
11
Resources
0
Prompts
0

MCP Local Assistant

A small Model Context Protocol (MCP) server that exposes useful tools for working with local files and datasets.

It is designed as a lightweight AI-powered local assistant: the tools can be called from MCP clients like the MCP Inspector, ChatGPT Desktop, or Claude Desktop.

Features

  • File utilities:

    • write_note(filename, text) – append text to a file in data/
    • list_data_files() – list files in data/
    • read_text_file(filename) – read file content
    • delete_file(filename) – delete a file in data/
    • search_in_file(filename, query) – search within one file
    • search_in_all_files(query) – search across all text files
    • summarize_text_file(filename, max_chars) – simple text summary

  • CSV helpers (no external dependencies):

    • describe_dataset(filename) – rows, columns, missing values
    • preview_dataset(filename, n) – first n rows as JSON-friendly dicts

  • Utility tools:

    • greet(name) – simple greeting
    • current_time() – current local time

All tools are typed and exposed via MCP so that an LLM can discover and call them.

Requirements

  • Python 3.10+
  • pip for installing dependencies

Install Python dependencies with:

pip install -r requirements.txt