stack-over-mcp

ygfest/stack-over-mcp

3.2

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

An MCP server that integrates with Stack Overflow to provide programming solutions, reducing hallucinations in LLM responses.

Tools
3
Resources
0
Prompts
0

Stack Over MCP

An MCP (Model Context Protocol) server that provides LLMs with tools to search Stack Overflow for programming solutions, helping ground responses in real-world code examples and reduce hallucinations.

Features

  • Smart Search: Search Stack Overflow questions with customizable filters
  • Detailed Question Analysis: Get complete question details with all answers
  • Tag-Based Discovery: Find popular solutions by programming language/framework
  • Real-time Results: Direct access to Stack Overflow's API
  • Rich Metadata: Scores, view counts, acceptance status, and more

Installation

  1. Clone and setup the project:

    cd stack-over-mcp
    python -m venv venv
    # On Windows:
    venv\Scripts\activate
    # On macOS/Linux:
    source venv/bin/activate
    
  2. Install dependencies:

    pip install -r requirements.txt
    

Usage

Running the MCP Server

python main.py

The server will start and provide three main tools:

1. search_stackoverflow

Search Stack Overflow questions with flexible filtering:

# Basic search
search_stackoverflow("python list comprehension")

# Advanced search with filters
search_stackoverflow(
    query="react hooks error",
    limit=5,
    sort="votes",
    tags=["javascript", "react"],
    accepted_only=True
)

Parameters:

  • query: Search terms
  • limit: Max results (1-100, default: 10)
  • sort: "relevance", "activity", "votes", or "creation"
  • tags: Filter by programming tags
  • accepted_only: Only questions with accepted answers

2. get_question_details

Get complete information about a specific question:

# Get question with all answers
get_question_details(question_id=12345678)

# Get just question metadata
get_question_details(question_id=12345678, include_answers=False)

3. search_by_tags

Find popular questions by technology tags:

# Find top Python questions
search_by_tags(tags=["python"], sort="votes", min_score=10)

# Find recent React + TypeScript questions
search_by_tags(
    tags=["javascript", "react", "typescript"],
    sort="activity",
    limit=5
)

Integration with MCP Clients

This server is designed to work with MCP-compatible clients like Claude Desktop. Add it to your MCP configuration:

{
  "mcpServers": {
    "stack-over-mcp": {
      "command": "python",
      "args": ["path/to/stack-over-mcp/main.py"]
    }
  }
}

Use Cases

  • Debugging: Find solutions to specific error messages
  • Learning: Discover best practices and common patterns
  • Code Review: Validate approaches against community solutions
  • Architecture: Find proven solutions for complex problems
  • API Usage: Get real examples of library/framework usage

API Rate Limits

The Stack Overflow API has rate limits:

  • Unauthenticated: 300 requests per day per IP
  • Authenticated: 10,000 requests per day (requires API key)

For production use, consider implementing API key authentication in stack_api.py.

Project Structure

stack-over-mcp/
ā”œā”€ā”€ main.py          # FastMCP server with tool definitions
ā”œā”€ā”€ stack_api.py     # Stack Overflow API client
ā”œā”€ā”€ requirements.txt # Python dependencies
└── README.md        # This file

Contributing

Feel free to enhance the server with additional features:

  • API key authentication for higher rate limits
  • Caching for frequently requested questions
  • Advanced filtering options
  • Support for other Stack Exchange sites

License

This project is open source and available under the MIT License.