home / mcp / reunemacacada mcp server

Reunemacacada MCP Server

Provides a Python-based MCP server that generates learning paths from diverse resources with caching, TF-IDF relevance filtering, and YouTube integration.

Installation
Add the following to your MCP client configuration file.

Configuration

View docs
{
  "mcpServers": {
    "cabrit0-mcp_server_reunemacacada": {
      "url": "https://reunemacacada.onrender.com"
    }
  }
}

You can generate structured Master Content Plans (MCPs) for any topic by running a server that gathers resources, organizes them into learning paths, and returns a standardized structure that your client applications can consume. This server supports multi-language outputs, efficient caching, asynchronous task handling, and TF-IDF relevance filtering to prioritize the most relevant resources for your topic.

How to use

Start by choosing a topic you want to build an MCP for. Use an MCP client to request a new learning path, either synchronously or asynchronously. You can specify the number of nodes, depth of the tree, and language preference to tailor the learning path. The server will fetch resources from diverse sources, filter them by relevance, and assemble them into a coherent sequence. You can also opt to include YouTube videos and categories to narrow the content to specific domains.

How to install

Prerequisites: Ensure you have Python 3.9+ and Node.js installed on your system. You will also need npm for managing JavaScript dependencies and a modern browser for headless scraping when needed.

Then follow these steps to set up and run the MCP server locally:

# 1) Clone the project
git clone https://github.com/yourusername/mcp_server.git
cd mcp_server

# 2) Create and activate a Python virtual environment
python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

# 3) Install Python dependencies
pip install -r requirements.txt

# 4) Install Node.js dependencies for optimized scraping
npm install

# 5) Ensure Chrome/Chromium is installed for Pyppeteer
# (follow your OS-specific instructions to install Chrome/Chromium)

Additional notes for installation and startup

If you prefer running the server in development mode, you can start the API server with a hot-reload option so changes are picked up immediately.

uvicorn main:app --reload --host 0.0.0.0 --port 8000

Additional sections

Configuration and endpoints are designed to be intuitive and developer-friendly. The server exposes a health check endpoint to verify availability and a suite of MCP-related endpoints to generate, monitor, and manage MCPs. If you want to clear cached results to force fresh data, you can trigger a cache clear with a pattern that matches the keys you want to remove.

Security and maintenance notes

Keep dependencies up to date and limit access to the MCP endpoints in production environments. Use environment isolation (virtual environments for Python and separate runtime environments for Node.js) to minimize interference between processes. Monitor resource usage during web scraping to avoid timeouts and maintain responsive performance.

Troubleshooting tips

If requests appear slow, check the cache status and verify that commonly requested topics are cached. For issues with headless scraping, ensure the Puppeteer instance pool is functioning and that Chrome/Chromium is accessible. Review logs for any rate limiting from external resources and adjust the TF-IDF filtering or scraping strategy accordingly.

Examples of usage patterns

Generate an MCP for a topic in Portuguese with a specific node count and language preference. If you want a broader topic tree but with a compact width, adjust min_width and max_height to balance depth and breadth.

Available tools

generate_mcp

Creates an MCP from a given topic, producing a structured learning path with configurable resources, nodes, and depth. Uses web search, TF-IDF filtering, and category detection.

generate_mcp_async

Initiates asynchronous MCP generation so long-running tasks run in the background while you continue using the interface.

status

Checks the status of an asynchronous MCP generation task and returns progress updates.

clear_cache

Clears cached MCP results based on a provided pattern to refresh data.

cache_stats

Provides statistics about the cache, including hit/mail ratios and domain-specific caches.