mirecekd/awsblogs-mcp

AWS Blogs MCP Server

MCP server for working with AWS Blog and News articles from api.aws-news.com.

Description

This MCP server provides tools for retrieving and filtering AWS blog articles and news from all AWS categories. The server is built using the FastMCP framework, is fully asynchronous (async/await), and supports SSE transport on port 8807. Data is fetched from the public API and processed using aiohttp and BeautifulSoup.

Architecture

• Entrypoint: main_sse.py – sets up the environment and starts the server.
• Server: src/awsblogs_mcp_server/server_sse.py – defines all MCP tools, handles argument parsing, and runs the FastMCP server.
• Data Processor: src/awsblogs_mcp_server/data_processor.py – handles API communication, filtering, caching, and HTML parsing.
• Caching: In-memory cache (5 minutes) for article lists (except search queries, which always fetch fresh data).
• Validation: Input parameters are validated for type, presence, and (where relevant) format. Only AWS articles (aws.amazon.com) are supported for full content download.

Available MCP Tools

1. get_todays_posts

Gets articles published today.

Parameters:

• post_type: "News", "Blog", or "Both" (default)
• limit: Maximum number of articles (default 20)

2. get_posts_by_date

Gets articles from a specific date range.

Parameters:

• from_date: From date in YYYY-MM-DD format (optional)
• to_date: To date in YYYY-MM-DD format (optional)
• days_back: Number of days back from today (alternative to from_date)
• post_type: "News", "Blog", or "Both" (default)
• limit: Maximum number of articles (default 50)

3. get_posts_by_category

Gets articles from a specific category.

Parameters:

• category: Category name (e.g., "Big Data", "Machine Learning", "Industries")
• post_type: "News", "Blog", or "Both" (default)
• days_back: Number of days back from today (default 30)
• limit: Maximum number of articles (default 30)

4. search_posts

Searches articles by text query using the API's search functionality (searches in title, URL, and slug).

Parameters:

• query: Search query (searches in title, URL, and slug)
• post_type: "News", "Blog", or "Both" (default)
• days_back: Number of days back from today (default 90)
• limit: Maximum number of articles (default 25)

Note: Search queries always fetch fresh data and do not use cache.

5. get_categories

Gets a list of all available article categories.

6. get_latest_posts

Gets the latest articles.

Parameters:

• post_type: "News", "Blog", or "Both" (default)
• limit: Maximum number of articles (default 20)
• days_back: Number of days back from today (default 7)

7. get_popular_posts

Gets popular articles (marked as popular=true).

Parameters:

• post_type: "News", "Blog", or "Both" (default)
• days_back: Number of days back from today (default 30)
• limit: Maximum number of articles (default 15)

8. get_article_content

Downloads full article content from a given URL.

Parameters:

• url: Article URL (must be from aws.amazon.com)

Return values:

• title: Article title
• content: Full article text
• description: Meta description
• author: Author (if available)
• published_date: Publication date
• content_length: Content length in characters
• word_count: Word count

Note: Only AWS articles (aws.amazon.com) are supported for full content download.

Available Categories

The server supports filtering by these categories:

• Architecture
• AWS Cloud Operations
• AWS for Games
• AWS Insights
• AWS Marketplace
• AWS News
• AWS Partner Network
• AWS Smart Business
• Big Data
• Business Intelligence
• Business Productivity
• Cloud Enterprise Strategy
• Cloud Financial Management
• Compute
• Contact Center
• Containers
• Database
• Desktop & Application Streaming
• Developer Tools
• DevOps & Developer Productivity
• Front-End Web & Mobile
• HPC
• IBM and Red Hat
• Industries
• Integration & Automation
• Internet of Things
• Machine Learning
• Media
• Messaging & Targeting
• Microsoft Workloads on AWS
• Migration and Modernization
• .NET on AWS
• Networking & Content Delivery
• Open Source
• Public Sector
• Quantum Computing
• SAP
• Security
• Spatial Computing
• Startups
• Storage
• Supply Chain & Logistics
• Training & Certification

Caching

• Article lists are cached in memory for 5 minutes (except for search queries).
• Search queries always fetch fresh data from the API.
• Cache is invalidated automatically after timeout or when parameters change.

Installation and Running

Docker (recommended)

1. Run from GitHub Container Registry:

   docker run -p 8807:8807 ghcr.io/mirecekd/awsblogs-mcp
   

2. Build Docker image locally:

   chmod +x build.sh
   ./build.sh
   

3. Run using docker-compose:

   docker-compose up awsblogs-mcp-sse
   

4. Or local Docker build:

   docker run -p 8807:8807 awsblogs-mcp-server:sse
   

Local Development

1. Install dependencies:

   pip install -e .
   

2. Run server:

   python main_sse.py --host 0.0.0.0 --port 8807
   

MCP Client Configuration

To use in an MCP client (e.g., Cline), add to configuration:

{
  "mcpServers": {
    "awsblogs": {
      "type": "sse",
      "url": "http://localhost:8807/sse/"
    }
  }
}

API Endpoint

The server uses the public API: https://api.aws-news.com/articles

Features

• ✅ SSE transport on port 8807
• ✅ Filter by article type (News/Blog)
• ✅ Filter by category
• ✅ Date filtering (date range, days back)
• ✅ Text search (API-powered)
• ✅ Cache mechanism (5 minutes, except for search)
• ✅ Structured responses
• ✅ Docker support
• ✅ Asynchronous architecture (aiohttp, FastMCP)
• ✅ HTML parsing with BeautifulSoup

Project Structure

awsblogs-mcp/
├── src/awsblogs_mcp_server/
│   ├── __init__.py
│   ├── server_sse.py          # SSE MCP server (tools, server logic)
│   └── data_processor.py      # API client, filtering, caching, HTML parsing
├── main_sse.py               # SSE entry point
├── Dockerfile.sse            # Docker image for SSE
├── docker-compose.yml        # Docker Compose configuration
├── build.sh                  # Build script
├── pyproject.toml            # Python project configuration
├── README.md                 # This file
└── assets/
    ├── aws_blogs.png
    └── n8n-workflow.json

Development

For development we recommend:

1. Fork the repository
2. Create a new branch for the feature
3. Test locally using python main_sse.py
4. Test Docker build using ./build.sh
5. Create a pull request

License

MIT License - see LICENSE file.

Author

Miroslav Dvořák (mirecekd@gmail.com)

Usage Examples

N8N Workflow Integration

This MCP server can be easily integrated with N8N workflows for automated AWS news processing and notifications.

The screenshot above shows a real-world N8N workflow where the AWS Blogs MCP server is used as a tool within an AI Agent. The workflow consists of:

1. Chat Trigger - Receives user messages
2. AI Agent - Processes requests using Azure OpenAI
3. AWS Blogs MCP Tool - Fetches AWS articles based on user queries
4. Memory Buffer - Maintains conversation context

Example N8N Workflow: You can find a complete N8N workflow configuration in

How it works in practice:

• User asks: "Show me latest AWS articles about Machine Learning"
• AI Agent uses the MCP server to call search_posts with query "Machine Learning"
• Server returns relevant AWS ML articles with titles, URLs, and summaries
• AI Agent formats the response for the user with clickable links
• Conversation context is maintained for follow-up questions

This integration allows for natural language interactions with AWS content, making it easy to stay updated with the latest AWS developments through conversational AI.

Inspiration

Project inspired by aws-news-mcp-server by jritsema.