Berlin Services MCP Server - Enhanced

A production-grade Model Context Protocol (MCP) server for Berlin city services (Dienstleistungen). It features resilient caching, advanced search, PDF form discovery, and filling capabilities.

Features

• Comprehensive Tools: Search, browse, and get detailed information on Berlin services.
• Form Management: Discover, analyze, and fill PDF forms for various services.
• Resilient Caching: Optimized for performance with intelligent fallback for offline use.
• Remote Sync: Supports both local and remote (Cloud) deployments.

Technical Architecture

1. Data Sourcing

The server pulls data from the official Berlin City Services API

• Live Fetching: Tools like search_services and get_service_details hit the live API.
• Resilient Caching: Implements a dual-layer cache (Memory + Disk) to ensure high performance and offline reliability.
• Smart Fallback: If the API is down and no cache exists, it falls back to a minimal set of core services to remain functional.

2. Advanced PDF Processing

Powered by PyMuPDF (fitz), the server provides sophisticated document handling:

• Field Extraction: Automatically detects fillable fields in PDF forms hosted on the city's servers.
• Intelligent Filling: Maps user data to PDF fields using smart matching and common translation fragments.
• Loop Protection: Includes a LoopProtector to prevent redundant tool executions and LLM loops.
• Visual Previews: Can render specific pages of a PDF to JPEG for a quick in-chat snapshot (useful for verification without leaving the chat).

3. Deployment Modes

• Local Mode: Optimized for use with Claude Desktop. In this mode, open_file_locally can be used to open filled PDFs directly with your system's default viewer.
• Remote/Cloud Mode: Designed for deployment on platforms like Fly.io or Railway. Replaces local file opening with secure synchronization and in-chat previews.

Prerequisites

• Python 3.11 or higher
• uv (recommended) or pip

Local Setup

Using uv (Recommended)

1. Install dependencies and sync the environment:

   uv sync
   

2. Run the server as a module:

   export PYTHONPATH=$PYTHONPATH:$(pwd)/src && uv run python -m berlin_mcp.main
   

   Note: If you've installed the package via uv sync, you can also just run:

   uv run berlin-mcp
   

Using pip

1. Install dependencies:

   pip install -r requirements.txt
   

2. Run the server:

   export PYTHONPATH=$PYTHONPATH:$(pwd)/src && python -m berlin_mcp.main
   

Claude Desktop Configuration

Add the following to your Claude Desktop configuration file. Replace /Users/mananshah/Desktop/Berlin-Services-MCP-Server with the actual absolute path to the folder.

macOS

File location: ~/Library/Application Support/Claude/claude_desktop_config.json

{
  "mcpServers": {
    "berlin-services": {
      "command": "uv",
      "args": [
        "run",
        "--directory",
        "/Users/YourUser/Desktop/Berlin-Services-MCP-Server",
        "python",
        "-m",
        "berlin_mcp.main"
      ],
      "env": {
        "PYTHONPATH": "src"
      }
    }
  }
}

Windows

File location: %APPDATA%\Claude\claude_desktop_config.json

{
  "mcpServers": {
    "berlin-services": {
      "command": "uv",
      "args": [
        "run",
        "--directory",
        "C:\\Users\\YourUser\\Desktop\\Berlin-Services-MCP-Server",
        "python",
        "-m",
        "berlin_mcp.main"
      ],
      "env": {
        "PYTHONPATH": "src"
      }
    }
  }
}

Tools and Resources

The server exposes the following tools to help you navigate and manage Berlin city services:

Tool	Description
search_services	Advanced search for Berlin services with paging and filters.
get_service_details	Get complete details about a service including forms and prerequisites.
get_service_forms	Get all forms associated with a service with detailed metadata.
search_forms	Search for forms across all services.
get_forms_by_type	Get all forms of a specific type across services.
get_service_prerequisites	Get all prerequisites and requirements for a service.
get_service_checklist	Get a comprehensive checklist for completing a service application.
analyze_form_for_filling	Download a PDF form and extract its fillable fields.
perform_form_filling	Fill a PDF form and save locally.
download_filled_form	Retrieve a filled PDF form.
get_form_visual_preview	Render a page of a filled PDF as an image.
open_file_locally	Open a file on the user's local system (Local setup only).
delete_filled_form	Manually delete a filled form from the server.
browse_services_by_category	Browse services by category.
find_online_services	Find all services available online.
get_service_locations	Get all locations for a service.
get_api_status	Get API and cache status diagnostics.
clear_cache	Clear the service cache and force refresh.

Security Note

PDF forms filled on remote servers should be downloaded and then deleted using the delete_filled_form tool to ensure data privacy.