CLDGeminiPDF

You run a Model Context Protocol (MCP) server that lets Claude Desktop analyze PDF documents by leveraging Gemini AI models. It provides seamless PDF access via a Filesystem MCP server and runs a dedicated Java-based analyzer to extract text and perform intelligent analysis, giving you flexible model choices and easy deployment.

How to use

You use this MCP server by pairing it with Claude Desktop through MCP configuration. The filesystem server gives you access to PDFs stored in a directory you specify, while the CLDGeminiPDF service runs the Gemini models to analyze those PDFs. You can pick from multiple Gemini models to balance speed and depth of analysis. Start an analysis by instructing Claude Desktop to locate a PDF file via the filesystem server and then request analysis through the CLDGeminiPDF server. You can also specify the model you want to use for the analysis to tailor results.

For best results, point Claude Desktop to a directory you control and ensure your API key for Gemini is configured. Use the available model options to optimize response quality and cost based on your needs.

How to install

Prerequisites you need in place before starting:

• Java 11 or higher
• Maven for building from source
• Google AI Studio API Key
• Claude Desktop application
• Filesystem MCP Server as a required dependency

Option 1: Download Pre-built JAR (Recommended)

# 1) Download the latest pre-built JAR
# 2) Save it to a convenient location on your system

Option 2: Build from Source

git clone <your-repository-url>
cd CLDGeminiPDF
mvn clean compile assembly:single

This build creates a JAR file with all dependencies in the target directory.

Get Your Gemini API Key

Visit Google AI Studio, sign in, click Get API Key, and copy the API key for use in environment variables.

Configuration and run details

Configure Claude Desktop to run both the filesystem server and the CLDGeminiPDF server. Use explicit paths to your PDFs and the JAR you built or downloaded.

{
  "mcpServers": {
    "filesystem": {
      "command": "npx",
      "args": [
        "-y",
        "@modelcontextprotocol/server-filesystem",
        "/path/to/your/documents"
      ]
    },
    "cldgeminipdf": {
      "command": "java",
      "args": [
        "-jar",
        "/path/to/CLDGeminiPDF.v1.0.0.jar"
      ],
      "env": {
        "GEMINI_API_KEY": "your_api_key_here",
        "GEMINI_MODEL": "gemini-2.0-flash"
      }
    }
  }
}

Notes for placement and paths you must customize: replace /path/to/your/documents with the directory containing your PDFs, and replace /path/to/CLDGeminiPDF.v1.0.0.jar with the actual path to the JAR file you downloaded or built.

Available Gemini models

You can choose from several Gemini models to suit your needs. The list includes latest high-performance previews and stable options such as gemini-2.5 and gemini-2.0, plus gemma-3 models. Availability depends on your API key and usage tier in Google AI Studio.

Usage examples

Analyze a PDF by full path: Please analyze this research paper: file:///Users/username/Documents/research_paper.pdf Focus on the methodology and conclusions.

Find a PDF by name in your Documents directory and analyze it: Find the "research_paper.pdf" file in the Documents directory and analyze it using Gemini.

Ask for available models: What Gemini models are available for PDF analysis?

Troubleshooting

Common issues you may encounter include missing API keys, incorrect file paths, or API errors. Ensure your Gemini API key is set in Claude Desktop, verify the PDF file path is accessible, and restart Claude Desktop after configuration changes.

If you see Gemini API rate limit errors, wait and retry, or choose a model with higher rate limits according to your plan.

Development

Building from source requires Maven and Java. The project uses Jackson for JSON processing, Apache PDFBox for PDF text extraction, and the MCP Java SDK for communications.

Available tools

pdf_analysis

Analyzes and extracts content from PDFs using Gemini models and returns insights.

model_selection

Choose from Gemini models to balance performance and cost for analysis.

filesystem_access

Access PDFs through the Filesystem MCP server to locate documents within a specified directory.