Read PDF As Text

This MCP server provides efficient PDF text extraction, search, and metadata retrieval for Claude Code. It lets you process large PDFs without flooding Claude's context, enabling focused analysis and faster results.

How to use

You will use a client that can load MCP servers to access three core capabilities: read-pdf, search-pdf, and pdf-metadata. Configure the server in your MCP client, then invoke the tools by asking for the action you want—reading text, searching within the document, or extracting metadata. Use these practical patterns to accomplish common tasks without loading the full PDF into Claude Code.

Read-pdf lets you pull the text from a PDF file with optional page ranges and text cleaning. You can decide whether you want metadata included in the output. Search-pdf lets you look for specific phrases and control whether the search is case-sensitive or matches whole words. pdf-metadata retrieves essential information like the title, author, and page count.

Typical workflows include: first fetch metadata to understand the document, then search for relevant terms, and finally read specific sections to extract the needed content. For example, you might read pages 10–15 with cleaned text when you know where the relevant material lies.

How to install

Prerequisites you need before installing: Node.js and npm should be available on your system.

Option A: Install from npm (recommended) install globally

npm install -g @fabriqa.ai/pdf-reader-mcp

After installation, the MCP server is available globally. You can configure it by running the provided setup steps.

Option B: Install from source

git clone https://github.com/hancengiz/read_pdf_as_text_mcp.git
cd read_pdf_as_text_mcp
npm install

If you install from source, you will use the included configuration script to register the MCP server system-wide.

Configuration and setup notes

Configure the MCP server in your Claude client using the npx-based command shown here. This enables the server to run locally and be reachable by your Claude Code workspace.

# Add the MCP server using Claude Code CLI
claude mcp add pdf-reader npx @fabriqa.ai/pdf-reader-mcp@latest

# Or use the convenience script to update configuration
npx @fabriqa.ai/pdf-reader-mcp/update-config.js

If you prefer manual configuration, add this block to your Claude configuration file (paths may vary by platform). This uses npx to run the MCP server from the globally installed package.

{
  "mcpServers": {
    "pdf-reader": {
      "command": "npx",
      "args": [
        "@fabriqa.ai/pdf-reader-mcp@latest"
      ]
    }
  }
}

Additional setup for Claude Desktop

On macOS, you can place the configuration in the Claude Desktop settings at the specified path for your platform. This ensures the MCP server is available when Claude Desktop starts.

# macOS example path
~/Library/Application Support/Claude/claude_desktop_config.json

{
  "mcpServers": {
    "pdf-reader": {
      "command": "npx",
      "args": [
        "@fabriqa.ai/pdf-reader-mcp@latest"
      ]
    }
  }
}

Usage reference for the tools

read-pdf extracts text from a PDF file with optional page ranges and text cleaning. You can include metadata in the output.

search-pdf provides text search within a PDF. You can enable case-sensitive search or whole-word matching.

pdf-metadata retrieves document metadata such as title, author, and page count.

Troubleshooting and tips

If the server does not appear in Claude Code, verify the MCP command is correctly configured and that Node.js is installed and in your PATH. Ensure dependencies are installed if you built from source and restart Claude Code.

PDFs without extractable text (such as scanned images) may yield limited results. Try enabling clean_text to improve formatting and readability.

Available tools

read-pdf

Extracts full text content from a PDF file with optional page ranges and an option to clean and normalize the text. Include metadata if requested.

search-pdf

Searches for a specific query within a PDF file, with optional case sensitivity and whole-word matching.

pdf-metadata

Retrieves detailed metadata from a PDF, including title, author, page count, and dates.