MCP Server Integration

Model Context Protocol server for Claude Code and Cursor. Integrate Doclayer document processing directly in your AI coding assistant.

← Back to Integrations

MCP Server

Document processing for AI coding assistants

The Doclayer MCP server enables developers using Claude Code or Cursor to integrate document processing directly in their IDE. Upload documents, search content, create templates, and more - all through natural conversation.

View on PyPI GitHub Repository

Quick Start (5 minutes)

Install the MCP server

Install the Doclayer MCP server via pip or uvx:

# Via pip
pip install doclayer-mcp

# Or via uvx (recommended for MCP)
uvx doclayer-mcp

Configure your IDE

Add Doclayer to your MCP configuration file.

Get your API key

Restart your IDE

Restart Claude Code or Cursor to load the MCP server. Then ask your AI assistant to use the quickstart prompt.

IDE Configuration

Claude Code

~/.claude/claude_desktop_config.json

{
  "mcpServers": {
    "doclayer": {
      "command": "uvx",
      "args": ["doclayer-mcp"],
      "env": {
        "DOCLAYER_API_KEY": "your-api-key"
      }
    }
  }
}

Cursor

~/.cursor/mcp.json

{
  "mcpServers": {
    "doclayer": {
      "command": "uvx",
      "args": ["doclayer-mcp"]
    }
  }
}

Available Tools

The MCP server exposes 13 tools that your AI assistant can use to interact with Doclayer:

setup_auth

Configure API credentials

check_connection

Verify API connectivity

upload_document

Upload a document for processing

get_document_status

Check processing status

get_document_extractions

Retrieve extraction results

search_documents

Vector search across documents

list_templates

List available agent templates

get_template

Get template details

list_projects

List your projects

create_project

Create a new project

generate_api_key

Create new API key

create_custom_template

Scaffold a new template

validate_template

Validate template manifest

Built-in Prompts

Pre-configured prompts to help you get started quickly:

quickstart

5-minute integration guide

integration-advisor

Analyze your project and suggest integration

template-creator

Guide to create custom templates

troubleshoot

Diagnose common issues

best-practices

Integration best practices

Example: Processing an Invoice

Simply ask your AI assistant to process a document. The MCP server handles the rest.

You: Upload invoice.pdf and extract the key details

Claude: I'll upload and process that invoice for you.
[Uses upload_document tool]
[Uses get_document_status tool to wait for processing]
[Uses get_document_extractions tool]

Here are the extracted details:
- Vendor: Acme Corp
- Invoice #: INV-2024-001
- Amount: $1,234.56
- Due Date: 2024-02-15

Environment Variables

Variable	Description	Default
DOCLAYER_API_KEY	Your Doclayer API key	None
DOCLAYER_BASE_URL	API base URL	https://api.doclayer.ai
DOCLAYER_TENANT_ID	Tenant/organization ID	None
DOCLAYER_NO_KEYRING	Disable system keyring	false

Prerequisites

Python 3.11 or higher
Claude Code or Cursor with MCP support
Doclayer account with API key

Install from PyPI

pip install doclayer-mcp

Try n8n Instead

Visual workflow automation