# Quick Start Guide
Get up and running with Document Parser MCP in minutes!
## Installation
### 1. Setup Environment
```bash
# Clone repository
git clone <repository-url>
cd document-parser-mcp
# Create virtual environment
python -m venv venv
# Activate virtual environment
source venv/bin/activate # macOS/Linux
# venv\Scripts\activate # Windows
# Install dependencies
pip install -r requirements.txt
pip install docling
```
### 2. Optional: Enhanced Features
```bash
# For Apple Silicon MLX acceleration
pip install docling[mlx]
# For OCR support
pip install easyocr
# For tesseract OCR (macOS)
brew install tesseract
```
## First Run
### Test the Installation
```bash
# Start the server
python -m document_parser
```
The server will start and listen for MCP connections via stdio.
## Integration with Claude Desktop
### 1. Configure Claude Desktop
Edit your Claude Desktop config file:
- **macOS**: `~/Library/Application Support/Claude/claude_desktop_config.json`
- **Windows**: `%APPDATA%\Claude\claude_desktop_config.json`
Add this configuration:
```json
{
"mcpServers": {
"document-parser": {
"command": "python",
"args": ["-m", "document_parser"],
"cwd": "/absolute/path/to/document-parser-mcp"
}
}
}
```
**Important**: Replace `/absolute/path/to/document-parser-mcp` with the actual path!
### 2. Restart Claude Desktop
Close and reopen Claude Desktop completely.
### 3. Test the Integration
In a new Claude conversation, try:
```
Can you parse this document for me?
```
Then upload a PDF, DOCX, or image file. The document parser should automatically process it!
## Basic Usage Examples
### Parse a Local PDF
```json
{
"name": "parse_document",
"arguments": {
"source": "/path/to/document.pdf"
}
}
```
### Parse from URL
```json
{
"name": "parse_document",
"arguments": {
"source": "https://arxiv.org/pdf/2408.09869"
}
}
```
### Parse with Specific Pipeline
```json
{
"name": "parse_document",
"arguments": {
"source": "scan.pdf",
"pipeline": "vlm",
"options": {
"ocr_enabled": true
}
}
}
```
### Advanced Options
```json
{
"name": "parse_document_advanced",
"arguments": {
"source": "document.pdf",
"pipeline": "standard",
"ocr_enabled": true,
"table_accuracy_mode": "accurate",
"pdf_backend": "dlparse_v4"
}
}
```
## Configuration
### Create Custom Config
Copy the default configuration:
```bash
cp config.yaml my_config.yaml
```
Edit `my_config.yaml` to customize settings.
Run with custom config:
```bash
python -m document_parser --config my_config.yaml
```
### Key Settings
```yaml
# Server settings
server:
max_concurrent_jobs: 3 # Adjust based on your RAM
# Processing settings
processing:
default_pipeline: standard
# OCR settings
ocr:
engine: easyocr
languages: [eng] # Add more: [eng, spa, fra]
# Performance
performance:
max_memory_gb: 6 # Adjust for your system
enable_mlx_acceleration: true # Apple Silicon only
```
## Troubleshooting
### Issue: "ModuleNotFoundError: No module named 'docling'"
**Solution:**
```bash
pip install docling
```
### Issue: Server doesn't appear in Claude Desktop
**Solutions:**
1. Check the config file path is correct
2. Ensure the `cwd` path is absolute, not relative
3. Restart Claude Desktop completely
4. Check server starts without errors: `python -m document_parser`
### Issue: Out of memory errors
**Solution:**
Reduce `max_concurrent_jobs` and `max_memory_gb` in config:
```yaml
server:
max_concurrent_jobs: 1
processing:
performance:
max_memory_gb: 4
```
### Issue: Poor OCR quality
**Solution:**
1. Specify correct language:
```yaml
processing:
ocr:
languages: [eng, spa] # Add your languages
```
2. Try different OCR engine:
```yaml
processing:
ocr:
engine: tesseract # or easyocr
```
## Next Steps
- Read the [full README](README.md) for all features
- Explore [configuration options](docs/CONFIGURATION.md)
- Check [supported formats](docs/FORMATS.md)
- Run the [test suite](tests/)
## Getting Help
- Check [Troubleshooting](#troubleshooting) section above
- Review [Docling documentation](https://docling-project.github.io/docling/)
- Open an issue on GitHub
Happy parsing! 🚀
