# MCP File Search Server
A Model Context Protocol (MCP) server that provides intelligent file search capabilities for local directories. This server can search by file type, filename patterns, and file content using natural language queries.
## Features
- 🔍 **Natural Language Search**: Use plain English to describe what files you're looking for
- 📁 **Multi-Type Search**: Search by file extension, filename keywords, and file content
- 🤖 **AI-Powered Parsing**: Uses OpenAI GPT to intelligently parse search requests
- 📄 **Multiple File Formats**: Supports PDF, Word docs, Excel, JSON, CSV, and text files
- ⚡ **Fast Search**: Efficient file system traversal with smart filtering
- 🎯 **Relevance Scoring**: Results ranked by relevance to your query
## Installation
1. **Install dependencies:**
```bash
uv sync
```
2. **Set up environment variables:**
```bash
cp .env.example .env
# Edit .env and add your OpenAI API key
```
3. **Run the setup script:**
```bash
python setup_mcp_server.py
```
## Usage
### As MCP Server
Add to your MCP client configuration:
```json
{
"mcpServers": {
"file-search": {
"command": "python",
"args": ["/path/to/mcp_file_search_server.py"],
"env": {}
}
}
}
```
### Available Tools
#### `search_files`
Search for files in a local directory using natural language.
**Parameters:**
- `folder_path` (required): Absolute path to search directory
- `search_prompt` (required): Natural language search description
- `max_results` (optional): Maximum results to return (default: 10)
**Examples:**
```json
{
"folder_path": "/Users/john/Documents",
"search_prompt": "pdf files about machine learning",
"max_results": 5
}
```
```json
{
"folder_path": "/Users/john/Projects",
"search_prompt": "python scripts with neural network code",
"max_results": 10
}
```
### Standalone Usage
You can also use the search functionality directly:
```python
from fastmcp_file_search import search_files
from models import SearchRequest
request = SearchRequest(
folder_path="/path/to/search",
search_prompt="find all PDF files about AI",
max_results=10
)
results = search_files(request)
for result in results:
print(f"Found: {result['file_name']}")
```
### Web UI
Run the Streamlit web interface:
```bash
streamlit run file_search_ui.py
```
## Supported File Types
- **Documents**: PDF, Word (.docx, .doc), Excel (.xlsx, .xls)
- **Data**: JSON, CSV
- **Code**: Python (.py), JavaScript (.js), HTML, CSS, XML
- **Text**: Plain text, Markdown (.md), YAML (.yml), etc.
## Search Examples
- `"pdf files about machine learning"`
- `"python scripts with neural network code"`
- `"excel spreadsheets containing budget data"`
- `"json configuration files"`
- `"word documents from last month"`
- `"text files with API documentation"`
## Configuration
### Environment Variables
- `OPENAI_API_KEY`: Your OpenAI API key (required)
- `OPENAI_ORG_ID`: Your OpenAI organization ID (optional)
### Search Behavior
- Uses AND logic by default (files must match all criteria)
- Searches file extensions, filenames, and content
- Excludes system directories (.git, .venv, __pycache__, etc.)
- Limits content search to first 50KB of each file
## Architecture
```
mcp_file_search_server.py # MCP server implementation
├── fastmcp_file_search.py # Main search orchestration
├── models.py # Data models
├── utils.py # LLM parsing and utilities
├── search_functions.py # Individual search functions
└── file_search_ui.py # Web interface
```
## Troubleshooting
1. **"Import mcp could not be resolved"**
- Install the MCP package: `pip install mcp`
2. **"LLM parsing failed"**
- Check your OpenAI API key in `.env`
- Verify internet connection
3. **"No files found"**
- Check folder path exists and is readable
- Try broader search terms
- Verify file types exist in target directory
## Project Structure
```
├── mcp_file_search_server.py # Main MCP server implementation
├── models.py # Pydantic data models
├── utils.py # LLM integration and utilities
├── search_functions.py # Individual search operations
├── fastmcp_file_search.py # Main search orchestration
├── file_search_ui.py # Streamlit web interface
├── test_official_client.py # Official MCP client test
├── test_mcp_client.py # JSON-RPC test client
├── mcp_config.json # MCP server configuration
├── pyproject.toml # Project dependencies
├── README.md # This file
└── USAGE_GUIDE.md # Detailed usage instructions
```
## Development
To extend the server:
1. Add new search functions in `search_functions.py`
2. Update the search orchestration in `fastmcp_file_search.py`
3. Add new tools to `mcp_file_search_server.py`
## License
MIT License - see LICENSE file for details.
