Enables searching, listing, and reading files directly from Google Drive with specialized tools for analyzing High Performance Computing (HPC) log files, identifying errors, warnings, and resource issues.
Google Drive HPC Log Analyzer
A Model Context Protocol (MCP) server that brings Google Drive integration to Claude Desktop, with specialized tools for analyzing High Performance Computing (HPC) log files.
Overview
This MCP server enables Claude to directly access your Google Drive files and intelligently analyze HPC job logs, helping you quickly identify errors, warnings, and resource issues without leaving the Claude Desktop interface.
Features
Google Drive Integration: Search, list, and read files directly from Google Drive
HPC Log Analysis: Automatic detection of common HPC issues:
Out of memory (OOM) errors
Job timeouts and time limit issues
Failed jobs and fatal errors
Resource allocation problems
Smart Search: Quickly find log files (.log, .out, .err) across your Drive
Read-Only Access: Secure, read-only access to your Google Drive
Persistent Authentication: One-time OAuth setup with automatic token refresh
Prerequisites
Python 3.7+
Claude Desktop installed
Google Cloud account (free tier works fine)
pip for package management
Quick Start
1. Clone the Repository
2. Install Dependencies
3. Set Up Google Cloud Credentials
Follow the detailed SETUP_GUIDE.md to:
Create a Google Cloud project
Enable the Google Drive API
Download OAuth credentials as
credentials.json
4. Configure Claude Desktop
Run the automatic configuration script:
Or manually edit ~/Library/Application Support/Claude/claude_desktop_config.json
:
5. Authenticate
Run the server once to authenticate:
This will open your browser for Google OAuth authentication.
6. Restart Claude Desktop
Quit and restart Claude Desktop. Look for the š icon to confirm the MCP server is loaded.
Usage Examples
Once set up, you can ask Claude things like:
Available Tools
The MCP server provides these tools to Claude:
list_drive_files
Search and list files in Google Drive with optional queries and folder filtering.
Parameters:
query
(str): Search query (e.g., "name contains 'log'")max_results
(int): Maximum number of results (default: 10)folder_id
(str, optional): Specific folder ID to search within
read_drive_file
Read the contents of a file from Google Drive.
Parameters:
file_id
(str): Google Drive file ID
analyze_hpc_log
Analyze an HPC log file and provide insights.
Parameters:
file_id
(str): Google Drive file ID of the log file
Returns:
File metadata (name, size, line count)
List of errors found
List of warnings found
Actionable suggestions for fixing issues
search_hpc_logs
Search specifically for HPC log files (.log, .out, .err extensions).
Parameters:
search_term
(str): Search term (default: "*.log")folder_id
(str, optional): Folder to search withinmax_results
(int): Maximum results (default: 20)
Security
Read-Only Access: The server only requests read permissions to Google Drive
Local Credentials: All authentication tokens are stored locally on your machine
No Third-Party Access: Your Google Drive data is accessed directly by Claude running locally
Gitignored Secrets:
credentials.json
andtoken.pickle
are excluded from version control
Important: Never commit credentials.json
or token.pickle
to public repositories.
File Structure
Troubleshooting
"credentials.json not found"
Make sure you've downloaded the OAuth client credentials from Google Cloud Console and placed them in the project directory.
"Module not found" errors
Install dependencies: pip install -r requirements.txt
MCP tools don't appear in Claude
Verify
claude_desktop_config.json
syntax is valid JSONCheck that the file path is absolute, not relative
Restart Claude Desktop completely
Authentication issues
Delete token.pickle
and run python server.py
to re-authenticate.
See SETUP_GUIDE.md for more detailed troubleshooting.
Development
Testing Authentication
Running the Server Directly
Contributing
Contributions are welcome! Feel free to:
Report bugs
Suggest new features
Submit pull requests
License
MIT License - feel free to use and modify as needed.
Resources
Acknowledgments
Built with FastMCP - a Python framework for creating MCP servers.
This server cannot be installed
local-only server
The server can only run on the client's local machine because it depends on local resources.
Enables Claude to directly access Google Drive files and analyze High Performance Computing (HPC) log files. Automatically detects common HPC issues like memory errors, job timeouts, and resource allocation problems.