Helmfile MCP Server
A FastMCP server for executing Helmfile commands through the Model Context Protocol (MCP). This server provides a standardized interface for managing Helmfile operations, allowing AI assistants to help with Helmfile deployments, configurations, and troubleshooting.

Key Features
- Command Execution: Execute any Helmfile command with proper validation and error handling
- Synchronization: Specialized tool for synchronizing Helmfile releases
- Async Operations: Asynchronous command execution for better performance
- Command Piping: Support for Unix pipe operations to filter and transform command output
- Progress Tracking: Real-time progress updates through MCP context
- Timeout Support: Configurable command timeouts to prevent hanging operations
- Structured Errors: Detailed error responses with proper error codes and messages
Installation
Prerequisites
- Python 3.11 or higher
- Helmfile installed and available in PATH
- Access to a Kubernetes cluster
Install from Source
- Clone the repository:
git clone [email protected]:snowsky/mcp-helmfile.git
cd mcp-helmfile
- Install dependencies:
uv pip install -e .
Claude Desktop App Configuration
To configure the Claude desktop app to work with the MCP servers, you can use the following settings in your claude_desktop_config.json file:
{
"mcpServers": {
"filesystem": {
"command": "npx",
"args": [
"-y",
"@modelcontextprotocol/server-filesystem",
"LOCAL_DIRECTORY_ALLOWED_ACCESSED_BY_CLAUDE_DESKTOP"
]
},
"helmfile": {
"command": "uv",
"args": [
"--directory",
"LOCAL_DIRECTORY_THIS_REPO",
"run",
"server.py"
]
}
}
}
Explanation of Configuration
-
filesystem: This configuration uses
npxto run the server for filesystem operations, specifying directories for Desktop and Downloads. -
helmfile: This configuration uses
uvto run the Helmfile server. The--directoryargument specifies the path to the MCP Helmfile project, andrun server.pyindicates the script to execute.
Use with Claude Desktop App
After adding this configuration, restart the Claude desktop app to ensure it picks up the new settings. You can then interact with the MCP servers as needed.
Copy helmfile.yaml file to the above directory LOCAL_DIRECTORY_ALLOWED_ACCESSED_BY_CLAUDE_DESKTOP.
For more information on troubleshooting and advanced configurations, please refer to the debugging documentation.
Run in python
Starting the Server
python -m mcp-helmfile.server
Available Tools
1. execute_helmfile
A general-purpose tool for executing any Helmfile command.
Parameters:
command: Complete Helmfile command to execute (including any pipes and flags)timeout: Maximum execution time in seconds (default: 300)ctx: Optional MCP context for request tracking
Example:
result = await execute_helmfile(
command="list",
timeout=60,
ctx=context
)
2. sync_helmfile
A specialized tool for synchronizing Helmfile releases.
Parameters:
helmfile_path: Path to the Helmfile configuration filenamespace: Optional namespace to targettimeout: Maximum execution time in seconds (default: 300)ctx: Optional MCP context for request tracking
Example:
result = await sync_helmfile(
helmfile_path="/path/to/helmfile.yaml",
namespace="production",
timeout=300,
ctx=context
)
Response Format
All tools return a dictionary with the following structure:
{
"status": "success" | "error",
"output": "Command output if successful",
"error": {
"code": "Error code",
"message": "Error message"
} # Only present if status is "error"
}
Development
Running Tests
pytest
Building and Publishing
- Build the package:
uv build
- Publish to PyPI:
uv publish
Security Considerations
- Commands are executed with proper validation
- Dangerous operations like apply/destroy require confirmation
- Environment-specific commands are validated against allowed environments
- Command timeouts prevent resource exhaustion
- Proper error handling and reporting
License
Contributing
Contributions are welcome! Please feel free to submit a Pull Request.
Recommend MCP Servers 💡
mcp-v8
MCP server that exposes a V8 JavaScript runtime as a tool for AI agents like Claude and Cursor. Supports persistent heap snapshots via S3 or local filesystem, and is ready for integration with modern AI development environments.
LaurieWired/GhidraMCP
MCP Server for Ghidra
mckinsey/vizro
A MCP server that helps create Vizro dashboards and charts with LLMs.
microsandbox/microsandbox
An MCP server providing secure execution environment for untrusted code through microVM isolation with instant startup
@first-to-fly/code-cleanup
A MCP server for cleaning up code files using Google's Generative AI, with stash backup and custom prompts.
@charlesmuchene/pref-editor-mcp-server
Edit Android app user preferences in real-time using natural language commands via MCP protocol