AWS CLI MCP Server

# AWS CLI MCP Server [](https://github.com/ClockworkCubed/aws-cli-mcp/actions/workflows/test.yml) A Model Context Protocol (MCP) server that acts as a thin wrapper around the local AWS CLI command, providing AI assistants with secure access to AWS services through the familiar AWS CLI interface. ## Overview This MCP server allows AI assistants to interact with AWS services by proxying commands to the local AWS CLI installation. It provides a secure, controlled way for AI systems to manage AWS resources while leveraging existing AWS CLI configurations and credentials. ## Features - **Thin Wrapper**: Direct proxy to local AWS CLI commands - **Security**: Respects existing AWS CLI credentials and permissions - **Compatibility**: Works with all AWS CLI commands and services - **Configuration**: Uses existing AWS CLI profiles and configurations ## Prerequisites - AWS CLI installed and configured locally - Valid AWS credentials (via `aws configure`, IAM roles, or environment variables) ## Installation ### Prerequisites 1. **Python 3.10+** - Required for FastMCP compatibility 2. **AWS CLI** - Must be installed and configured locally ```bash # Install via homebrew (macOS) brew install awscli # Or via pip pip install awscli # Configure with your credentials aws configure ``` ### Install Dependencies ```bash # Clone the repository git clone <repository-url> cd aws-cli-mcp # Install dependencies pip install -e . ``` ## Usage ### Running the MCP Server Start the MCP server using Python: ```bash python aws_mcp.py ``` The server will start and log: ``` INFO Starting MCP server 'aws-cli-mcp' with transport 'stdio' ``` The server runs in stdio mode and communicates via the Model Context Protocol over standard input/output. ### Using with MCP Clients Once running, the server exposes one tool: - **`execute_aws_command_tool`** - Execute any AWS CLI command with structured JSON output #### Tool Parameters ```json { "args": ["service", "command", "--option", "value"] } ``` #### Example Usage ```json { "tool": "execute_aws_command_tool", "arguments": { "args": ["sts", "get-caller-identity"] } } ``` #### Response Format ```json { "stdout": "{\n \"UserId\": \"AIDACKCEVSQ6C2EXAMPLE\",\n \"Account\": \"123456789012\",\n \"Arn\": \"arn:aws:iam::123456789012:user/example\"\n}", "stderr": "", "exitCode": 0 } ``` ### Connecting to LLMs The server can be used with any MCP-compatible LLM client: 1. **Claude Desktop**: Add to your MCP configuration 2. **Custom MCP Clients**: Connect via stdio transport 3. **Development**: Use MCP SDKs to interact programmatically ### Error Handling The server provides helpful error messages: - **AWS CLI not installed**: Clear installation instructions - **Invalid commands**: Full AWS CLI error details preserved - **Credential issues**: AWS CLI error messages passed through - **JSON parsing**: Graceful handling of non-JSON responses ## Status ✅ **Fully Functional** - MCP server implementation complete with: - FastMCP protocol compliance - AWS CLI tool registration - Command-line server startup - Comprehensive error handling ## Security Considerations - The server only has access to AWS resources that the local AWS CLI can access - All AWS CLI security mechanisms (IAM policies, MFA, etc.) remain in effect - No additional authentication is required beyond existing AWS CLI setup