Cloud IoT

You can run an MCP (Model Context Protocol) server that exposes a defined OpenAPI specification as an MCP endpoint. This enables you to access the Cloud IoT API through MCP clients, with a focused server process you can start locally or remotely and configure via environment variables.

How to use

Start the server in a local process and connect your MCP client to either the remote HTTP endpoint or the local stdio interface. Use the HTTP option when you want to proxy or expose the OpenAPI spec at a remote URL. Use the stdio option when you want to run the server as a local process and interact through standard input/output streams. You can configure the server at runtime using environment variables to select config sources or inline JSON strings.

How to install

Prerequisites you need before installing the server are listed here so you can prepare your environment.

# Ensure you have Python 3.9+ installed
python --version

# Install dependencies (if you are not using a dev container)
pip install -e ".[dev]"

# Alternatively, install with uv
uv pip install --editable ".[dev]"

Run the server in stdio mode

To run the MCP server locally in stdio mode, use the following command. This starts the server as a Python process that communicates via standard input and output.

python mcp_server/main.py stdio

Configure environment variables

The server reads its configuration from environment variables. The key variables shown here let you point the server to a configuration source or provide inline configuration data.

CONFIG_PATH=/path/to/mcp_server/mcp_config.json
CONFIG='{"example": "value"}'
SECURITY=your_security_parameters_here

Notes on running tests and tooling

Development uses linting, formatting, static analysis, and tests to maintain code quality. Run the linting, formatting, and test commands as part of your workflow to ensure you have clean, reliable code.

Additional configuration and security notes

The server supports both HTTP and stdio connection methods. For HTTP, you point to a remote OpenAPI URL and let the server act as a proxy or gateway. For stdio, you start a local process and interact via standard streams. If you plan to expose sensitive APIs or keys, use the SECURITY variable to provide necessary credentials and limit access accordingly.

Building and publishing

When you are ready to package or publish your MCP server, use the build and publish steps defined for the project tooling to create distributables.

Example usage scenarios

• Connect an MCP client to the HTTP endpoint that uses a pre-defined OpenAPI URL to access Cloud IoT API functionality.

Available tools

lint

Check code quality and style using linting rules to enforce a consistent codebase.

format

Format source code to a standard style using a formatter.

staticAnalysis

Run static analysis tools (mypy, bandit, semgrep) to detect potential issues.

tests

Run test suites with coverage to ensure functionality and reliability.

precommit

Pre-commit hooks enforce checks before every commit to maintain quality.