This document outlines how to test the MCP Cloud Server using unit tests, MCP Inspect, and Claude Desktop.
- Python 3.8 or higher
- Environment variables set up in
.env
file - MCP Inspect installed (
pip install mcp-inspect
) - Claude Desktop or Claude API access
To run the automated unit tests:
python src/test_mcp_server.py
This will:
- Start the MCP server
- Run tests against the API endpoints
- Shut down the server when tests are complete
MCP Inspect is a utility to interact with MCP servers directly.
First, start the MCP server:
python src/main.py
Once the server is running, you can use MCP Inspect to:
- Check server information:
mcp-inspect info --url http://localhost:7008
- List available functions:
mcp-inspect functions --url http://localhost:7008
- Call a specific function (example with list_buckets):
mcp-inspect call --url http://localhost:7008 --function list_buckets --args '{"provider": "aws"}'
Claude Desktop can be used to test the integration with AI.
- Start the MCP server:
python src/main.py
- Connect Claude Desktop to your MCP server:
- Open Claude Desktop
- Go to Settings > MCP
- Add a new MCP connection
- Enter URL:
http://localhost:7008
- Name it:
mcp-cloud
- Click "Connect"
Try these prompts to test the cloud storage functionality:
-
Basic check:
Can you check what cloud providers are available on the MCP server?
-
List buckets:
Could you list all my cloud storage buckets?
-
Get specific bucket details:
Show me details about my bucket named [BUCKET_NAME]
- If tests fail with connection errors, ensure the server is running on port 7008
- Check environment variables if authentication fails
- Review server logs for more detailed error information