First, a brief overview of what this is for.
Model Context Protocol (MCP) is an open standard developed by Anthropic to standardize the interaction between large language models (LLMs) and external data, tools, and systems. Essentially, MCP enables AI models to “see” and “use” your local computer or external cloud services, significantly enhancing their usefulness.
Therefore, I propose an implementation of an MCP server that provides access to the SimpleOne server via the Table REST API.
Installation is straightforward: download the archive, extract it, carefully read the README.md, and start using it 
.
mcp-simpleone.zip (5.7 MB)
Below is a brief guide on how to use it with a practical example: accessing the server from qwen code. In principle, you can connect it to any IDE or agent. However, qwen code offers attractive free limits: up to 1,000 requests and unlimited tokens per day.
For the examples below, I’ll use Windows and the path c:\AI\mcp-simpleone\ where you extract the archive contents.
Now, configure and start the MCP server.
Navigate to the server folder.
-
Copy the template
.env.exampleto.env: -
Fill in the configuration:
SIMPLEONE_URL=https://your-instance.simpleone.ru
SIMPLEONE_API_KEY=your-api-key-here
NODE_TLS_REJECT_UNAUTHORIZED=0 # For self-signed certificates
LOG_LEVEL=trace # For maximum verbosity logs
Or, using Basic Auth:
SIMPLEONE_URL=https://your-instance.simpleone.ru
SIMPLEONE_BASIC_USER=your-username
SIMPLEONE_BASIC_PASSWORD=your-password
NODE_TLS_REJECT_UNAUTHORIZED=0 # For self-signed certificates
LOG_LEVEL=trace # For maximum verbosity logs
Start the server using SSE transport:
npm run start:sse
Leave the server running — this is required for testing purposes.
Open another console and verify the server is working:
Info (list of tools):
curl http://localhost:3000/info
You should see a response from the server.
Now create a folder for a test project and navigate into it. Launch qwen code:
qwen
On first launch, you’ll need to authenticate.
Now initialize the project:
/init
Configure the connection to the MCP server. You could consult the documentation, but we’ll make it simpler — ask qwen to configure it for us.
Enter the request:
Connect to the MCP server. The server is located at c:\AI\mcp-simpleone. Use SSE protocol. Save settings in the project configuration.
If you omit the last sentence, settings will be saved globally and MCP will be used by default — which we don’t want yet.
Qwen will automatically scan the server files and extract the necessary configuration data. You may receive clarifying questions.
Configuration completed.
Now restart qwen code and enter:
/quit
Upon exit, you’ll receive a session summary.
Return to the current session:
qwen --resume eca1b86d-bbf4-44f6-a880-b6b164c59fd8
Test the MCP server. Enter the command:
/mcp
and check which MCP connections are active.
Select “simpleone” and view available tools.
Exit the menu.
Now test functionality. For example, enter for testing:
Call table_read with tableName="itsm_incident" limit=5
Note: The AI demonstrated initiative.
Now make the example more complex. Enter:
Find employees whose last name starts with "Асосков". Display results as a table with fields: "Last Name, First Name, Login, Email".
Use the following additional information: the employees table is named "employee", and the last name column is "last_name".
After several attempts, the result was successfully found.
PS. The entire server code was generated by the qwen AI. Therefore, before using in production, a full code review is required.
PSS. The project is available at https://github.com/0aen0/mcp-simpleone
