Working with documentation via MCP#

The MCP server lets you connect published K2 Cloud documentation to an AI client that supports the Model Context Protocol. The client will be able to search the documentation, read the pages it finds, and include source links in its response.

Connecting#

To connect, use the following parameters:

  • server address: https://mcp-docs.k2.cloud/mcp;

  • transport: Streamable HTTP.

  1. Open the MCP server settings in the AI client.

  2. Add a server with the Streamable HTTP transport.

  3. Specify the address https://mcp-docs.k2.cloud/mcp.

  4. Save the settings and enable the MCP server if it is not already enabled.

Parameter names depend on the client you use. If the client supports the mcpServers configuration, an entry may look as follows:

{
  "mcpServers": {
    "k2-docs": {
      "url": "https://mcp-docs.k2.cloud/mcp"
    }
  }
}

Use#

The MCP server provides documentation in Russian and English. After connecting, ask questions in natural language. If necessary, you can specify the documentation language and additional task conditions, for example:

  • “Explain how to create an instance. Include links to the sources.”

  • “Find recommendations for creating a high-availability PostgreSQL service.”

  • “Find the instructions for creating a VPN connection and list the main steps.”

Verifying the connection#

To check the server without an AI client, run the following request:

curl -fsS https://mcp-docs.k2.cloud/mcp \
  -H 'Content-Type: application/json' \
  -H 'Accept: application/json, text/event-stream' \
  --data '{"jsonrpc":"2.0","id":1,"method":"tools/list","params":{}}'

The response must include a list of tools. If the client does not connect, make sure the address ends with /mcp and that Streamable HTTP is selected as the transport.