Skip to main content
MCP server monitors verify that Model Context Protocol servers are running, responsive, and returning valid tool listings. Use them to monitor AI tool infrastructure that your agents and applications depend on.
Define this in code. YAML format · Terraform

​
When to use MCP server monitors

  • AI tool infrastructure — verify MCP servers that power Cursor, Claude Desktop, or custom agents
  • Tool availability — confirm specific tools are listed and available
  • Protocol compliance — validate MCP protocol version and capabilities
  • Tool count drift — detect unexpected changes to the tool catalog

​
How it works

  1. DevHelm launches the MCP server using the configured command and arguments
  2. A connection is established over the MCP protocol
  3. Assertions evaluate the server’s response — capabilities, tool listings, protocol version
  4. Response time is measured from launch to successful initialization

​
Quick example

devhelm monitors create \
  --name "DevHelm MCP" \
  --type MCP_SERVER \
  --url uvx \
  --frequency 300 \
  --regions us-east

monitors:
  - name: DevHelm MCP
    type: MCP_SERVER
    config:
      command: uvx
      args: ["devhelm-mcp-server"]
      env:
        DEVHELM_API_TOKEN: ${DEVHELM_MCP_TOKEN}
    frequencySeconds: 300
    regions: [us-east]
    assertions:
      - type: mcp_connects
        severity: fail
      - type: mcp_tool_available
        toolName: list_monitors
        severity: fail
With the CLI, --url is the command that launches the MCP server. Arguments and environment variables aren’t settable via flags — use YAML config-as-code (config.args, config.env) or the API for the full configuration, as shown above.

​
Next steps

Configuration

Command, args, env, and all MCP assertions.

MCP Server SDK

The DevHelm MCP server for AI agents.