secure_run_command

What secure_run_command does on Code MCP Server

AI agents invoke secure_run_command to trigger actions in Code MCP Server. What it does depends on the arguments the agent supplies, and its effects often reach beyond the immediate call: builds kicked off, notifications sent, workflows started.

Why secure_run_command is rated High

This tool executes arbitrary commands (albeit restricted to an allowlist via 'add_allowed_command'). Execution of commands is inherently Execute-category, as command outcomes are argument-dependent and can have broad side effects on the system. While restricted by a whitelist mechanism, the tool still permits running shell commands with externally visible consequences.

From the tool's definition The tool description states it "Execute[s] a command" which directly indicates code/command execution. The phrase "with security restrictions (only allowed commands)" shows it runs external operations whose effects depend on arguments, fitting the Execute…

The rule that runs secure_run_command safely

PolicyLayer is an MCP gateway: it sits between your AI agents and Code MCP Server, and checks every tool call against a rule you set before the call runs. Nothing changes on the server itself. For secure_run_command, this is the rule to start with:

{
  "version": "1",
  "default": "deny",
  "tools": {
    "secure_run_command": {
      "limits": [
        {
          "counter": "secure_run_command_rate",
          "window": "minute",
          "max": 10,
          "scope": "grant"
        }
      ]
    }
  }
}

Questions about secure_run_command