Skip to main content
Glama
claude-proxmox-controller

claude-proxmox-controller

run_command

Execute commands inside a running Proxmox VM through the QEMU guest agent and return the output, with optional user and stdin support.

Instructions

Run a program inside a running VM through the QEMU guest agent and get its output (for example ["systemctl", "status", "nginx"], or ["sh", "-c", "df -h"]). It runs as root (SYSTEM on Windows) unless user is given.

Tool call template

Replace each placeholder with a value matching the input schema. Only required parameters are included; add optional parameters as needed.

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {
      "host": "<host>",
      "vmid": "<vmid>",
      "command": "<command>",
      "reason": "<reason>"
    },
    "name": "run_command"
  }
}

Input Schema

NameRequiredDescriptionDefault
hostYesThe host's id, from list_hosts.
userNoLinux guests: run it as this user of the VM, from their home folder with their HOME — so what it makes is theirs, not root's.
vmidYesThe guest's ID (VMID).
inputNoText given to the program's standard input.
reasonYesWhy you are doing this, in a sentence. The user sees it in the approval window and in the log.
commandYes
timeout_sNo

Schema Changelog

Changes observed during successful MCP inspections.

  1. First observedv0.1.0

TDQS

A3.8/5.0
Behavior4/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Adds real behavioral context beyond the annotations: execution via the QEMU guest agent, default privilege of root/SYSTEM, and the privilege/ownership implications of supplying a user. It does not describe what the returned output looks like (stdout vs stderr, exit code) or that calls are gated by a user approval window, which the reason parameter hints at only in the schema. Note the tension between arbitrary root-level execution and destructiveHint=false, though the description never asserts non-destructiveness, so this is not a contradiction.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness5/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Two tight sentences, front-loaded with the action and transport, followed by the privilege caveat. The inline examples replace several sentences of explanation with no wasted wording.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness3/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For a high-impact arbitrary-execution tool with no output schema, the description should say something about what comes back (stdout, exit code, truncation) and about timeout_s behavior. The privilege and command-format coverage is good, but the return-value gap and the unexplained approval/reason flow leave an agent under-informed at call time.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters4/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema coverage is 71%, so most parameters are already documented, but the description adds the crucial argv semantics that the bare array schema does not: commands are passed as an argument vector and require ['sh','-c',...] for shell syntax. It also clarifies that user changes both the executing identity and HOME/ownership, which is more than the schema's phrasing.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose4/5

Does the description clearly state what the tool does and how it differs from similar tools?

States a specific verb and resource ('Run a program inside a running VM through the QEMU guest agent') plus the outcome ('get its output'), with two concrete command-format examples. The 'inside a running VM' scoping implicitly separates it from the host-level sibling host_command, though it never names that sibling.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines3/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

Implies the prerequisite (a running VM with a reachable guest agent) and tells the agent when the user parameter matters ('runs as root (SYSTEM on Windows) unless user is given'). However it never says when to prefer this over host_command or the backup/guest-management siblings, nor what to do if the guest agent is unavailable.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.