Skip to main content
Glama

Run natural-language task on device

device_run_task

Have the on-device AI agent (Google's ARTEMIS) perform a task described in plain English, e.g. 'Open Settings, enable Developer options, and report the Android build number'. Waits for completion (up to min(wait_seconds, 25) seconds on this remote server) and returns the task with its steps and result; if it is still running, poll with task_status. profile 'flash' is fast for simple UI actions; 'pro' plans and verifies multi-step tasks. AI is included (no model key needed); optionally ?model_key=… on the connector URL uses your own key. One task per device at a time.

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": {
      "device_id": "<device_id>",
      "instruction": "<instruction>"
    },
    "name": "device_run_task"
  }
}

Input Schema

NameRequiredDescriptionDefault
profileNoflash
device_idYes
instructionYes
wait_secondsNo0 = return immediately with the queued task

Schema Changelog

Changes observed during successful MCP inspections.

  1. First observed

TDQS

A4.2/5.0
Behavior4/5

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

With no annotations provided, the description carries the behavioral burden well: it explains the wait cap on the remote server, the return of task steps and result, polling via task_status, model-key handling, and the one-task-per-device limit. It does not explicitly describe side effects or whether on-device actions are reversible, which keeps it from a 5.

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?

The description is front-loaded with the purpose, followed by operational details in a compact paragraph. Every sentence contributes useful context, including profile choice, wait behavior, polling, model-key handling, and concurrency.

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

Completeness4/5

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

For a complex task-execution tool with no output schema and no annotations, the description covers the key operational details: waiting, return value shape, polling fallback, profile semantics, and concurrency limits. It is not fully complete because it does not explain device_id requirements or side-effect expectations in more depth.

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

Parameters3/5

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

Schema description coverage is only 25%, so the description must compensate for undocumented parameters. It adds useful meaning for profile and wait behavior and gives an instruction example, but it leaves device_id largely unexplained beyond the one-task-per-device constraint.

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

Purpose5/5

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

The description gives a specific verb and resource: have the on-device AI agent ARTEMIS perform a plain-English task. It includes a concrete example and distinguishes the tool from task_status by explaining that task_status is for polling when execution is still running.

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

Usage Guidelines4/5

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

It clearly indicates when to use the tool and how to choose between the flash and pro profiles, and it names task_status as the polling alternative. It also states the one-task-per-device constraint, though it does not explicitly contrast the tool with direct-input siblings such as device_input.

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

Try in Browser

Glama MCP Gateway

Add one secure layer between your agents and this server.

Resources