Skip to main content
Glama

check_system

Detect GPU hardware and verify Vulkan acceleration. Reports GPU name, VRAM, Vulkan binary status, and recommends optimal Whisper model for hardware.

Instructions

Detect GPU hardware and verify Vulkan acceleration is available. Reports GPU name, VRAM, whether the Vulkan binary is installed, and recommends the best Whisper model for your hardware. Run this if you want to confirm GPU acceleration is working or diagnose why it isn't.

Input Schema

TableJSON Schema
NameRequiredDescriptionDefault

No arguments

Implementation Reference

  • Registration of the 'check_system' tool in ListToolsRequestSchema handler. Defines the tool name, description (detect GPU hardware, VRAM, Vulkan status, recommended model), and empty inputSchema.
    {
      name: "check_system",
      description:
        "Detect GPU hardware and verify Vulkan acceleration is available. " +
        "Reports GPU name, VRAM, whether the Vulkan binary is installed, " +
        "and recommends the best Whisper model for your hardware. " +
        "Run this if you want to confirm GPU acceleration is working or diagnose why it isn't.",
      inputSchema: { type: "object", properties: {} },
    },
  • Handler for the 'check_system' tool in CallToolRequestSchema. Calls hasVulkanDll() to check for ggml-vulkan.dll, calls detectGpus() to get GPU info via wmic, formats output with GPU name, VRAM, recommended model, and Vulkan status.
    // check_system
    // -------------------------------------------------------------------------
    if (name === "check_system") {
      const vulkan = hasVulkanDll();
      const gpus = await detectGpus();
    
      let gpuLines = "";
      if (gpus.length === 0) {
        gpuLines = "⚠️  No GPU detected via wmic — this may indicate a driver issue.\n";
      } else {
        for (const gpu of gpus) {
          const vramStr = formatVram(gpu.vramBytes);
          gpuLines += `🖥️  GPU:   ${gpu.name}\n`;
          gpuLines += `💾  VRAM:  ${vramStr} (reported by Windows — may be half of actual on AMD cards)\n`;
          if (gpu.vramBytes > 0) {
            gpuLines += `📦  Recommended model: ${recommendedModel(gpu.vramBytes)}\n`;
          }
          gpuLines += "\n";
        }
      }
    
      const vulkanLine = vulkan
        ? `✅ Vulkan binary:  ggml-vulkan.dll found — GPU acceleration is active`
        : `❌ Vulkan binary:  ggml-vulkan.dll NOT found — whisper is running CPU-only\n\n` +
          `   To enable GPU acceleration:\n` +
          `   Download whisper-vulkan-win-x64.zip from:\n` +
          `   https://github.com/eviscerations/whisper-windows-mcp/releases\n` +
          `   Extract to: ${dirname(WHISPER_CLI_PATH)}`;
    
      return {
        content: [{
          type: "text",
          text: `System check\n${"─".repeat(40)}\n\n${gpuLines}${vulkanLine}`,
        }],
      };
    }
  • The detectGpus() helper function used by check_system. Runs 'wmic path win32_VideoController get name,AdapterRAM /format:csv' to enumerate GPUs and their VRAM.
    interface GpuInfo {
      name: string;
      vramBytes: number;
    }
    
    async function detectGpus(): Promise<GpuInfo[]> {
      try {
        const { stdout } = await execFileAsync(
          "wmic",
          ["path", "win32_VideoController", "get", "name,AdapterRAM", "/format:csv"],
          { windowsHide: true }
        );
        const gpus: GpuInfo[] = [];
        for (const line of stdout.split(/\r?\n/)) {
          const trimmed = line.trim();
          if (!trimmed || trimmed.startsWith("Node") || trimmed.startsWith(",AdapterRAM")) continue;
          const parts = trimmed.split(",");
          if (parts.length < 3) continue;
          const vramBytes = parseInt(parts[1] ?? "0", 10) || 0;
          const name = (parts[2] ?? "").trim();
          if (name && name !== "Name") gpus.push({ name, vramBytes });
        }
        return gpus;
      } catch {
        return [];
      }
    }
  • The formatVram() helper used by check_system to format VRAM bytes into human-readable string.
    function formatVram(bytes: number): string {
      if (!bytes || bytes < 1024 * 1024) return "Unknown";
      const gb = bytes / (1024 * 1024 * 1024);
      return gb >= 1 ? `${gb.toFixed(1)} GB` : `${Math.round(bytes / (1024 * 1024))} MB`;
    }
  • The recommendedModel() helper used by check_system to recommend a Whisper model based on VRAM size.
    function recommendedModel(vramBytes: number): string {
      const gb = vramBytes / (1024 * 1024 * 1024);
      if (gb >= 6) return "large-v3-turbo (ggml-large-v3-turbo.bin) — ~6x faster than large-v3, minimal accuracy loss for English";
      if (gb >= 4) return "medium.en (ggml-medium.en.bin) — good fit for your VRAM";
      if (gb >= 2) return "small.en (ggml-small.en.bin) — safe choice for your VRAM";
      return "base.en (ggml-base.en.bin) — recommended for limited VRAM";
    }
Behavior4/5

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

The description explains what the tool does and what it reports, providing good behavioral context. However, since no annotations are present, it would benefit from explicitly stating that the tool is read-only and does not modify system state, though this is implied by its diagnostic nature.

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 extremely concise: two sentences cover purpose, outputs, and usage context. Information is front-loaded, and every sentence adds value.

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

Completeness5/5

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

Given zero parameters and no output schema, the description provides complete context about what the tool checks and reports (GPU name, VRAM, Vulkan status, model recommendation). No gaps remain for typical use.

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?

The tool has zero parameters, so description cannot add parameter-specific meaning. Per guidelines, baseline score is 4, and the description does not add misleading or irrelevant information.

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 clearly states the tool detects GPU hardware and verifies Vulkan acceleration. It lists specific outputs (GPU name, VRAM, Vulkan status, model recommendation). Among sibling tools, none duplicate this functionality, making the purpose distinct and unambiguous.

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?

The description explicitly advises when to use the tool ('if you want to confirm GPU acceleration is working or diagnose why it isn't'). While it does not list alternatives or exclusions, the context is clear and sufficient for most use cases.

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

Install Server

Other Tools

Latest Blog Posts

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/eviscerations/whisper-windows-mcp'

If you have feedback or need assistance with the MCP directory API, please join our Discord server