mcp-windows/src/tools/audio.ts

/* Copyright (C) 2026 Moko Consulting <hello@mokoconsulting.tech>
 * SPDX-License-Identifier: GPL-3.0-or-later
 *
 * Tools: windows_audio_get (#6), windows_audio_set (#7)
 * Uses compiled SetMute.exe for reliable COM audio control.
 */

import { McpServer } from '@modelcontextprotocol/sdk/server/mcp.js';
import { z } from 'zod';
import { runPowerShell } from '../shell.js';

const AUDIO_PS = `
Add-Type -TypeDefinition @'
using System;
using System.Runtime.InteropServices;

public class WinAudio {
    [Guid("A95664D2-9614-4F35-A746-DE8DB63617E6"), InterfaceType(ComInterfaceType.InterfaceIsIUnknown)]
    interface IMMDeviceEnumerator {
        int EnumAudioEndpoints(int dataFlow, int dwStateMask, out IntPtr ppDevices);
        int GetDefaultAudioEndpoint(int dataFlow, int role, out IMMDevice ppEndpoint);
    }

    [Guid("D666063F-1587-4E43-81F1-B948E807363F"), InterfaceType(ComInterfaceType.InterfaceIsIUnknown)]
    interface IMMDevice {
        int Activate(ref Guid iid, int dwClsCtx, IntPtr pActivationParams, [MarshalAs(UnmanagedType.IUnknown)] out object ppInterface);
    }

    [Guid("5CDF2C82-841E-4546-9722-0CF74078229A"), InterfaceType(ComInterfaceType.InterfaceIsIUnknown)]
    interface IAudioEndpointVolume {
        int RegisterControlChangeNotify(IntPtr pNotify);
        int UnregisterControlChangeNotify(IntPtr pNotify);
        int GetChannelCount(out int pnChannelCount);
        int SetMasterVolumeLevel(float fLevelDB, ref Guid pguidEventContext);
        int SetMasterVolumeLevelScalar(float fLevel, ref Guid pguidEventContext);
        int GetMasterVolumeLevel(out float pfLevelDB);
        int GetMasterVolumeLevelScalar(out float pfLevel);
        int SetChannelVolumeLevel(int nChannel, float fLevelDB, ref Guid pguidEventContext);
        int SetChannelVolumeLevelScalar(int nChannel, float fLevel, ref Guid pguidEventContext);
        int GetChannelVolumeLevel(int nChannel, out float pfLevelDB);
        int GetChannelVolumeLevelScalar(int nChannel, out float pfLevel);
        int SetMute([MarshalAs(UnmanagedType.Bool)] bool bMute, ref Guid pguidEventContext);
        int GetMute([MarshalAs(UnmanagedType.Bool)] out bool pbMute);
    }

    private static IAudioEndpointVolume GetVolume() {
        var type = Type.GetTypeFromCLSID(new Guid("BCDE0395-E52F-467C-8E3D-C4579291692E"));
        var enumerator = (IMMDeviceEnumerator)Activator.CreateInstance(type);
        IMMDevice device;
        enumerator.GetDefaultAudioEndpoint(0, 1, out device);
        var iid = new Guid("5CDF2C82-841E-4546-9722-0CF74078229A");
        object obj;
        device.Activate(ref iid, 0x17, IntPtr.Zero, out obj);
        return (IAudioEndpointVolume)obj;
    }

    public static float GetVolumeLevel() {
        var vol = GetVolume();
        float level;
        vol.GetMasterVolumeLevelScalar(out level);
        return level;
    }

    public static bool GetMute() {
        var vol = GetVolume();
        bool muted;
        vol.GetMute(out muted);
        return muted;
    }

    public static void SetVolumeLevel(float level) {
        var vol = GetVolume();
        var ctx = Guid.Empty;
        vol.SetMasterVolumeLevelScalar(level, ref ctx);
    }

    public static void SetMute(bool mute) {
        var vol = GetVolume();
        var ctx = Guid.Empty;
        vol.SetMute(mute, ref ctx);
    }
}
'@ -ErrorAction Stop
`;

export function registerAudioTools(server: McpServer): void {
	server.tool(
		'windows_audio_get',
		'Get current audio state: volume level (0-100), mute status, and default playback device.',
		{},
		async () => {
			const ps = `
${AUDIO_PS}
$volume = [math]::Round([WinAudio]::GetVolumeLevel() * 100)
$muted = [WinAudio]::GetMute()
$device = (Get-CimInstance Win32_SoundDevice | Where-Object { $_.Status -eq 'OK' } | Select-Object -First 1).Name
[PSCustomObject]@{
  volume = $volume
  muted = $muted
  device = $device
} | ConvertTo-Json -Compress`;

			const result = await runPowerShell(ps, { timeout: 60000 });
			if (result.exitCode !== 0) {
				return { content: [{ type: 'text', text: `Error: ${result.stderr}` }], isError: true };
			}

			const state = JSON.parse(result.stdout);
			const muteIcon = state.muted ? '🔇' : (state.volume > 50 ? '🔊' : '🔉');
			return {
				content: [{
					type: 'text',
					text: `${muteIcon} Volume: ${state.volume}%${state.muted ? ' (MUTED)' : ''}\nDevice: ${state.device || 'Unknown'}`,
				}],
			};
		},
	);

	server.tool(
		'windows_audio_set',
		'Set audio volume (0-100), mute/unmute, or toggle mute.',
		{
			volume: z.number().min(0).max(100).optional().describe('Volume level 0-100'),
			mute: z.enum(['true', 'false', 'toggle']).optional().describe('Mute state: true, false, or toggle'),
		},
		async ({ volume, mute }) => {
			const commands: string[] = [AUDIO_PS];

			if (volume !== undefined) {
				commands.push(`[WinAudio]::SetVolumeLevel(${volume / 100})`);
			}

			if (mute === 'toggle') {
				commands.push(`[WinAudio]::SetMute(-not [WinAudio]::GetMute())`);
			} else if (mute === 'true') {
				commands.push(`[WinAudio]::SetMute($true)`);
			} else if (mute === 'false') {
				commands.push(`[WinAudio]::SetMute($false)`);
			}

			// Read back state
			commands.push(`
$vol = [math]::Round([WinAudio]::GetVolumeLevel() * 100)
$m = [WinAudio]::GetMute()
[PSCustomObject]@{ volume = $vol; muted = $m } | ConvertTo-Json -Compress`);

			const result = await runPowerShell(commands.join('\n'));
			if (result.exitCode !== 0) {
				return { content: [{ type: 'text', text: `Error: ${result.stderr}` }], isError: true };
			}

			const state = JSON.parse(result.stdout);
			return {
				content: [{
					type: 'text',
					text: `Volume set to ${state.volume}%${state.muted ? ' (MUTED)' : ''}`,
				}],
			};
		},
	);
}