v0.1.0 — Initial Release

First public release of audio-mcp — a local MCP server for macOS that captures microphone + system audio into explicit sessions and exposes raw WAV audio to AI agents.

Features

• Mic + system audio capture via a bundled Swift helper (ScreenCaptureKit + AVFoundation)
• Stereo output for capture: "both" sessions: L=mic, R=system — lossless, single file
• Live polling — get_audio works on active sessions so agents can stream audio out of a live recording
• 7 MCP tools: start_session, stop_session, get_audio, list_sessions, get_session, list_audio_sources, delete_session
• Signed + notarized Swift helper binary — no Gatekeeper prompt
• Session-based — never a background daemon, explicit start/stop

Install

Via npx (recommended):

```json
{
"mcpServers": {
"audio": {
"command": "npx",
"args": ["-y", "audio-mcp"]
}
}
}
```

Via global install:

```sh
npm install -g audio-mcp
```

Requirements

• macOS 13 (Ventura) or later
• Node.js 18 or later
• Microphone + Screen Recording permissions (granted to your MCP client on first session)

What it does NOT do

No transcription, no STT, no network calls. Raw WAV audio only.

Links

• npm: https://www.npmjs.com/package/audio-mcp
• README: https://github.com/BugorBN/audio-mcp#readme

License

MIT