RAPP  /  installer  /  shortcuts  /  brainstem-voice
🎙️

Brainstem Voice

The 5-action Apple Shortcut that turns Siri into a brainstem client. "Hey Siri, ask Brainstem…"

🟡 Author phase. The protocol is locked and the infrastructure is in place. The signed .shortcut file lands here once authored on a Mac with Shortcuts.app. Until then, follow the build steps below — it takes 5 minutes.

What it does

You say "Hey Siri, ask Brainstem [thing]" (or tap the Shortcut). Siri dictates → POSTs to your brainstem's /chat endpoint → extracts the voice_response field → speaks it aloud. The brainstem already emits a TTS-shaped voice slot for exactly this; the Shortcut is just the harness.

Surfaces it covers

📱 iPhone 📱 iPad 💻 Mac ⌚ Apple Watch 🔊 HomePod 🚗 CarPlay

One bundle. Six surfaces. Zero native code.

Build it (5 minutes)

Open Shortcuts.app on macOS or iOS. File → New Shortcut. Name it Brainstem Voice. Add these 5 actions in order:

  1. Ask for Input Prompt: "What do you want to ask?". Input Type: Text. (When triggered from Siri or Watch, this becomes the dictation prompt automatically.)
  2. Get Contents of URL URL: {your_brainstem_url}/chat. Method: POST. Headers: Content-Type: application/json. Request Body: JSON with user_input = Provided Input and conversation_history = empty list.
  3. Get Dictionary Value Get: Value. Key: voice_response. From: Contents of URL.
  4. If (optional fallback) If Dictionary Value is empty, fetch response instead. (Used when voice mode is off.)
  5. Speak Text Text: Dictionary Value. Voice: system default. Wait Until Finished: on.

Tap the play button to test. Type or speak something. Confirm the brainstem's voice response is spoken back.

Then sign + ship

  1. File → Export… → save as brainstem-voice.shortcut.
  2. Sign for public install: bash installer/shortcuts/sign.sh brainstem-voice.shortcut
  3. Share → Copy iCloud Link → paste it in this directory's README.md.

Configure on first run

The Shortcut prompts once and stores the answers as Shortcut variables:

VariableWhatExample
brainstem_urlHTTP root of your brainstemhttp://192.168.1.42:7071
auth_tokenBearer token (Tier 2 only)empty for Tier 1
voice_modeWhether to expect a voice slotdefault true

To change later, edit the Shortcut's variables in Shortcuts.app — no need to rebuild.

Why this and not a native app

A native watchOS / iOS app would need an Apple Developer Program enrollment ($99/yr), App Store review per release, watchOS-specific code, per-OS-version regression testing. This Shortcut needs none of it. The brainstem already emits the voice slot. The Shortcut is harness, not product.

The constitutional posture is captured in the Surfaces vault note: reach new form factors with web tech and OS-native primitives, not native code.