feat: Add voice input with pluggable backend (Gemini zero-install + Whisper)

conductor/check-pr-status.md

@@ -0,0 +1,11 @@
+# Objective
+
+Check PR status and implement a fix if there are any issues.
+
+# Implementation steps
+
+1. Exit **Plan Mode** with `Shift+Tab`.
+2. Verify the PR is mergeable with `gh pr view`. If there are conflicts, fetch
+   and merge upstream/main.
+3. Verify the tests are passing with `gh pr checks` and resolve and investigate
+   any failing checks.

docs/cli/settings.md

@@ -85,6 +85,15 @@ they appear in the UI.
 | Error Verbosity                      | `ui.errorVerbosity`                    | Controls whether recoverable errors are hidden (low) or fully shown (full).                                                                                       | `"low"` |
 | Screen Reader Mode                   | `ui.accessibility.screenReader`        | Render output in plain-text to be more screen reader accessible                                                                                                   | `false` |
 
+### Voice
+
+| UI Label                    | Setting                  | Description                                                                                                                                                  | Default     |
+| --------------------------- | ------------------------ | ------------------------------------------------------------------------------------------------------------------------------------------------------------ | ----------- |
+| Enable Voice Input          | `voice.enabled`          | Enable voice input support. Note: Voice input is not natively supported in WSL2 (Windows Subsystem for Linux).                                               | `false`     |
+| Transcription Backend       | `voice.provider`         | Set transcription backend: "gemini" (default, zero-install) or "whisper" (local).                                                                            | `"gemini"`  |
+| Whisper Binary Path         | `voice.whisperPath`      | Path to the whisper executable. Only used when provider is "whisper".                                                                                        | `undefined` |
+| Silence Detection Threshold | `voice.silenceThreshold` | RMS energy threshold (0–1000) below which audio is discarded as silence. Lower values allow quieter speech such as whispering. 0 disables silence detection. | `80`        |
+
 ### IDE
 
 | UI Label | Setting       | Description                  | Default |

docs/reference/commands.md

@@ -255,6 +255,31 @@ Slash commands provide meta-level control over the CLI itself.
     - **Description:** List configured MCP servers and tools with descriptions
       and schemas.
 
+### `/voice`
+
+- **Description:** Manage voice input configuration and inspect current voice
+  settings.
+- **Shortcuts:** Press `Space Space` on an empty prompt to start or stop
+  recording. Press **Esc** while recording to cancel.
+- **Sub-commands:**
+  - **`enable`**:
+    - **Description:** Enable voice input.
+  - **`disable`**:
+    - **Description:** Disable voice input.
+  - **`provider [gemini|whisper]`**:
+    - **Description:** Set the transcription backend.
+  - **`sensitivity <0-1000>`**:
+    - **Description:** Set the silence detection threshold. `0` disables silence
+      filtering.
+  - **`set-path <path>`**:
+    - **Description:** Set the path to the Whisper binary when using the local
+      Whisper backend.
+  - **`help`**:
+    - **Description:** Show voice command help.
+  - **Default behavior:**
+    - **Description:** Running `/voice` with no sub-command shows the current
+      voice settings.
+
 ### `/memory`
 
 - **Description:** Manage the AI's instructional context (hierarchical memory

docs/reference/configuration.md

@@ -409,6 +409,30 @@ their corresponding top-level category object in your `settings.json` file.
   - **Default:** `false`
   - **Requires restart:** Yes
 
+#### `voice`
+
+- **`voice.enabled`** (boolean):
+  - **Description:** Enable voice input support. Note: Voice input is not
+    natively supported in WSL2 (Windows Subsystem for Linux).
+  - **Default:** `false`
+
+- **`voice.provider`** (enum):
+  - **Description:** Set transcription backend: "gemini" (default, zero-install)
+    or "whisper" (local).
+  - **Default:** `"gemini"`
+  - **Values:** `"gemini"`, `"whisper"`
+
+- **`voice.whisperPath`** (string):
+  - **Description:** Path to the whisper executable. Only used when provider is
+    "whisper".
+  - **Default:** `undefined`
+
+- **`voice.silenceThreshold`** (number):
+  - **Description:** RMS energy threshold (0–1000) below which audio is
+    discarded as silence. Lower values allow quieter speech such as whispering.
+    0 disables silence detection.
+  - **Default:** `80`
+
 #### `ide`
 
 - **`ide.enabled`** (boolean):

packages/cli/src/config/settingsSchema.ts

@@ -902,6 +902,64 @@ const SETTINGS_SCHEMA = {
     },
   },
 
+  voice: {
+    type: 'object',
+    label: 'Voice Input',
+    category: 'General',
+    requiresRestart: false,
+    default: {},
+    description:
+      'Settings for voice input. Note: Voice input is not natively supported in WSL2 (Windows Subsystem for Linux).',
+    showInDialog: false,
+    properties: {
+      enabled: {
+        type: 'boolean',
+        label: 'Enable Voice Input',
+        category: 'General',
+        requiresRestart: false,
+        default: false,
+        description:
+          'Enable voice input support. Note: Voice input is not natively supported in WSL2 (Windows Subsystem for Linux).',
+        showInDialog: true,
+      },
+      provider: {
+        type: 'enum',
+        label: 'Transcription Backend',
+        category: 'General',
+        requiresRestart: false,
+        default: 'gemini',
+        description:
+          'Set transcription backend: "gemini" (default, zero-install) or "whisper" (local).',
+        showInDialog: true,
+        options: [
+          { value: 'gemini', label: 'Gemini (Cloud)' },
+          { value: 'whisper', label: 'Whisper (Local)' },
+        ],
+      },
+      whisperPath: {
+        type: 'string',
+        label: 'Whisper Binary Path',
+        category: 'General',
+        requiresRestart: false,
+        default: undefined as string | undefined,
+        description:
+          'Path to the whisper executable. Only used when provider is "whisper".',
+        showInDialog: true,
+      },
+      silenceThreshold: {
+        type: 'number',
+        label: 'Silence Detection Threshold',
+        category: 'General',
+        requiresRestart: false,
+        default: 80,
+        description:
+          'RMS energy threshold (0–1000) below which audio is discarded as silence. ' +
+          'Lower values allow quieter speech such as whispering. 0 disables silence detection.',
+        showInDialog: true,
+      },
+    },
+  },
+
   ide: {
     type: 'object',
     label: 'IDE',