Skip to main content
Claude Board includes a built-in voice assistant that lets you create tasks, check status, and manage your board using voice commands.

​
Getting Started

Click the microphone button (floating action button) in the bottom-right corner to open the assistant panel. You can also press Alt+V to toggle the microphone when the panel is open.

​
Available Commands

CommandWhat it does
”Create task”Starts a multi-step flow to create a new task
”List tasks”Reports task counts by status
”Change status”Guides you through moving a task to a new status
”Help”Lists all available commands
”Cancel”Stops any active operation

​
Task Creation Flow

When you say “create task”, the assistant walks you through:
  1. Title — “What should the task title be?”
  2. Description — “Would you like to add a description?” (say “skip” to skip)
  3. Type — “What type? Feature, bugfix, refactor, docs, test, or chore.”
  4. Priority — “Priority? None, low, medium, or high.”
  5. Confirm — “Shall I create it?” (say “yes” or “no”)

​
Features

​
Text-to-Speech

The assistant reads its responses aloud. Toggle the speaker icon in the panel header to mute/unmute.

​
Audio Visualizer

A real-time waveform visualizer appears in the panel when the microphone is active, showing audio input levels.

​
Sound Effects

  • Rising beep when listening starts
  • Falling beep when listening stops

​
Command Hints

When the assistant is idle, clickable suggestion chips appear for quick access to common commands.

​
Text Input

You can also type commands in the text input field at the bottom of the panel. Voice and text input work identically.

​
Keyboard Shortcut

Press Alt+V to toggle the microphone on/off without clicking.

​
Architecture

The voice assistant uses a modular plugin architecture: 
features/voice/
├── VoiceAssistant.jsx           # Slim UI shell (FAB + panel)
├── VoiceAssistantProvider.jsx   # React context + state machine
├── engine/
│   ├── ttsEngine.js             # Text-to-speech (Web Speech API)
│   ├── sttEngine.js             # Audio analyser for waveform
│   └── soundEffects.js          # Web Audio API beep effects
├── intent/
│   ├── intentParser.js          # Pattern-based intent detection
│   └── entityExtractors.js      # Extract type, priority, model
├── commands/
│   ├── commandRegistry.js       # Plugin registry
│   ├── createTaskCommand.js     # "Create task" flow
│   ├── listTasksCommand.js      # "List tasks"
│   ├── changeStatusCommand.js   # "Change status" flow
│   ├── helpCommand.js           # "Help"
│   └── cancelCommand.js         # "Cancel"
└── components/
    ├── ChatPanel.jsx            # Message list + controls
    ├── AudioVisualizer.jsx      # Canvas waveform
    └── CommandHints.jsx         # Suggestion chips

​
Adding Custom Commands

To add a new voice command:
  1. Create a new file in commands/ (e.g., myCommand.js)
  2. Use registerCommand() with patterns, flow states, and an execute function
  3. Import the file in commands/index.js
No other files need to change — the registry auto-discovers new commands.

​
Browser Support

The voice assistant requires browser support for the Web Speech API:
  • Chrome / Edge — Full support (recognition + synthesis)
  • Safari — Synthesis only (no recognition)
  • Firefox — Limited support
If speech recognition is not available, the microphone button is disabled but text input still works.

​
Voice Dictation in Task Modal

Separately from the assistant, the task creation modal also has mic buttons on the title and description fields for direct voice dictation. These work independently and are useful for quickly filling in form fields.