Skip to content

Latest commit

 

History

History
159 lines (133 loc) · 4.96 KB

File metadata and controls

159 lines (133 loc) · 4.96 KB

Claude Voice Commander - Operational Guide

Quick Start

  1. Run python main.py
  2. Focus on terminal with Claude CLI running (10-second countdown)
  3. Start speaking commands

Voice Commands Reference

Claude CLI Mode Control

Voice Command Action
"plan mode" Switch to plan mode (/plan)
"planning mode" Switch to plan mode (/plan)
"auto mode" Switch to auto mode (/auto)
"auto accept" Switch to auto mode (/auto)
"interactive mode" Switch to interactive mode (/interactive)
"ask before apply" Switch to interactive mode (/interactive)

Option Selection

Voice Command Sends
"option one" / "select one" 1
"option two" / "select two" 2
"option three" / "select three" 3
"option four" / "select four" 4
"option five" / "select five" 5

Confirmations

Voice Command Sends
"yes" / "accept" / "continue" y
"no" / "cancel" / "reject" n
"quit" / "exit" q
"help" h

Text Accumulation System

How It Works

  1. Start Speaking: Say any non-control command → system starts accumulating text
  2. Continue Speaking: Keep adding to your request → text gets combined
  3. Finalize: Say a finalization keyword → sends complete request to Claude

Finalization Keywords

  • "send to claude"
  • "execute command"
  • "process request"
  • "submit"

Example Usage

You: "Create a new Python function"
System: [ACCUMULATING] Create a new Python function
You: "that calculates the factorial of a number"
System: [ADDED] that calculates the factorial of a number
System: [FULL TEXT] Create a new Python function that calculates the factorial of a number
You: "send to claude"
System: → Sends complete request to Claude CLI

Voice Command Patterns

File Operations

Voice Pattern Example Generated Command
"open file [filename]" "open file main.py" "Open and edit the file main.py"
"create file [filename]" "create file test.js" "Create a new file named test.js"
"show directory contents" - "Show the current directory contents"

Code Operations

Voice Pattern Example Generated Command
"run [target]" "run tests" "Run tests"
"build [target]" "build project" "Build the project"
"debug [target]" "debug function" "Debug and fix function"

Git Operations

Voice Pattern Example Generated Command
"git status" - "Show git status"
"git commit [message]" "git commit added feature" "Git commit with message: added feature"
"git push" - "Git push changes"

Search Operations

Voice Pattern Example Generated Command
"find [term]" "find function getName" "Search for 'function getName' in the codebase"
"search for [term] in [location]" "search for TODO in src" "Search for 'TODO' in src"

Speech Recognition Tips

For Best Recognition

  • Speak clearly at moderate pace
  • Use natural pauses between command parts
  • Avoid background noise when possible
  • Use consistent phrases from the command tables

Common Speech Corrections

The system automatically corrects:

  • "clause" → "claude"
  • "clawed" → "claude"
  • "called" → "claude"
  • "dot py" → ".py"
  • "dot js" → ".js"

Number Words

Word Converts To
"one" "1"
"two" "2"
"three" "3"
"four" "4"
"five" "5"

Workflow Examples

Simple Command

You: "git status"
System: → Immediately sends "Show git status" to Claude

Complex Request (Text Accumulation)

You: "Write a Python function"
System: [ACCUMULATING] Write a Python function
You: "that reads a CSV file"
System: [ADDED] that reads a CSV file  
You: "and returns the data as a dictionary"
System: [ADDED] and returns the data as a dictionary
You: "execute command"
System: → Sends complete request to Claude

Mode Switching

You: "plan mode"
System: → Sends "/plan" to Claude CLI
You: "Create a REST API endpoint"
System: [ACCUMULATING] Create a REST API endpoint
You: "send to claude"
System: → Sends request to Claude (will create plan instead of executing)

Error Recovery

If Command Not Recognized

  • Speak more clearly and try again
  • Use exact phrases from command tables
  • Check if accumulation is active - say finalization keyword first

If Wrong Mode

  • Switch modes with voice: "auto mode", "plan mode", "interactive mode"
  • Use shortcuts: "yes" for accept, "no" for cancel

If System Unresponsive

  • Check terminal focus - system sends to focused window from startup
  • Restart application if needed
  • Check microphone permissions and settings