Say

Overview

Speak the bot's last reply aloud using a Bash-based text-to-speech script. It supports a customizable rate, defaulting to 300 words per minute, for accessible and hands-free interaction.

How This Skill Works

When invoked, the skill runs the speak.sh script with the current session ID and an optional rate argument. The script reads the last response, converts it to speech, and plays it back through the system's TTS engine. The key command is: ${CLAUDE_PLUGIN_ROOT}/skills/say/scripts/speak.sh ${CLAUDE_SESSION_ID} $1.

When to Use It

• Hear the bot's last reply instead of reading it.
• Multitasking or hands-free contexts (driving, cooking) requiring audio feedback.
• Accessibility needs for users who prefer audio output.
• Demos and QA to quickly verify responses via audio.
• Tuning pace and clarity by adjusting the speaking rate.

Quick Start

1. Step 1: Ensure CLAUDE_PLUGIN_ROOT and CLAUDE_SESSION_ID are set in your environment.
2. Step 2: Run the command with an optional rate, e.g. 350: ${CLAUDE_PLUGIN_ROOT}/skills/say/scripts/speak.sh ${CLAUDE_SESSION_ID} 350
3. Step 3: Listen to the spoken last response and adjust the rate as needed.

Best Practices

• Test with the default rate (300 wpm) before adjusting.
• Choose a rate that matches content length and listener comfort.
• Ensure spoken content does not reveal sensitive data; consider transcripts for audits.
• Verify the spoken output matches the last response for accuracy.
• Log spoken events for traceability and debugging.

Example Use Cases

• QA engineers listen to the last bot reply during scripted tests.
• A developer reviews long responses hands-free while coding.
• An accessibility demo showcases voice-first interaction with the bot.
• Support teams audit chat transcripts by listening to them.
• Product demos illustrate TTS playback of bot responses.

Frequently Asked Questions