Menu
Speak text aloud using local TTS. Use when the user wants Claude's output read aloud or to toggle auto-read mode.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
Capture the screen and inject a short text summary into Claude's context via an in-process VLM (llama-cpp-python). No daemon. Use to feed terminal/editor/browser state to Claude for debug, fix suggestions, or summarization with minimal token cost.
Transcribe speech using local STT. Use when the user wants voice input or offline audio transcription.
| name | speak |
| description | Speak text aloud using local TTS. Use when the user wants Claude's output read aloud or to toggle auto-read mode. |
| compatibility | Designed for Claude Code |
| metadata | {"allowed-tools":"Bash, Read","argument-hint":["text to speak or --toggle"],"stability":"development"} |
Speak text aloud using local text-to-speech.
/speak Hello world — speak specific text/speak --toggle — enable/disable auto-read mode/speak --voice en_GB-alan — use a specific voiceuv run python -m cc_tts.speak $ARGUMENTS
See .cc-senses.example.toml [tts] section for the full schema and CC_TTS_* env overrides.
Three paths from text to audio (Stop hook, stream-json pipe, PTY proxy). See docs/architecture.md for the comparison table and docs/adr/0001-tts-delivery-modes.md for the rationale.
Do not combine Stop hook + PTY proxy — causes double speaking.
For the full bidirectional flow (/speak --toggle → /voice → speak → response spoken back), see docs/UserStory.md Flow A.