This brings ElevenLabs text-to-speech to Claude with the same ease as macOS's built-in say command. You get voice responses in chat when users ask for them, plus fine control over delivery with audio tags like [whispers], [excited], and [short pause] for the v3 model. It handles the usual pain points: respelling for pronunciation, normalization flags for numbers and URLs, language hints for better output. The setup is clean if you're on Mac (brew install), needs an ElevenLabs API key, and defaults to their expressive v3 model. Honestly most useful when you want Claude to actually speak responses instead of just suggesting what to say, especially for character voices or accessibility scenarios.
npx -y skills add steipete/clawdis --skill sag --agent claude-codeInstalls into .claude/skills of the current project.
Select a file.
juliusbrussee/caveman
mattpocock/skills
shadcn/improve
obra/superpowers
forrestchang/andrej-karpathy-skills
vercel-labs/skills