
This handles AssemblyAI's speech-to-text APIs, streaming transcription, and their LLM Gateway for applying language models to audio. The skill exists because training data gets the authentication wrong (no Bearer prefix, just the raw API key), points to deprecated SDKs (Java, Go, and C# were discontinued in April 2025), and references the old LeMUR API that sunset in March 2026. It also catches prompting gotchas on Universal-3 Pro, like never using negative instructions and the mutual exclusivity of prompt versus keyterms_prompt. If you're building voice agents with LiveKit or Pipecat, or need medical transcription, PII redaction, or real-time streaming STT, this will save you from outdated examples and auth failures.
npx -y skills add assemblyai/assemblyai-skill --skill assemblyai --agent claude-codeInstalls into .claude/skills of the current project.
Select a file.