Three tools with complete input schemas and descriptions. text_to_speech and speech_to_text have detailed parameter documentation (stability, similarity_boost, output_format enums, etc.), but sound_effects lacks visible schema/description in source. All tools include cost warnings. Parameter descriptions are generally good (72-char baseline met), but descriptions could be more concise for LLM optimization. No output schemas documented. Error handling present but minimal guidance for recovery. Missing tool annotations (readOnlyHint/destructiveHint).

• Add complete schema and description for sound_effects tool. Include parameter types (text, duration, output_directory, etc.), constraints, and output structure.
• Document output schemas for all tools. E.g., text_to_speech returns {file_path: string, voice_name: string, duration_seconds: float}. This enables agents to chain calls.
• Reduce tool descriptions to 50-150 chars. Restructure as: '[ACTION] [WHEN] [RETURNS]'. E.g., 'Convert text to speech using ElevenLabs API. Use when user requests audio generation. Returns file path and voice metadata.'
• Add tool annotations: text_to_speech and sound_effects with destructiveHint=true (write files, incur costs); speech_to_text with readOnlyHint=true. Enables agent reasoning about side effects.
• Enhance error messages with recovery guidance. E.g., 'Voice not found: {voice_name}. Available voices: [list]. Call list_voices() to discover options.' Implement pattern:recovery-guide.