Three tools with Turkish-language descriptions and reasonable schemas, but significant gaps in parameter documentation and error handling. Tool names are action-oriented (generate_*, analyze_*) but descriptions lack clarity on when to use each vs. the others. Input schemas present but parameter descriptions sparse or missing. No output schema documentation. Error responses are generic ('error' field with text) rather than actionable recovery guidance. Content/copyright guards exist but not exposed as tool parameters or documented in descriptions.

• Expand analyze_emotion description to 100 - 200 chars explaining: 'Analyzes emotion, energy, BPM, key, and instruments from Turkish text WITHOUT generating lyrics. Use this for lightweight analysis or when you only need metadata. Returns {emotion, energy, bpm, key, instruments, vocal_style, suno_style_prompt}.'
• Add description to analyze_emotion's 'text' parameter: 'Turkish text to analyze (story, concept, or lyrics, 5 - 800 words).'
• Document output schemas for all three tools. E.g., generate_music_prompt returns {style_prompt: string, structured_lyrics: string, title_suggestions: [string], parameters: {emotion, energy, bpm, key, instruments, vocal_style}, copyright_similarity: number}.
• Enhance error responses with recovery guidance. E.g., copyright failure: 'Generated lyrics too similar to training data (similarity=0.87, threshold=0.75). Retry with different emotion or energy level, or call analyze_emotion first to adjust parameters.'
• Document content/copyright guard behavior in tool descriptions. E.g., 'Input is checked for harmful content; output is checked for copyright similarity (Jaccard ≥0.75 triggers retry). Max 3 retries before failure.'

Inferred effective spec: <=2025-11-25.

• Relies on Logging (deprecated) - log to stderr or use OpenTelemetry