Suno Song Creator

Core Concept

Suno does not read prompts like a human following instructions. Instead, it maps text into a probabilistic style-mesh, blending musical concepts based on co-occurrence patterns learned during training. Every word carries "statistical baggage" - associations that may not be intended.

Critical insight: "Pop" acts as a gravitational black hole. Nearly every genre (rock: 315B links, funk: 116B links, emo: 12.2B links) gets pulled toward pop unless actively counteracted through exclusions, unusual combinations, or strategic contrast.

When to Use This Skill

Use this skill to:

• Create complete Suno prompts from scratch
• Optimize existing prompts for better results
• Select appropriate models and parameters
• Write effective lyrics with proper meta tags
• Apply genre-specific production strategies
• Build consistent vocal personas
• Prevent common issues (lyric bleed, generic outputs)

Interactive Guidance Tools

This skill uses interactive tools to gather information and research musical styles:

AskUserQuestion Tool

Use AskUserQuestion to gather essential information through structured choices. This helps users clarify their vision and makes better recommendations.

When to use:

• Initial vision gathering (genre, mood, vocals)
• Model selection decisions
• Parameter configuration choices
• Style reference clarification
• When multiple valid approaches exist

When to ask vs. when to proceed:

Ask questions when:

• User provides vague or general request ("make me a song")
• Multiple valid approaches exist (genre, mood, model selection)
• User mentions preference but lacks specifics
• Choices have significant trade-offs to explain
• Clarification would improve the result

Proceed directly when:

• User provides detailed, specific requirements
• Artist reference given (research instead of asking)
• Only one appropriate approach exists
• User demonstrates expertise with terminology
• Request is clear and unambiguous

Example usage:

Question: "What's the primary genre for your song?"
Options:
- "Acoustic/Folk/Singer-Songwriter" (Natural vocals, intimate production)
- "Electronic/EDM/Synthwave" (Synthesized sounds, modern production)
- "Rock/Alternative" (Guitar-driven, raw energy)
- "Pop" (Polished, radio-ready hooks)

Artist and Song Research (Automated)

IMPORTANT: When user mentions artist reference, automatically launch song-researcher sub-agent for automated pattern analysis.

Trigger conditions:

• "I want a sad song like Phoebe Bridgers"
• "Something in Sabrina Carpenter's Manchild style"
• "Taylor Swift folklore era vibes"
• Any mention of artist name + song title or style reference

Automated research workflow:

1. Extract from user request:

   • Artist name (required)
   • Specific song name (if mentioned)
   • Style/mood hints from user's description

2. Launch song-researcher sub-agent via Task tool:

   Task tool:
     subagent_type: "song-researcher"
     description: "Research artist patterns"
     prompt: "Research [Artist] - [Song if mentioned]. User wants [style/mood description]."
   

3. Sub-agent performs automated research:

   • Genius.com: Fetches lyrics, structure, annotations for specific song or popular tracks
   • HookTheory/Chord sites: Searches for chord progressions and musical analysis
   • Spotify/Music sites: Finds genre tags, similar artists, production notes
   • Pattern extraction: Analyzes syllables, rhyme schemes, metaphors, emotional arc
   • Multi-song analysis: If specific song mentioned, also analyzes 2-3 other tracks for context
   • Confidence scoring: Rates research quality (90-100% comprehensive, 70-89% good, 50-69% basic, <50% insufficient)

4. Receive structured research report:

   • Research Quality section (confidence score, sources used, songs analyzed)
   • Primary Song Analysis (lyrical structure, thematic elements, musical context)
   • Artist Context (consistent patterns across songs, variations observed)
   • Recommendations for Suno Prompt (genre descriptors, vocal persona, lyrical guidance, production notes)
   • Research Limitations (transparently notes missing data)

5. Use research findings to inform subsequent steps:

   • Step 3 (Build Prompt): Genre, vocal, instrumentation, production from research recommendations
   • Step 5 (Write Lyrics): Syllable patterns, rhyme schemes, metaphor approach from analysis
   • Step 4 (Parameters): Model selection based on researched style

Benefits of automated research:

• ✅ No manual browsing required
• ✅ Structured pattern data every time
• ✅ Multi-song comparative analysis
• ✅ Confidence scoring shows data quality
• ✅ Graceful degradation when data limited
• ✅ Actionable recommendations ready for prompt building

Error handling:

• Artist not found → Sub-agent tries alternate spellings, returns partial results with low confidence
• Limited data → Uses available sources, notes limitations clearly
• Specific song unavailable → Falls back to analyzing popular songs, notes in report
• No data at all → Returns empty report, main agent asks user for style guidance

Important: Research is for understanding patterns and inspiration, not copying. Sub-agent extracts structural patterns; main agent creates completely original lyrics informed by these learned structures.

For standalone research: Users can also invoke /research-artist [Artist] independently to explore patterns before creating songs.

File Writing Tool

Use the Write tool to save complete prompts to organized project directories after creation. This enables:

Benefits:

• Organized prompt library by project/album
• Future reference and iteration on successful prompts
• Version control and prompt history tracking
• Reusable templates for similar songs
• Complete documentation of creative decisions

When files are saved:

• Automatically after completing all workflow steps
• Project-based directory structure in current working directory
• Single markdown file per song containing all prompt components
• Includes metadata, configuration, prompt, lyrics, and research notes

Practical Example: Interactive + Automated Research Workflow

User says: "I want a sad song like Phoebe Bridgers"

Step 1 - Ask clarifying questions:

Question: "What aspects of Phoebe Bridgers' style appeal to you?"
Header: "Style Focus"
Options:
- "Intimate, confessional lyrics" (Bedroom pop, vulnerable delivery)
- "Melancholic atmosphere" (Dreamy production, sad mood)
- "Indie folk instrumentation" (Acoustic guitar, minimal production)
- "All of the above" (Comprehensive Phoebe Bridgers approach)

User selects: "All of the above"

Step 2 - Automated research (song-researcher sub-agent):

Main agent launches research:

Task tool:
  subagent_type: "song-researcher"
  description: "Research Phoebe Bridgers patterns"
  prompt: "Research Phoebe Bridgers. User wants sad, intimate, indie folk style."

Sub-agent performs:

• WebSearch + WebFetch: "Phoebe Bridgers Motion Sickness lyrics Genius"
• WebSearch + WebFetch: "Phoebe Bridgers Kyoto lyrics Genius" (context song)
• WebSearch + WebFetch: "Phoebe Bridgers Scott Street lyrics Genius" (context song)
• WebSearch: "Phoebe Bridgers genre Spotify similar artists"
• WebSearch: "Phoebe Bridgers chord progressions"

Sub-agent returns structured report:

# Research Report: Phoebe Bridgers

## Research Quality
Confidence Score: 85% (Good)
Sources Used: Genius ✓, Chords ✗, Spotify/Context ✓
Songs Analyzed: 3 total (Motion Sickness, Kyoto, Scott Street)

## Artist Context

### Consistent Patterns
- Syllable counts: 8-10 in verses, 6-9 in chorus
- Rhyme: Loose/slant rhymes, conversational (ABCB common)
- Structure: Verse-Chorus-Verse-Chorus-Bridge-Final Chorus
- Metaphor: One central metaphor per song, concrete imagery
- Tone: Vulnerable-to-defiant arc, confessional
- Vocabulary: Specific details, conversational language

## Recommendations for Suno Prompt

Genre: "indie folk, 2020s bedroom pop, Phoebe Bridgers x Julien Baker sensibility"
Vocal: "soft female alto, whisper-to-belt range, confessional delivery, vulnerable yet defiant, slight breathiness"
Lyrical Guidance: 8-10 syllables verses, 6-9 chorus, loose ABCB rhyme, one central metaphor
Production: "lo-fi warmth, close-mic intimacy, fingerpicked acoustic guitar, minimal percussion"

Step 3 - Main agent uses research to build prompt: