Extract descriptive musical characteristics from any artist or band without using their name, building a vocabulary of sonic qualities for AI music generation, music description, or creative recombination.
3k tokens
context cost
the whole folder, loaded on every use
1
files
instructions only
0
copies elsewhere
how many repositories repackaged it
128
stars on the repo
on the repository, not the skill itself
Install
one command, takes just this skill from the repository
Extract descriptive musical characteristics from any artist or band without using their name, building a vocabulary of sonic qualities for AI music generation, music description, or creative recombination. Replace "sounds like [Artist]" with specific, technique-focused descriptions.
Core Principle
How, not who. Describe techniques, approaches, and sonic qualities rather than referencing artists. This enables:
Ethical AI music generation
Precise communication about sound
Creative recombination of elements
Genre-independent vocabulary
Quick Reference: Six Dimensions
| Dimension | What to Analyze |
|-----------|-----------------|
| Rhythmic Foundation | Drums, tempo, bass lines, time signatures |
List 5-10 standalone phrases usable in AI generation:
"..."
"..."
Ethical Guidelines
Do
Combine elements from multiple analyses
Focus on techniques and approaches
Build reusable vocabulary
Create novel fusions
Don't
Copy complete profiles directly
Replicate signature riffs/melodies
Use as "sounds like [Artist]" substitute
Claim to reproduce specific artists
Anti-Patterns
1. The Name Drop
Pattern: Using artist names as shorthand instead of technique descriptions. "Sounds like Radiohead" instead of describing the actual sonic qualities.
Why it fails: Defeats the entire purpose. Artist names are black boxes that convey different things to different people and may produce copyright issues in AI generation.
Fix: Never use artist names in final output. For every "sounds like X," unpack what that actually means in terms of rhythm, harmony, production, etc.
2. The Single Dimension
Pattern: Analyzing only one dimension (usually rhythm or production) while ignoring others. Producing incomplete profiles.
Why it fails: Musical identity emerges from interaction of all dimensions. A rhythmic profile without harmonic context is useless for generation.
Fix: Force yourself through all six dimensions. Even if an artist seems "about the guitar sound," their rhythmic choices matter.
3. The Genre Substitute
Pattern: Describing music by genre labels instead of techniques. "Post-punk" instead of describing what makes it post-punk.
Why it fails: Genre labels are contested categories, not techniques. AI systems need concrete instructions, not genre negotiations.
Fix: Treat genre labels as starting points requiring unpacking. What rhythmic, harmonic, and production choices define this genre for this artist?
4. The Representative Track Trap
Pattern: Analyzing one famous song and extrapolating to entire catalog. Missing range and evolution.
Why it fails: Artists vary. Their most famous song may not be representative. Analysis from one track produces narrow profiles.
Fix: Analyze 3-5 tracks from different periods and modes. Look for both constants and variations.
5. The Technical Overdose
Pattern: Including so much technical detail that prompts become unusable. Every possible parameter specified.
Why it fails: AI generation systems can't process unlimited context. Overly detailed prompts get truncated or confuse the model.
Fix: Distill to 5-10 essential phrases. Prioritize what makes this artist distinct rather than comprehensive.
Integration Points
Inbound:
From listening to music you want to analyze
Outbound:
To AI music generation prompts
To lyric-diagnostic for complete song analysis
Complementary:
lyric-diagnostic: Lyrical analysis (words)
This skill: Musical analysis (sounds)
How to use it
Copy the folder
Take jwynia/musical-dna from the repository into ~/.claude/skills for personal
use, or into .claude/skills inside a project.
Check the name does not clash
The agent identifies a skill by the name field in its header. Two skills with the
same name cannot sit side by side — one of them will be ignored.