| name | influence-manage |
| description | Manage the influences library database (influences_library.csv) for genetic algorithm mutations. Use when user wants to add influences, search for mutation ideas, track what's been used, or explore new sonic territory for prompt generation. |
| allowed-tools | Bash, Read, Write |
Influence Management Skill
This skill manages the influences library database containing 46 musical influences used for the genetic algorithm's mutation strategy (the 20% new DNA).
What it does
- Lists all influences (with optional filtering by category, status)
- Searches influences by text (name, elements, adaptation notes)
- Shows detailed information for a specific influence
- Adds new influences to the library
- Updates influences (marks as used, changes status)
- Shows statistics (count by category, by status)
- Helps track the evolution from Unexplored → Tested → Proven/Avoid
When to use
- User wants to add a new musical influence
- User asks "what mutation ideas do we have?"
- User wants to find influences by category (e.g., "show me acoustic instruments")
- User asks "what influences haven't we used yet?"
- User wants to mark an influence as used in a prompt
- User wants to track which influences are proven vs should be avoided
- User is generating new prompts and needs mutation inspiration
Influence Lifecycle
Influences progress through these states:
- Unexplored (initial state) - Not yet tried in any prompt
- Tested - Used in at least one generated prompt, awaiting user feedback
- Proven - User rated a prompt using this influence as excellent/very good
- Avoid - User feedback suggests this influence doesn't work well
Categories
The library organizes influences into these categories:
- Acoustic Instruments (14) - Dulcimer, harp, autoharp, celesta, steel pan, marimba, vibraphone, glockenspiel, mandolin, bouzouki, oud, sitar, lap steel, EBow guitar
- Electronic Effects (9) - Dub delay, tape echo, reverb plates, analog synth arpeggios, Mellotron, granular synthesis, ambient drones, vinyl crackle, tape saturation
- Rhythmic Elements (6) - Swing quantization, shuffle grooves, polyrhythmic layers, odd time signatures, clave patterns, gamelan rhythms
- Textures (7) - Field recordings (rain/forest/ocean), harmonic drones, sustained tones, melodic percussion, tuned drums
- Genre Influences (10) - Surf rock, krautrock, fourth world, Balearic beat, minimalism, space music, ambient dub, lo-fi hip hop, exotica, kosmische musik
Usage examples
cd .claude/skills/influence-manage/scripts
source ../../venv/bin/activate
python manage_influences.py --list
python manage_influences.py --list --category "Acoustic Instruments"
python manage_influences.py --list --status "Unexplored"
python manage_influences.py --search "saxophone"
python manage_influences.py --search "delay"
python manage_influences.py --search "gamelan"
python manage_influences.py --show 1
python manage_influences.py --show 12
python manage_influences.py --add \
--category "Acoustic Instruments" \
--name "Hang Drum" \
--elements "Meditative metallic tones, melodic percussion, harmonic overtones" \
--avoid "Busy melodic patterns, aggressive playing" \
--notes "Perfect for theta states, use gentle patterns for ambient texture"
python manage_influences.py --mark-used 12 --prompt-ids "45,46"
python manage_influences.py --update-status 12 --status "Proven"
python manage_influences.py --stats
Integration with Genetic Algorithm Workflow
When generating new prompts (using prompt-generate skill), this skill helps with the 20% Mutation strategy:
Finding Mutation Ideas
-
Random exploration: List unexplored influences by category
python manage_influences.py --list --status "Unexplored" --category "Electronic Effects"
-
Thematic search: Search for influences matching the time block mood
python manage_influences.py --search "theta"
python manage_influences.py --search "gentle"
python manage_influences.py --search "focus"
python manage_influences.py --search "harmonic"
-
Proven winners: Find influences that worked well before
python manage_influences.py --list --status "Proven"
After Generating & Testing Prompts
-
Mark as used: Track which influences were applied
python manage_influences.py --mark-used 7 --prompt-ids "51,52,53"
-
Update status based on ratings:
- If user rates prompt as excellent/very good → mark influence as "Proven"
- If user rates prompt poorly → mark influence as "Avoid"
- Otherwise → mark as "Tested"
python manage_influences.py --update-status 7 --status "Proven"
Key Principles
✅ DO:
- Use influences to inject diversity into prompt generation
- Track which influences work well vs should be avoided
- Mark influences as used to see what's fresh vs overused
- Search the library before adding duplicates
- Use adaptation notes to guide how to apply the influence
❌ DON'T:
- Add influences without clear "Elements_To_Use" and "Elements_To_Avoid"
- Forget to update status after testing prompts
- Use influences marked as "Avoid"
- Apply influences that don't fit the time block's purpose (e.g., aggressive sounds in theta-inducing morning warmup)
Example: Using Influences for Midday Refresh
Goal: Generate 6 prompts for Midday Refresh time block (20% mutations = ~1-2 prompts)
Step 1: Find suitable mutation candidates
python manage_influences.py --search "refresh"
python manage_influences.py --search "alert"
python manage_influences.py --list --category "Rhythmic Elements"
Step 2: Review candidates (Influence #8 - Glockenspiel)
- Elements to use: "Bright metallic sparkle, gentle high tones, crystalline clarity"
- Elements to avoid: "Dominant melodies, march-like patterns"
- Adaptation notes: "Subtle high-frequency accents for alertness during afternoon slump"
- Status: Unexplored
Step 3: Generate prompt using this influence
- Add glockenspiel as subtle accent in an electronic/acoustic hybrid
- Keep it as texture, not melody
- Target gentle alertness, not harsh stimulation
Step 4: After testing in Suno
python manage_influences.py --mark-used 8 --prompt-ids "41"
python manage_influences.py --update-status 8 --status "Proven"
Files used
scripts/manage_influences.py - Main management script (embedded in this skill)
scripts/csv_utils.py - CSV utilities for influences_library.csv (embedded in this skill)
../../../../influences_library.csv - Influences database (project root)
Related skills
- Use with
prompt-generate to apply the 20% mutation strategy
- Use with
prompt-rate to track which influences led to excellent prompts
- Use with
prompt-show to see which influences were used in a prompt
Current Library Status
As of 2025-11-07:
- Total influences: 46
- Status: All "Unexplored" (virgin territory!)
- Categories: Acoustic (14), Electronic (9), Rhythmic (6), Textures (7), Genre (10)
The library is a fresh canvas waiting to inject new sonic DNA into the prompt library.