Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Creates a new album with the correct directory structure and templates. Use IMMEDIATELY when the user says 'make a new album' or similar, before any discussion.
.claude/skills/bitwize-music-studio-new-album/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-05 | ✗→✓ | ▲ Improved | 1359% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 79% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 50% | 0% |
| case-15 | ✗→✓ | ▲ Improved | 77% | 0% |
| case-16 | ✗→✓ | ▲ Improved | -27% | 0% |
Input: $ARGUMENTS
Create a new album directory structure with all required files and templates.
You create the complete album directory structure based on config.
Expected formats:
<album-name> <genre> — standard album<album-name> documentary <genre> — true-story/documentary album (creates RESEARCH.md + SOURCES.md)Examples:
sample-album electronicmy-new-album hip-hopprotest-songs folkthe-heist documentary hip-hopValid genres: Any genre that has a directory under ${CLAUDE_PLUGIN_ROOT}/genres/. Use the slug form (lowercase, hyphenated) — e.g. deep-house, crust-punk, k-pop, hip-hop.
To check if a genre is valid, verify ${CLAUDE_PLUGIN_ROOT}/genres/{genre}/README.md exists.
Parsing logic:
documentary: album = arg1, genre = arg3, documentary = trueAfter parsing, if documentary flag was not set, ask: "Is this a documentary/true-story album? (This adds research and sources templates.)"
If arguments are missing, ask:
Usage: /new-album <album-name> <genre>
/new-album <album-name> documentary <genre>
Example: /new-album sample-album electronic
/new-album the-heist documentary hip-hop
/new-album night-drive deep-house
Genre must match a directory under genres/ (use slug form: deep-house, crust-punk, etc.)Call create_album_structure(album_slug, genre, documentary) — creates the complete directory structure with templates in one call.
{content_root}/artists/{artist}/albums/{genre}/{album-name}/tracks/ and promo/ directories with templatesdocumentary: true): also creates RESEARCH.md and SOURCES.md{created: bool, path: str, files: [...]}Note: Audio and documents directories are NOT created (those are created when needed by import-audio/import-art).
Report:
Created album: {album-name}
Location: {album_path}
Files created:
- README.md (album template)
- tracks/ (empty, ready for track files)
- promo/ (social media copy templates)
Next steps:
Option 1 - Interactive (Recommended):
Run /bitwize-music:album-conceptualizer to design your album concept
through the 7 Planning Phases.
Option 2 - Manual:
1. Edit README.md with your album concept
2. Create tracks with /import-track or manually in tracks/
Tip: For OST/soundtrack albums with a mix of vocal and instrumental
tracks, the album-conceptualizer will ask about the vocal/instrumental
split per track. Set `instrumental: true` in track frontmatter for
instrumental tracks — they skip the lyrics workflow and go directly
to /bitwize-music:suno-engineer.Config file missing:
Error: Config not found at ~/.bitwize-music/config.yaml
Run /configure to set up.Invalid genre:
Error: Invalid genre "{genre}"
No genre directory found at genres/{genre}/. Use a valid genre slug (e.g. hip-hop, deep-house, grindcore).
Check genres/INDEX.md for the full list.Album already exists (any genre — slugs are globally unique):
Error: Album slug '{album-name}' already exists under genre '{existing-genre}': {existing_path}
Album slugs are unique across all genres — choose a different name, or
rename the existing album with /bitwize-music:rename.Templates not found:
Error: Templates not found. Is the plugin installed correctly?
Expected at: ${CLAUDE_PLUGIN_ROOT}/templates//new-album sample-album electronicConfig has:
yamlpaths: content_root: ~/bitwize-music artist: name: bitwize
Result:
Created album: sample-album
Location: ~/bitwize-music/artists/bitwize/albums/electronic/sample-album/
Files created:
- README.md (album template)
- tracks/ (empty, ready for track files)
Next steps:
Option 1 - Interactive (Recommended):
Run /bitwize-music:album-conceptualizer to design your album concept
through the 7 Planning Phases.
Option 2 - Manual:
1. Edit README.md with your album concept
2. Create tracks with /import-track or manually in tracks/If user mentions this is a documentary or true-story album:
/new-album the-heist documentary hip-hopThe create_album_structure(album_slug, genre, documentary=true) call automatically creates RESEARCH.md and SOURCES.md from templates.
Report:
Created album: the-heist (documentary)
Location: ~/bitwize-music/artists/bitwize/albums/hip-hop/the-heist/
Files created:
- README.md (album template)
- RESEARCH.md (research template)
- SOURCES.md (sources template)
- tracks/ (empty, ready for track files)Wrong:
bash# Manual mkdir, config reading, template copying cat ~/.bitwize-music/config.yaml mkdir -p ~/music-projects/artists/bitwize/albums/... cp templates/album.md ...
Right:
# Single MCP call handles everything
create_album_structure(album_slug, genre, documentary)The MCP tool reads config, resolves paths, creates directories, and copies templates automatically.
Any genre with a directory under genres/ is valid. Use the most specific genre that fits:
bash/new-album my-album boom-bap # has its own genre directory /new-album my-album deep-house # specific subgenre /new-album my-album grindcore # specific subgenre /new-album my-album hip-hop # broad category also works
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-09 | pass→pass | 5,933 | 2,676 | -55% | 1 | 1 | 0% | 623 | 1,986 | +219% | 0 | 0 | — |
case-14 | pass→pass | 11,107 | 2,180 | -80% | 1 | 1 | 0% | 1,688 | 1,921 | +14% | 0 | 0 | — |
case-01 | fail→fail | 14,654 | 7,576 | -48% | 1 | 1 | 0% | 2,347 | 2,413 | +3% | 0 | 0 | — |
case-02 | fail→fail | 30,101 | 7,460 | -75% | 1 | 1 | 0% | 2,936 | 1,924 | -34% | 0 | 0 | — |
case-03 | fail→fail | 18,977 | 4,070 | -79% | 1 | 1 | 0% | 3,137 | 2,134 | -32% | 0 | 0 | — |
case-04 | pass→pass | 10,816 | 6,793 | -37% | 1 | 1 | 0% | 1,725 | 2,646 | +53% | 0 | 0 | — |
case-05 | fail→pass | 2,080 | 9,392 | +352% | 1 | 1 | 0% | 224 | 3,269 | +1359% | 0 | 0 | — |
case-06 | fail→fail | 15,223 | 5,266 | -65% | 1 | 1 | 0% | 2,101 | 1,861 | -11% | 0 | 0 | — |
case-07 | fail→fail | 4,031 | 6,488 | +61% | 1 | 1 | 0% | 584 | 1,903 | +226% | 0 | 0 | — |
case-08 | fail→fail | 8,497 | 8,589 | +1% | 1 | 1 | 0% | 1,145 | 2,953 | +158% | 0 | 0 | — |
case-10 | fail→fail | 14,003 | 7,866 | -44% | 1 | 1 | 0% | 2,146 | 2,138 | -0% | 0 | 0 | — |
case-11 | fail→pass | 7,321 | 3,788 | -48% | 1 | 1 | 0% | 1,299 | 2,331 | +79% | 0 | 0 | — |
case-12 | fail→pass | 9,050 | 3,111 | -66% | 1 | 1 | 0% | 1,389 | 2,079 | +50% | 0 | 0 | — |
case-13 | pass→pass | 7,977 | 1,987 | -75% | 1 | 1 | 0% | 1,372 | 1,877 | +37% | 0 | 0 | — |
case-15 | fail→pass | 6,945 | 3,191 | -54% | 1 | 1 | 0% | 1,208 | 2,141 | +77% | 0 | 0 | — |
case-16 | fail→pass | 19,004 | 4,227 | -78% | 1 | 1 | 0% | 2,932 | 2,146 | -27% | 0 | 0 | — |
case-17 | fail→pass | 12,758 | 3,469 | -73% | 1 | 1 | 0% | 1,997 | 2,149 | +8% | 0 | 0 | — |
case-18 | fail→pass | 8,255 | 3,118 | -62% | 1 | 1 | 0% | 1,378 | 2,103 | +53% | 0 | 0 | — |
case-23 | pass→pass | 11,788 | 2,452 | -79% | 1 | 1 | 0% | 1,758 | 1,939 | +10% | 0 | 0 | — |
case-19 | fail→pass | 19,603 | 2,292 | -88% | 1 | 1 | 0% | 1,981 | 1,908 | -4% | 0 | 0 | — |
case-20 | fail→fail | 13,284 | 5,243 | -61% | 1 | 1 | 0% | 1,897 | 2,586 | +36% | 0 | 0 | — |
case-21 | fail→pass | 11,637 | 3,176 | -73% | 1 | 1 | 0% | 1,763 | 2,141 | +21% | 0 | 0 | — |
case-22 | fail→pass | 10,097 | 3,462 | -66% | 1 | 1 | 0% | 1,545 | 2,167 | +40% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 23 cases were attempted, and 19 counted toward the lift figure. The other 4 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +43 percentage points is the difference between those two pass rates over the 19 comparable cases. 2 cases got worse with the skill loaded, and they are included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.