Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Extract all chapters from an EPUB file into separate markdown files. Use when the user wants to split an EPUB into individual chapter files, extract EPUB chapters, or convert an ebook to separate markdown documents.
.claude/skills/aiskillstore-epub-chapter-extractor/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | -72% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -57% | 0% |
| case-13 | ✗→✓ | ▲ Improved | -20% | 0% |
| case-16 | ✗→✓ | ▲ Improved | 41% | 0% |
| case-18 | ✗→✓ | ▲ Improved | -55% | 0% |
Extract each chapter from an EPUB file into its own markdown file.
When the user wants to extract chapters from an EPUB, run the extraction script with uv:
bashcd ~/.claude/skills/epub-chapter-extractor && uv run --with ebooklib --with beautifulsoup4 --with html2text --with lxml python extract_chapters.py "/path/to/book.epub" [output_dir]
If output_dir is omitted, creates a folder named after the EPUB in the same directory.
User: "Extract chapters from /Users/yourname/Books/mybook.epub"
bashcd ~/.claude/skills/epub-chapter-extractor && uv run --with ebooklib --with beautifulsoup4 --with html2text --with lxml python extract_chapters.py "/Users/yourname/Books/mybook.epub"
Output files will be at /Users/yourname/Books/mybook/:
01_introduction.md02_chapter_one.mdAfter extraction, open the output folder:
bashopen /Users/yourname/Books/mybook
Each chapter file contains:
markdown# Chapter Title [Chapter content in markdown format]
Files are numbered for proper sorting: 01_, 02_, etc.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-08 | fail→fail | 12,643 | 5,798 | -54% | 1 | 1 | 0% | 2,521 | 552 | -78% | 0 | 0 | — |
case-01 | fail→fail | 11,875 | 5,619 | -53% | 1 | 1 | 0% | 2,476 | 654 | -74% | 0 | 0 | — |
case-02 | fail→fail | 9,535 | 5,388 | -43% | 1 | 1 | 0% | 2,077 | 641 | -69% | 0 | 0 | — |
case-03 | fail→pass | 8,896 | 1,523 | -83% | 1 | 1 | 0% | 1,948 | 547 | -72% | 0 | 0 | — |
case-04 | fail→fail | 9,355 | 4,417 | -53% | 1 | 1 | 0% | 1,887 | 593 | -69% | 0 | 0 | — |
case-05 | fail→pass | 9,267 | 5,243 | -43% | 1 | 1 | 0% | 1,684 | 726 | -57% | 0 | 0 | — |
case-06 | fail→fail | 13,609 | 5,038 | -63% | 1 | 1 | 0% | 2,817 | 606 | -78% | 0 | 0 | — |
case-07 | pass→fail | 10,714 | 4,807 | -55% | 1 | 1 | 0% | 2,217 | 575 | -74% | 0 | 0 | — |
case-09 | fail→fail | 14,467 | 5,225 | -64% | 1 | 1 | 0% | 2,998 | 539 | -82% | 0 | 0 | — |
case-10 | fail→fail | 8,520 | 6,315 | -26% | 1 | 1 | 0% | 1,796 | 752 | -58% | 0 | 0 | — |
case-11 | fail→fail | 12,261 | 5,014 | -59% | 1 | 1 | 0% | 2,305 | 634 | -72% | 0 | 0 | — |
case-12 | fail→fail | 5,077 | 5,824 | +15% | 1 | 1 | 0% | 231 | 654 | +183% | 0 | 0 | — |
case-13 | fail→pass | 5,082 | 2,224 | -56% | 1 | 1 | 0% | 843 | 677 | -20% | 0 | 0 | — |
case-14 | pass→pass | 4,545 | 1,776 | -61% | 1 | 1 | 0% | 649 | 531 | -18% | 0 | 0 | — |
case-15 | pass→pass | 3,589 | 1,866 | -48% | 1 | 1 | 0% | 534 | 598 | +12% | 0 | 0 | — |
case-16 | fail→pass | 2,296 | 1,687 | -27% | 1 | 1 | 0% | 365 | 514 | +41% | 0 | 0 | — |
case-17 | pass→fail | 9,892 | 7,497 | -24% | 1 | 1 | 0% | 2,037 | 839 | -59% | 0 | 0 | — |
case-18 | fail→pass | 10,910 | 3,079 | -72% | 1 | 1 | 0% | 1,920 | 868 | -55% | 0 | 0 | — |
case-19 | pass→fail | 16,858 | 6,006 | -64% | 1 | 1 | 0% | 3,556 | 671 | -81% | 0 | 0 | — |
case-20 | fail→fail | 6,506 | 10,051 | +54% | 1 | 1 | 0% | 1,324 | 2,261 | +71% | 0 | 0 | — |
case-21 | pass→pass | 5,865 | 11,688 | +99% | 1 | 1 | 0% | 1,073 | 2,637 | +146% | 0 | 0 | — |
case-22 | pass→pass | 10,132 | 11,004 | +9% | 1 | 1 | 0% | 1,892 | 2,426 | +28% | 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. 22 cases were attempted, and 10 counted toward the lift figure. The other 12 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 +9 percentage points is the difference between those two pass rates over the 10 comparable cases. 3 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.