Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Comprehensive PDF manipulation toolkit for extracting text and tables, creating new PDFs, merging/splitting documents, and handling forms. When Claude needs to fill in a PDF form or programmatically process, generate, or analyze PDF documents at scale.
.claude/skills/dokhacgiakhoa-pdf/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -61% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -68% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -75% | 0% |
| case-04 | ✓→✓ | ＝ Same ✓ | -22% | 0% |
| case-05 | ✓→✓ | ＝ Same ✓ | 14% | 0% |
This guide covers essential PDF processing operations using Python libraries and command-line tools. For advanced features, JavaScript libraries, and detailed examples, see reference.md. If you need to fill out a PDF form, read forms.md and follow its instructions.
pythonfrom pypdf import PdfReader, PdfWriter # Read a PDF reader = PdfReader("document.pdf") print(f"Pages: {len(reader.pages)}") # Extract text text = "" for page in reader.pages: text += page.extract_text()
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 10,108 | 6,927 | -31% | 1 | 1 | 0% | 1,744 | 673 | -61% | 0 | 0 | — |
case-02 | fail→pass | 18,682 | 7,033 | -62% | 1 | 1 | 0% | 2,325 | 741 | -68% | 0 | 0 | — |
case-03 | fail→pass | 28,004 | 9,129 | -67% | 1 | 1 | 0% | 4,394 | 1,095 | -75% | 0 | 0 | — |
case-04 | pass→pass | 11,538 | 7,850 | -32% | 1 | 1 | 0% | 1,231 | 955 | -22% | 0 | 0 | — |
case-05 | pass→pass | 9,314 | 2,109 | -77% | 1 | 1 | 0% | 671 | 766 | +14% | 0 | 0 | — |
case-06 | pass→pass | 17,223 | 9,324 | -46% | 1 | 1 | 0% | 2,039 | 1,165 | -43% | 0 | 0 | — |
case-07 | pass→pass | 12,645 | 8,109 | -36% | 1 | 1 | 0% | 2,487 | 962 | -61% | 0 | 0 | — |
case-08 | pass→pass | 8,436 | 7,908 | -6% | 1 | 1 | 0% | 620 | 926 | +49% | 0 | 0 | — |
case-09 | pass→pass | 17,916 | 12,052 | -33% | 1 | 1 | 0% | 2,180 | 1,612 | -26% | 0 | 0 | — |
case-10 | pass→pass | 14,954 | 9,658 | -35% | 1 | 1 | 0% | 1,656 | 1,262 | -24% | 0 | 0 | — |
case-11 | pass→pass | 13,995 | 7,315 | -48% | 1 | 1 | 0% | 1,554 | 804 | -48% | 0 | 0 | — |
case-12 | pass→pass | 14,703 | 7,318 | -50% | 1 | 1 | 0% | 1,824 | 880 | -52% | 0 | 0 | — |
case-13 | pass→pass | 15,075 | 6,886 | -54% | 1 | 1 | 0% | 1,924 | 749 | -61% | 0 | 0 | — |
case-14 | pass→pass | 15,672 | 8,358 | -47% | 1 | 1 | 0% | 1,648 | 1,024 | -38% | 0 | 0 | — |
case-15 | pass→pass | 4,415 | 7,391 | +67% | 1 | 1 | 0% | 784 | 765 | -2% | 0 | 0 | — |
case-16 | pass→pass | 10,752 | 8,413 | -22% | 1 | 1 | 0% | 1,201 | 1,004 | -16% | 0 | 0 | — |
case-17 | pass→pass | 4,240 | 7,166 | +69% | 1 | 1 | 0% | 707 | 754 | +7% | 0 | 0 | — |
case-18 | pass→pass | 9,326 | 8,078 | -13% | 1 | 1 | 0% | 910 | 849 | -7% | 0 | 0 | — |
case-19 | pass→pass | 15,008 | 2,565 | -83% | 1 | 1 | 0% | 1,791 | 861 | -52% | 0 | 0 | — |
case-20 | pass→pass | 20,371 | 6,897 | -66% | 1 | 1 | 0% | 3,262 | 1,808 | -45% | 0 | 0 | — |
case-21 | pass→pass | 13,847 | 6,973 | -50% | 1 | 1 | 0% | 1,659 | 710 | -57% | 0 | 0 | — |
case-22 | pass→pass | 11,962 | 7,490 | -37% | 1 | 1 | 0% | 1,125 | 790 | -30% | 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. The headline lift of +14 percentage points is the difference between those two pass rates over the 22 comparable cases.
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.