Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Processes PDF files. Extracts text and tables, fills forms, merges and splits documents, batch-processes files, converts to images, and generates PDFs programmatically. Use when working with .pdf files. Do NOT use for Word documents, spreadsheets, or presentations.
.claude/skills/telagod-processing-pdfs/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-23 | ✗→✓ | ▲ Improved | -11% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -8% | 0% |
| case-22 | ✓→✓ | ＝ Same ✓ | 15% | 0% |
| case-21 | ✓→✓ | ＝ Same ✓ | -3% | 0% |
| case-01 | ✓→✓ | ＝ Same ✓ | -9% | 0% |
Essential PDF operations using Python libraries and CLI tools.
| Task | Best Tool | Reference | |------|-----------|-----------| | Merge / split / metadata / rotate | pypdf | recipes.md | | Extract text (layout preserved) | pdfplumber | recipes.md | | Extract tables | pdfplumber | recipes.md | | Create new PDF | reportlab | recipes.md | | Batch CLI ops | qpdf / pdftk | recipes.md | | OCR scanned PDFs | pytesseract + pdf2image | advanced.md | | Add watermark / extract images / encrypt | pypdf / pdfimages | advanced.md | | Fill PDF forms | pdf-lib / pypdf | FORMS.md | | Advanced pypdfium2 / pdf-lib JS | — | REFERENCE.md |
pythonfrom pypdf import PdfReader reader = PdfReader("document.pdf") print(f"Pages: {len(reader.pages)}") text = "".join(page.extract_text() for page in reader.pages)
| Library | Use for | |---------|---------| | pypdf | Merge, split, metadata, encryption, rotation | | pdfplumber | Text extraction with layout, tables | | reportlab | Generate PDFs programmatically | | pdf2image + pytesseract | OCR scanned documents | | qpdf / pdftk (CLI) | Batch ops, no Python needed |
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-22 | pass→pass | 9,077 | 7,339 | -19% | 1 | 1 | 0% | 1,574 | 1,809 | +15% | 0 | 0 | — |
case-23 | fail→pass | 25,539 | 16,124 | -37% | 1 | 1 | 0% | 3,668 | 3,273 | -11% | 0 | 0 | — |
case-21 | pass→pass | 14,640 | 10,943 | -25% | 1 | 1 | 0% | 2,441 | 2,364 | -3% | 0 | 0 | — |
case-01 | pass→pass | 9,081 | 5,430 | -40% | 1 | 1 | 0% | 1,639 | 1,492 | -9% | 0 | 0 | — |
case-02 | pass→pass | 15,256 | 8,817 | -42% | 1 | 1 | 0% | 2,509 | 2,086 | -17% | 0 | 0 | — |
case-03 | pass→pass | 15,022 | 5,238 | -65% | 1 | 1 | 0% | 2,366 | 1,325 | -44% | 0 | 0 | — |
case-04 | pass→pass | 18,442 | 11,938 | -35% | 1 | 1 | 0% | 2,836 | 2,820 | -1% | 0 | 0 | — |
case-05 | pass→pass | 15,725 | 7,834 | -50% | 1 | 1 | 0% | 2,676 | 1,787 | -33% | 0 | 0 | — |
case-06 | pass→pass | 14,738 | 7,531 | -49% | 1 | 1 | 0% | 2,504 | 1,848 | -26% | 0 | 0 | — |
case-07 | pass→pass | 10,340 | 3,289 | -68% | 1 | 1 | 0% | 1,806 | 1,025 | -43% | 0 | 0 | — |
case-08 | pass→pass | 8,572 | 5,246 | -39% | 1 | 1 | 0% | 1,381 | 1,323 | -4% | 0 | 0 | — |
case-09 | pass→pass | 13,215 | 8,752 | -34% | 1 | 1 | 0% | 2,168 | 1,960 | -10% | 0 | 0 | — |
case-10 | pass→pass | 2,394 | 2,438 | +2% | 1 | 1 | 0% | 382 | 879 | +130% | 0 | 0 | — |
case-11 | fail→pass | 12,963 | 8,295 | -36% | 1 | 1 | 0% | 2,186 | 2,002 | -8% | 0 | 0 | — |
case-12 | pass→pass | 7,154 | 5,157 | -28% | 1 | 1 | 0% | 1,284 | 1,434 | +12% | 0 | 0 | — |
case-13 | pass→pass | 7,403 | 5,559 | -25% | 1 | 1 | 0% | 1,358 | 1,517 | +12% | 0 | 0 | — |
case-14 | pass→pass | 9,652 | 7,267 | -25% | 1 | 1 | 0% | 1,705 | 1,780 | +4% | 0 | 0 | — |
case-15 | pass→pass | 6,435 | 4,560 | -29% | 1 | 1 | 0% | 1,219 | 1,386 | +14% | 0 | 0 | — |
case-16 | pass→pass | 6,884 | 5,921 | -14% | 1 | 1 | 0% | 1,049 | 1,384 | +32% | 0 | 0 | — |
case-17 | pass→pass | 6,375 | 3,754 | -41% | 1 | 1 | 0% | 856 | 1,017 | +19% | 0 | 0 | — |
case-18 | pass→pass | 7,808 | 3,554 | -54% | 1 | 1 | 0% | 1,393 | 1,188 | -15% | 0 | 0 | — |
case-19 | pass→pass | 14,712 | 9,894 | -33% | 1 | 1 | 0% | 1,950 | 2,250 | +15% | 0 | 0 | — |
case-20 | pass→pass | 8,758 | 5,573 | -36% | 1 | 1 | 0% | 1,549 | 1,486 | -4% | 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. The headline lift of +9 percentage points is the difference between those two pass rates over the 23 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.