Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Converts external documents (PDF, DOCX, PPTX, XLSX, HTML) into editable markdown. Use when ingesting external files for rewriting or project integration.
.claude/skills/athola-doc-importer/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | 22% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 17% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 156% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -10% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 266% | 0% |
Import external documents into editable markdown.
to convert into project documentation
rewriting or remediation
markdown documentation
tome:papersmemory-palace:knowledge-intake
scribe:doc-generatorremediation mode directly
Determine the source document:
Apply the leyline:document-conversion protocol:
After conversion, normalize the markdown:
# style, not setext underlines)leyline:markdown-formatting
headers/footers, watermarks, repeated logos)
Apply the leyline:content-sanitization checklist:
Write the converted markdown to the target location. Default: same directory as source, with .md extension. Ask the user for target path if ambiguous.
If the user wants polishing or rewriting:
Skill(scribe:doc-generator) in Remediationmode on the imported file
application, and quality gates
Offer this step. Do not assume the user wants remediation.
The imported markdown should:
# Title from the document title references(note: image files may need separate handling)
<!-- REVIEW: conversion artifact -->
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | pass→pass | 5,390 | 2,253 | -58% | 1 | 1 | 0% | 709 | 936 | +32% | 0 | 0 | — |
case-02 | fail→fail | 4,043 | 13,490 | +234% | 1 | 1 | 0% | 481 | 2,905 | +504% | 0 | 0 | — |
case-03 | fail→pass | 14,416 | 15,129 | +5% | 1 | 1 | 0% | 2,694 | 3,298 | +22% | 0 | 0 | — |
case-04 | fail→pass | 7,120 | 2,947 | -59% | 1 | 1 | 0% | 901 | 1,056 | +17% | 0 | 0 | — |
case-05 | pass→pass | 5,100 | 3,241 | -36% | 1 | 1 | 0% | 748 | 1,203 | +61% | 0 | 0 | — |
case-06 | fail→pass | 4,065 | 5,549 | +37% | 1 | 1 | 0% | 602 | 1,543 | +156% | 0 | 0 | — |
case-07 | fail→fail | 14,882 | 4,992 | -66% | 1 | 1 | 0% | 1,922 | 849 | -56% | 0 | 0 | — |
case-08 | pass→fail | 10,647 | 5,004 | -53% | 1 | 1 | 0% | 1,817 | 1,070 | -41% | 0 | 0 | — |
case-09 | pass→fail | 16,343 | 7,355 | -55% | 1 | 1 | 0% | 2,960 | 1,734 | -41% | 0 | 0 | — |
case-10 | fail→pass | 17,482 | 11,234 | -36% | 1 | 1 | 0% | 2,871 | 2,575 | -10% | 0 | 0 | — |
case-11 | fail→pass | 4,772 | 13,310 | +179% | 1 | 1 | 0% | 743 | 2,722 | +266% | 0 | 0 | — |
case-12 | fail→fail | 6,303 | 2,487 | -61% | 1 | 1 | 0% | 1,054 | 1,027 | -3% | 0 | 0 | — |
case-13 | fail→pass | 9,865 | 9,493 | -4% | 1 | 1 | 0% | 993 | 2,182 | +120% | 0 | 0 | — |
case-14 | pass→pass | 7,851 | 21,207 | +170% | 1 | 1 | 0% | 1,320 | 2,805 | +113% | 0 | 0 | — |
case-15 | fail→pass | 13,247 | 9,513 | -28% | 1 | 1 | 0% | 2,379 | 2,160 | -9% | 0 | 0 | — |
case-16 | pass→fail | 9,570 | 2,192 | -77% | 1 | 1 | 0% | 1,466 | 959 | -35% | 0 | 0 | — |
case-17 | pass→fail | 21,882 | 9,305 | -57% | 1 | 1 | 0% | 3,475 | 1,220 | -65% | 0 | 0 | — |
case-18 | pass→pass | 5,899 | 12,483 | +112% | 1 | 1 | 0% | 954 | 2,489 | +161% | 0 | 0 | — |
case-19 | pass→pass | 9,410 | 13,922 | +48% | 1 | 1 | 0% | 1,705 | 2,916 | +71% | 0 | 0 | — |
case-20 | fail→pass | 3,857 | 19,523 | +406% | 1 | 1 | 0% | 610 | 2,322 | +281% | 0 | 0 | — |
case-21 | fail→fail | 4,401 | 2,213 | -50% | 1 | 1 | 0% | 661 | 882 | +33% | 0 | 0 | — |
case-22 | fail→pass | 5,690 | 4,293 | -25% | 1 | 1 | 0% | 1,068 | 1,257 | +18% | 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 20 counted toward the lift figure. The other 2 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 +23 percentage points is the difference between those two pass rates over the 20 comparable cases. 4 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.