Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Template-native manuscript planning, outline, drafting, revision, formatting, citation check, and AI-use disclosure routing. USE WHEN the user asks to write, outline, revise, format, or prepare a paper inside the Research Project Template.
.claude/skills/docxology-template-academic-paper/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | 134% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 20% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -19% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -6% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -17% | 0% |
Template-native paper workflow. It uses project manuscript sources, generated variables, validation gates, and evidence registries rather than a copied external academic agent prompt.
docs/_generated/active_projects.md.projects/<project>/manuscript/, generated variables, figure captions, and citekeys; use [MATERIAL GAP] instead of fabricating missing evidence.[[VAR:...]], registry-backed labels, citekeys, and project-defined cross-reference tokens over hard-coded numbers.prose-quality) and tighten any flagged stock phrasing, em-dash over-use, or uniform sentence length. Treat it as advisory (revise, do not blindly delete); it is a deterministic clean-room distillation of the ARS writing-quality check, not a model judgment.bashuv run python -m infrastructure.validation.cli prerender projects/<project>/manuscript --repo-root . uv run python -m infrastructure.reference.citation validate projects/<project>/manuscript/references.bib uv run python -m infrastructure.reference.verification verify projects/<project>/manuscript/references.bib --live --as-of-year <year> uv run python -m infrastructure.validation.cli prose-quality projects/<project>/manuscript uv run python -m infrastructure.validation.cli markdown projects/<project>/manuscript --repo-root . --strict uv run python -m infrastructure.validation.cli integrity output/<project>/
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 3,356 | 5,864 | +75% | 1 | 1 | 0% | 543 | 1,078 | +99% | 0 | 0 | — |
case-02 | fail→fail | 10,893 | 6,305 | -42% | 1 | 1 | 0% | 1,809 | 1,674 | -7% | 0 | 0 | — |
case-03 | fail→pass | 6,837 | 6,132 | -10% | 1 | 1 | 0% | 525 | 1,226 | +134% | 0 | 0 | — |
case-04 | fail→pass | 8,140 | 4,524 | -44% | 1 | 1 | 0% | 1,246 | 1,500 | +20% | 0 | 0 | — |
case-05 | fail→pass | 12,271 | 4,930 | -60% | 1 | 1 | 0% | 1,977 | 1,610 | -19% | 0 | 0 | — |
case-06 | fail→pass | 10,145 | 5,656 | -44% | 1 | 1 | 0% | 1,664 | 1,556 | -6% | 0 | 0 | — |
case-07 | fail→pass | 7,549 | 3,077 | -59% | 1 | 1 | 0% | 1,513 | 1,258 | -17% | 0 | 0 | — |
case-08 | fail→pass | 9,715 | 2,042 | -79% | 1 | 1 | 0% | 1,656 | 1,180 | -29% | 0 | 0 | — |
case-09 | fail→pass | 8,760 | 1,842 | -79% | 1 | 1 | 0% | 1,461 | 1,084 | -26% | 0 | 0 | — |
case-10 | fail→pass | 15,684 | 2,682 | -83% | 1 | 1 | 0% | 2,839 | 1,248 | -56% | 0 | 0 | — |
case-11 | fail→pass | 15,914 | 2,244 | -86% | 1 | 1 | 0% | 2,974 | 1,231 | -59% | 0 | 0 | — |
case-12 | fail→pass | 7,899 | 1,811 | -77% | 1 | 1 | 0% | 1,260 | 981 | -22% | 0 | 0 | — |
case-13 | fail→pass | 9,012 | 3,962 | -56% | 1 | 1 | 0% | 1,615 | 1,390 | -14% | 0 | 0 | — |
case-14 | fail→pass | 13,295 | 3,949 | -70% | 1 | 1 | 0% | 2,014 | 1,448 | -28% | 0 | 0 | — |
case-15 | fail→pass | 12,444 | 5,705 | -54% | 1 | 1 | 0% | 2,094 | 1,669 | -20% | 0 | 0 | — |
case-16 | fail→pass | 13,747 | 8,468 | -38% | 1 | 1 | 0% | 2,169 | 2,256 | +4% | 0 | 0 | — |
case-17 | pass→pass | 6,502 | 2,437 | -63% | 1 | 1 | 0% | 947 | 1,164 | +23% | 0 | 0 | — |
case-18 | fail→pass | 11,176 | 4,853 | -57% | 1 | 1 | 0% | 1,950 | 1,629 | -16% | 0 | 0 | — |
case-19 | pass→pass | 8,906 | 2,038 | -77% | 1 | 1 | 0% | 1,383 | 1,102 | -20% | 0 | 0 | — |
case-20 | fail→fail | 2,365 | 6,422 | +172% | 1 | 1 | 0% | 333 | 1,815 | +445% | 0 | 0 | — |
case-21 | fail→fail | 4,368 | 7,907 | +81% | 1 | 1 | 0% | 703 | 2,319 | +230% | 0 | 0 | — |
case-22 | pass→pass | 8,488 | 11,147 | +31% | 1 | 1 | 0% | 1,273 | 2,734 | +115% | 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 +67 percentage points is the difference between those two pass rates over the 22 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.