Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Run validation CLI, prerender, markdown/PDF/integrity gates, and QA workflows for the Research Project Template. USE WHEN validate manuscript, check PDF for ?? refs, prerender gate, link checker, output integrity, or pre-commit validation — even without validation_quality prompt.
.claude/skills/docxology-template-validation-quality/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | -31% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 103% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 798% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -24% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -45% | 0% |
Prefer actual CLI entrypoints in CLAUDE.md — not illustrative Python stubs.
output/<project>/.infrastructure.validation.cli markdown and/or prerender.infrastructure.reference.citation validate on .bib.infrastructure.validation.cli links --repo-root .cli pdf output/<project>/pdf/cli integrity output/<project>/bashuv run python -m infrastructure.validation.cli prerender projects/<project>/manuscript --repo-root . uv run python -m infrastructure.validation.cli markdown projects/<project>/manuscript --repo-root . --strict uv run python -m infrastructure.reference.citation validate projects/<project>/manuscript/references.bib uv run python -m infrastructure.validation.cli links --repo-root . uv run python -m infrastructure.validation.cli pdf output/<project>/pdf/ uv run python -m infrastructure.validation.cli integrity output/<project>/ uv run python -m infrastructure.prose.cli report projects/<project>/manuscript
infrastructure/validation/AGENTS.md| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 4,078 | 7,366 | +81% | 1 | 1 | 0% | 241 | 971 | +303% | 0 | 0 | — |
case-02 | fail→fail | 6,274 | 5,885 | -6% | 1 | 1 | 0% | 197 | 1,018 | +417% | 0 | 0 | — |
case-03 | fail→fail | 1,992 | 5,990 | +201% | 1 | 1 | 0% | 225 | 943 | +319% | 0 | 0 | — |
case-04 | fail→pass | 12,269 | 4,445 | -64% | 1 | 1 | 0% | 2,051 | 1,411 | -31% | 0 | 0 | — |
case-05 | fail→pass | 5,066 | 5,879 | +16% | 1 | 1 | 0% | 856 | 1,737 | +103% | 0 | 0 | — |
case-06 | fail→pass | 4,195 | 7,176 | +71% | 1 | 1 | 0% | 198 | 1,778 | +798% | 0 | 0 | — |
case-22 | pass→pass | 2,234 | 3,861 | +73% | 1 | 1 | 0% | 347 | 1,206 | +248% | 0 | 0 | — |
case-07 | fail→pass | 10,085 | 4,538 | -55% | 1 | 1 | 0% | 1,768 | 1,337 | -24% | 0 | 0 | — |
case-08 | fail→pass | 16,044 | 4,677 | -71% | 1 | 1 | 0% | 2,641 | 1,445 | -45% | 0 | 0 | — |
case-09 | fail→pass | 9,789 | 2,012 | -79% | 1 | 1 | 0% | 1,730 | 809 | -53% | 0 | 0 | — |
case-10 | fail→pass | 12,212 | 6,670 | -45% | 1 | 1 | 0% | 2,188 | 1,894 | -13% | 0 | 0 | — |
case-11 | fail→fail | 8,057 | 6,982 | -13% | 1 | 1 | 0% | 1,393 | 826 | -41% | 0 | 0 | — |
case-12 | pass→pass | 7,366 | 3,096 | -58% | 1 | 1 | 0% | 1,307 | 977 | -25% | 0 | 0 | — |
case-13 | pass→pass | 5,693 | 4,322 | -24% | 1 | 1 | 0% | 882 | 1,357 | +54% | 0 | 0 | — |
case-14 | fail→pass | 11,559 | 2,928 | -75% | 1 | 1 | 0% | 1,835 | 993 | -46% | 0 | 0 | — |
case-15 | fail→fail | 12,081 | 7,064 | -42% | 1 | 1 | 0% | 2,383 | 1,743 | -27% | 0 | 0 | — |
case-16 | pass→pass | 9,931 | 4,533 | -54% | 1 | 1 | 0% | 1,340 | 1,371 | +2% | 0 | 0 | — |
case-17 | pass→pass | 9,842 | 3,229 | -67% | 1 | 1 | 0% | 1,775 | 1,079 | -39% | 0 | 0 | — |
case-18 | fail→pass | 12,565 | 4,920 | -61% | 1 | 1 | 0% | 1,344 | 1,371 | +2% | 0 | 0 | — |
case-19 | fail→pass | 9,572 | 2,181 | -77% | 1 | 1 | 0% | 1,708 | 829 | -51% | 0 | 0 | — |
case-20 | fail→pass | 6,242 | 1,473 | -76% | 1 | 1 | 0% | 816 | 710 | -13% | 0 | 0 | — |
case-21 | pass→pass | 22,337 | 9,815 | -56% | 1 | 1 | 0% | 2,377 | 2,140 | -10% | 0 | 0 | — |
case-23 | pass→pass | 5,390 | 4,149 | -23% | 1 | 1 | 0% | 1,085 | 1,069 | -1% | 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, and 18 counted toward the lift figure. The other 5 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 +48 percentage points is the difference between those two pass rates over the 18 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.