Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Fonds, rules, and tools integration exemplar — read resource pools, apply governance rules, validate tool manifests, and run combined integration.
.claude/skills/docxology-template-pools-rules-tools/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-17 | ✗→✓ | ▲ Improved | 9% | 0% |
| case-01 | ✗→✓ | ▲ Improved | 203% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 24% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -8% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 12% | 0% |
Project-scoped skill for the in-repo exemplar at projects/templates/template_pools_rules_tools/. Load this when wiring or validating the three-resource architecture (fonds, rules, tools).
fonds/templates/,rules/templates/, and tools/templates/ without writing back to those trees.
None/empty + logged warnings).check_bibliography_overlap()) orproving a tool's entrypoint actually executes (not just that the file exists).
bash# From the repository root uv run pytest projects/templates/template_pools_rules_tools/tests/ \ --cov=projects/templates/template_pools_rules_tools/src --cov-fail-under=90 uv run python projects/templates/template_pools_rules_tools/scripts/01_validate_sources.py uv run python projects/templates/template_pools_rules_tools/scripts/02_run_integration.py uv run python projects/templates/template_pools_rules_tools/scripts/03_generate_manuscript.py uv run python projects/templates/template_pools_rules_tools/scripts/04_validate_strong_rules.py uv run python projects/templates/template_pools_rules_tools/scripts/05_generate_figures.py uv run python projects/templates/template_pools_rules_tools/scripts/z_generate_manuscript_variables.py uv run python scripts/pipeline/stage_02_analysis.py --project templates/template_pools_rules_tools uv run python scripts/pipeline/stage_03_render.py --project templates/template_pools_rules_tools uv run python scripts/runner/execute_pipeline.py --project templates/template_pools_rules_tools --core-only
fonds/, rules/, or tools/ from project src/.src/. Scripts orchestrate; readers stay standalone-friendly.pytest.mark.skipif when files are absent.output/ via pipeline stages; do not hand-edit copied deliverables.src/* resolves the monorepo root four levels above each module file.manuscript/preamble.md needs a closed fence. An unfenced file silently drops every multi-line directive (geometry, hypersetup, tcolorbox) via a whitelist fallback — only single-line \usepackage{}/\newcommand{} survive. Always verify with extract_preamble() after editing.{{UPPERCASE_KEY}}, never lowercase-dotted. infrastructure/rendering/manuscript_injection.py's regex only matches uppercase; a {{integration.foo}}-style token silently never resolves. The pipeline also requires scripts/z_generate_manuscript_variables.py to exist by that exact name — run_manuscript_variable_script() no-ops if it's absent.src/figures/figure_support.py::INTEGRATION_FIGURE_SPECS (re-exported from src/figures/figures.py) is the figure-provenance source. It does not auto-insert images — every figure still needs a real {#fig:label} block in the manuscript body, or its @fig: cross-reference renders as a literal ??.05_generate_figures.py binds the eight real manuscript labels to the PNGs produced in that run and writes output/figures/figure_registry.json; a missing referenced PNG aborts publication instead of yielding a partial registry.
noteBox/warningBox) break Beamer slide generation even though they render fine in the combined PDF — pandoc's per-section slide splitting can separate \begin/\end across frame boundaries. Prefer plain Markdown emphasis for callouts.run.sh/validate.sh subprocess tests are real but environment-guarded. timeout/gtimeout are absent on stock macOS — the code-executor tests skip there by design; don't "fix" a SKIPPED result by removing the skipif guard.AGENTS.mdREADME.mdTODO.mdfonds/AGENTS.mdrules/AGENTS.mdtools/AGENTS.md| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-17 | fail→pass | 16,202 | 9,940 | -39% | 1 | 1 | 0% | 1,871 | 2,042 | +9% | 0 | 0 | — |
case-01 | fail→pass | 30,026 | 12,962 | -57% | 1 | 1 | 0% | 994 | 3,009 | +203% | 0 | 0 | — |
case-02 | fail→pass | 40,209 | 15,235 | -62% | 1 | 1 | 0% | 2,462 | 3,050 | +24% | 0 | 0 | — |
case-03 | fail→pass | 28,167 | 18,818 | -33% | 1 | 1 | 0% | 3,838 | 3,523 | -8% | 0 | 0 | — |
case-04 | fail→pass | 16,930 | 10,241 | -40% | 1 | 1 | 0% | 1,860 | 2,091 | +12% | 0 | 0 | — |
case-05 | pass→pass | 19,270 | 11,572 | -40% | 1 | 1 | 0% | 2,392 | 2,392 | 0% | 0 | 0 | — |
case-06 | fail→pass | 19,523 | 12,499 | -36% | 1 | 1 | 0% | 2,205 | 2,432 | +10% | 0 | 0 | — |
case-07 | pass→pass | 14,015 | 8,931 | -36% | 1 | 1 | 0% | 1,468 | 1,907 | +30% | 0 | 0 | — |
case-08 | fail→pass | 10,978 | 10,397 | -5% | 1 | 1 | 0% | 980 | 2,015 | +106% | 0 | 0 | — |
case-09 | pass→pass | 16,132 | 9,279 | -42% | 1 | 1 | 0% | 1,816 | 1,931 | +6% | 0 | 0 | — |
case-10 | pass→pass | 14,038 | 8,907 | -37% | 1 | 1 | 0% | 1,639 | 1,911 | +17% | 0 | 0 | — |
case-11 | fail→pass | 19,592 | 11,948 | -39% | 1 | 1 | 0% | 2,382 | 2,485 | +4% | 0 | 0 | — |
case-12 | fail→pass | 16,296 | 8,106 | -50% | 1 | 1 | 0% | 1,989 | 1,596 | -20% | 0 | 0 | — |
case-13 | fail→pass | 13,864 | 7,668 | -45% | 1 | 1 | 0% | 1,578 | 1,585 | +0% | 0 | 0 | — |
case-14 | fail→pass | 10,086 | 6,997 | -31% | 1 | 1 | 0% | 847 | 1,555 | +84% | 0 | 0 | — |
case-15 | fail→pass | 13,730 | 7,343 | -47% | 1 | 1 | 0% | 1,473 | 1,624 | +10% | 0 | 0 | — |
case-16 | fail→pass | 18,739 | 10,198 | -46% | 1 | 1 | 0% | 2,369 | 2,108 | -11% | 0 | 0 | — |
case-18 | fail→pass | 15,157 | 11,507 | -24% | 1 | 1 | 0% | 1,719 | 2,299 | +34% | 0 | 0 | — |
case-19 | fail→pass | 19,130 | 15,348 | -20% | 1 | 1 | 0% | 2,329 | 2,828 | +21% | 0 | 0 | — |
case-20 | pass→pass | 19,396 | 22,612 | +17% | 1 | 1 | 0% | 2,719 | 4,309 | +58% | 0 | 0 | — |
case-21 | pass→pass | 17,283 | 18,268 | +6% | 1 | 1 | 0% | 2,166 | 3,645 | +68% | 0 | 0 | — |
case-22 | pass→pass | 17,780 | 14,971 | -16% | 1 | 1 | 0% | 2,356 | 3,207 | +36% | 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 21 counted toward the lift figure. The other 1 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 +68 percentage points is the difference between those two pass rates over the 21 comparable cases.
The publisher has shipped newer versions since this run, so these numbers describe v2, not the version currently listed.
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.
| Model | Method | Date | Lift |
|---|---|---|---|
| gemini-3.6-flash | verified | 8/4/2026 | +68% |
Other measured skills in the registry, with their headline benchmark lift.