Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Build monthly analytics reports for this Python data analytics repo. Use when asked to analyze monthly sales, revenue, customer activity, retention, churn, or produce a summary report from processed CSV data.
.claude/skills/kellyclaw-ai-monthly-analytics-report/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-05 | ✗→✓ | ▲ Improved | -66% | 0% |
| case-14 | ✗→✓ | ▲ Improved | -34% | 0% |
| case-15 | ✗→✓ | ▲ Improved | -37% | 0% |
| case-16 | ✗→✓ | ▲ Improved | -12% | 0% |
| case-19 | ✗→✓ | ▲ Improved | -43% | 0% |
Use this skill when creating or updating a monthly business analytics report.
Expected files:
data/processed/sales_clean.csvdata/processed/customers_clean.csvRequired columns in sales_clean.csv:
order_idcustomer_idorder_dategross_revenuediscount_amountnet_revenuepandas.reports/charts/.reports/monthly/YYYY-MM.md.Before finishing:
bashpytest ruff check .
If the report numbers changed, mention which source files and date ranges were used.
The final report should include:
md# Monthly Analytics Report: YYYY-MM ## Executive Summary ## Key Metrics ## Revenue Trends ## Customer Activity ## Notes and Caveats
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 4,127 | 5,827 | +41% | 1 | 1 | 0% | 215 | 697 | +224% | 0 | 0 | — |
case-02 | fail→fail | 15,898 | 5,158 | -68% | 1 | 1 | 0% | 2,481 | 463 | -81% | 0 | 0 | — |
case-03 | fail→fail | 4,702 | 6,152 | +31% | 1 | 1 | 0% | 314 | 719 | +129% | 0 | 0 | — |
case-04 | pass→pass | 7,311 | 3,353 | -54% | 1 | 1 | 0% | 1,298 | 845 | -35% | 0 | 0 | — |
case-05 | fail→pass | 9,754 | 1,697 | -83% | 1 | 1 | 0% | 1,729 | 584 | -66% | 0 | 0 | — |
case-06 | pass→pass | 5,651 | 1,925 | -66% | 1 | 1 | 0% | 1,025 | 567 | -45% | 0 | 0 | — |
case-11 | fail→fail | 5,060 | 2,576 | -49% | 1 | 1 | 0% | 869 | 590 | -32% | 0 | 0 | — |
case-07 | fail→fail | 5,561 | 1,813 | -67% | 1 | 1 | 0% | 872 | 610 | -30% | 0 | 0 | — |
case-08 | fail→fail | 9,510 | 2,652 | -72% | 1 | 1 | 0% | 1,812 | 782 | -57% | 0 | 0 | — |
case-09 | fail→fail | 3,521 | 1,766 | -50% | 1 | 1 | 0% | 510 | 581 | +14% | 0 | 0 | — |
case-10 | pass→pass | 8,336 | 2,172 | -74% | 1 | 1 | 0% | 1,483 | 715 | -52% | 0 | 0 | — |
case-12 | pass→pass | 6,569 | 2,360 | -64% | 1 | 1 | 0% | 1,117 | 649 | -42% | 0 | 0 | — |
case-13 | fail→fail | 5,706 | 15,098 | +165% | 1 | 1 | 0% | 988 | 520 | -47% | 0 | 0 | — |
case-14 | fail→pass | 5,594 | 2,062 | -63% | 1 | 1 | 0% | 1,009 | 668 | -34% | 0 | 0 | — |
case-15 | fail→pass | 4,459 | 1,588 | -64% | 1 | 1 | 0% | 736 | 464 | -37% | 0 | 0 | — |
case-16 | fail→pass | 3,580 | 1,305 | -64% | 1 | 1 | 0% | 568 | 497 | -13% | 0 | 0 | — |
case-17 | pass→pass | 9,182 | 3,479 | -62% | 1 | 1 | 0% | 1,634 | 935 | -43% | 0 | 0 | — |
case-18 | pass→pass | 7,866 | 2,206 | -72% | 1 | 1 | 0% | 1,246 | 712 | -43% | 0 | 0 | — |
case-19 | fail→pass | 9,163 | 8,453 | -8% | 1 | 1 | 0% | 1,863 | 1,054 | -43% | 0 | 0 | — |
case-20 | fail→fail | 8,719 | 4,836 | -45% | 1 | 1 | 0% | 1,558 | 1,048 | -33% | 0 | 0 | — |
case-21 | pass→pass | 2,501 | 3,793 | +52% | 1 | 1 | 0% | 504 | 893 | +77% | 0 | 0 | — |
case-22 | pass→pass | 9,685 | 4,695 | -52% | 1 | 1 | 0% | 1,620 | 1,145 | -29% | 0 | 0 | — |
case-23 | pass→fail | 8,208 | 4,631 | -44% | 1 | 1 | 0% | 1,375 | 1,032 | -25% | 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 20 counted toward the lift figure. The other 3 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 +17 percentage points is the difference between those two pass rates over the 20 comparable cases. 1 case got worse with the skill loaded, and it is 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.
| Model | Method | Date | Lift |
|---|---|---|---|
| gemini-3.6-flash | verified | 8/3/2026 | +5% |
Other measured skills in the registry, with their headline benchmark lift.