Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Lightweight Google Sheets integration with standalone OAuth authentication. No MCP server required. Full read/write access.
.claude/skills/google-sheets-automation/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | -34% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -19% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 7% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 17% | 0% |
| case-16 | ✗→✓ | ▲ Improved | 467% | 0% |
This package contains instructions, not an OAuth client or executable integration. Discover the host's available tools and their actual schemas first. Authentication, token storage, account support and permissions are provided by that integration; do not assume automatic login or a particular keyring implementation.
If no suitable integration is available, prepare the content or an explicit implementation plan and report that live access is unavailable. Do not invent a local command or claim a remote edit succeeded.
Update a supplied sales range using literal values. Preview the exact rows and range, perform the authorized update, then compare read-back values and totals while verifying adjacent formulas remain intact.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 10,101 | 4,980 | -51% | 1 | 1 | 0% | 1,419 | 1,488 | +5% | 0 | 0 | — |
case-02 | fail→fail | 6,558 | 6,757 | +3% | 1 | 1 | 0% | 1,125 | 2,429 | +116% | 0 | 0 | — |
case-03 | fail→pass | 11,845 | 2,867 | -76% | 1 | 1 | 0% | 2,296 | 1,515 | -34% | 0 | 0 | — |
case-04 | fail→pass | 10,275 | 1,934 | -81% | 1 | 1 | 0% | 1,801 | 1,453 | -19% | 0 | 0 | — |
case-05 | fail→pass | 9,105 | 2,021 | -78% | 1 | 1 | 0% | 1,418 | 1,523 | +7% | 0 | 0 | — |
case-06 | fail→fail | 6,657 | 4,435 | -33% | 1 | 1 | 0% | 1,204 | 1,385 | +15% | 0 | 0 | — |
case-07 | fail→fail | 5,422 | 5,083 | -6% | 1 | 1 | 0% | 1,103 | 1,422 | +29% | 0 | 0 | — |
case-08 | fail→fail | 8,096 | 4,541 | -44% | 1 | 1 | 0% | 1,534 | 1,324 | -14% | 0 | 0 | — |
case-09 | pass→pass | 9,217 | 4,149 | -55% | 1 | 1 | 0% | 1,996 | 1,889 | -5% | 0 | 0 | — |
case-10 | fail→fail | 8,167 | 4,471 | -45% | 1 | 1 | 0% | 653 | 1,444 | +121% | 0 | 0 | — |
case-11 | fail→pass | 12,139 | 6,330 | -48% | 1 | 1 | 0% | 1,589 | 1,852 | +17% | 0 | 0 | — |
case-12 | fail→fail | 6,780 | 5,970 | -12% | 1 | 1 | 0% | 1,255 | 1,598 | +27% | 0 | 0 | — |
case-13 | fail→fail | 7,389 | 6,825 | -8% | 1 | 1 | 0% | 1,546 | 1,585 | +3% | 0 | 0 | — |
case-14 | fail→fail | 6,928 | 6,681 | -4% | 1 | 1 | 0% | 1,403 | 1,682 | +20% | 0 | 0 | — |
case-15 | fail→fail | 7,677 | 4,481 | -42% | 1 | 1 | 0% | 1,563 | 1,390 | -11% | 0 | 0 | — |
case-16 | fail→pass | 3,019 | 6,851 | +127% | 1 | 1 | 0% | 422 | 2,393 | +467% | 0 | 0 | — |
case-17 | fail→fail | 3,230 | 4,231 | +31% | 1 | 1 | 0% | 470 | 1,360 | +189% | 0 | 0 | — |
case-18 | fail→fail | 8,079 | 7,064 | -13% | 1 | 1 | 0% | 1,667 | 1,561 | -6% | 0 | 0 | — |
case-19 | fail→fail | 8,508 | 4,824 | -43% | 1 | 1 | 0% | 1,845 | 2,009 | +9% | 0 | 0 | — |
case-20 | fail→pass | 9,538 | 2,575 | -73% | 1 | 1 | 0% | 1,805 | 1,594 | -12% | 0 | 0 | — |
case-21 | fail→fail | 7,898 | 6,134 | -22% | 1 | 1 | 0% | 1,608 | 1,560 | -3% | 0 | 0 | — |
case-22 | fail→fail | 5,531 | 7,965 | +44% | 1 | 1 | 0% | 996 | 1,803 | +81% | 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 9 counted toward the lift figure. The other 13 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 +27 percentage points is the difference between those two pass rates over the 9 comparable cases. 6 cases got worse with the skill loaded, and they are included in that figure.
The publisher has shipped newer versions since this run, so these numbers describe v1, 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.
Other measured skills in the registry, with their headline benchmark lift.