Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Deprecated: forwards to bmad-build-auto. Do not use unless invoked by name
.claude/skills/bmad-code-org-bmad-dev-auto/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 148% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 150% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -2% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 21% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -41% | 0% |
{project-root}/_bmad/custom/bmad-dev-auto.toml{project-root}/_bmad/custom/bmad-dev-auto.user.tomlbmad-dev-auto is deprecated. Redirecting to bmad-build-auto. Please use bmad-build-auto in the future., invoke bmad-build-auto exactly once with the user's original input verbatim, then execute no further steps in this shim.{project-root}/_bmad/custom/bmad-dev-auto.toml becomes {project-root}/_bmad/custom/bmad-build-auto.toml.{project-root}/_bmad/custom/bmad-dev-auto.user.toml becomes {project-root}/_bmad/custom/bmad-build-auto.user.toml.bmad-dev-auto is deprecated. Redirecting to bmad-build-auto. Please use bmad-build-auto in the future., invoke bmad-build-auto exactly once with the user's original input verbatim, then execute no further steps in this shim.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 8,286 | 9,796 | +18% | 1 | 1 | 0% | 858 | 2,132 | +148% | 0 | 0 | — |
case-02 | fail→fail | 5,071 | 7,825 | +54% | 1 | 1 | 0% | 452 | 1,542 | +241% | 0 | 0 | — |
case-03 | fail→pass | 5,437 | 10,629 | +95% | 1 | 1 | 0% | 827 | 2,064 | +150% | 0 | 0 | — |
case-04 | fail→pass | 23,111 | 3,667 | -84% | 1 | 1 | 0% | 1,046 | 1,023 | -2% | 0 | 0 | — |
case-05 | fail→pass | 12,311 | 4,896 | -60% | 1 | 1 | 0% | 1,077 | 1,305 | +21% | 0 | 0 | — |
case-06 | fail→pass | 11,378 | 5,186 | -54% | 1 | 1 | 0% | 1,889 | 1,120 | -41% | 0 | 0 | — |
case-07 | fail→pass | 8,574 | 4,554 | -47% | 1 | 1 | 0% | 1,269 | 950 | -25% | 0 | 0 | — |
case-08 | fail→fail | 8,554 | 17,677 | +107% | 1 | 1 | 0% | 1,336 | 3,749 | +181% | 0 | 0 | — |
case-09 | fail→fail | 13,423 | 9,508 | -29% | 1 | 1 | 0% | 2,076 | 1,958 | -6% | 0 | 0 | — |
case-10 | pass→pass | 5,984 | 4,903 | -18% | 1 | 1 | 0% | 1,022 | 1,150 | +13% | 0 | 0 | — |
case-11 | pass→pass | 5,521 | 5,588 | +1% | 1 | 1 | 0% | 882 | 937 | +6% | 0 | 0 | — |
case-12 | pass→pass | 12,866 | 7,790 | -39% | 1 | 1 | 0% | 2,100 | 1,824 | -13% | 0 | 0 | — |
case-13 | pass→pass | 33,388 | 7,212 | -78% | 1 | 1 | 0% | 1,705 | 1,623 | -5% | 0 | 0 | — |
case-14 | pass→pass | 10,627 | 9,483 | -11% | 1 | 1 | 0% | 1,701 | 934 | -45% | 0 | 0 | — |
case-15 | pass→pass | 19,387 | 4,049 | -79% | 1 | 1 | 0% | 1,917 | 1,116 | -42% | 0 | 0 | — |
case-16 | fail→pass | 6,824 | 3,683 | -46% | 1 | 1 | 0% | 1,004 | 1,006 | +0% | 0 | 0 | — |
case-17 | fail→pass | 23,653 | 8,344 | -65% | 1 | 1 | 0% | 1,727 | 1,236 | -28% | 0 | 0 | — |
case-18 | fail→pass | 20,148 | 3,854 | -81% | 1 | 1 | 0% | 2,105 | 917 | -56% | 0 | 0 | — |
case-19 | fail→pass | 10,596 | 3,832 | -64% | 1 | 1 | 0% | 1,561 | 1,089 | -30% | 0 | 0 | — |
case-20 | fail→pass | 11,451 | 5,146 | -55% | 1 | 1 | 0% | 1,465 | 1,282 | -12% | 0 | 0 | — |
case-21 | fail→pass | 10,105 | 6,264 | -38% | 1 | 1 | 0% | 1,430 | 1,359 | -5% | 0 | 0 | — |
case-22 | fail→fail | 26,814 | 3,396 | -87% | 1 | 1 | 0% | 3,644 | 947 | -74% | 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 +55 percentage points is the difference between those two pass rates over the 22 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.