Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Review code changes with several independent reviewers in parallel, then triage and present the findings. Use when the user says "run code review" or "review this code"
.claude/skills/bmad-code-org-bmad-code-review/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | 45% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -47% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 61% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 42% | 0% |
| case-12 | ✗→✓ | ▲ Improved | -53% | 0% |
Run the following command exactly once without changing the current working directory. Replace {project-root} with the absolute path to the project root and {skill-root} with the absolute path to this skill's directory:
bashuv run --no-cache "{project-root}/_bmad/scripts/render_skill.py" --project-root "{project-root}" --skill "{skill-root}"
quick or thorough), append --set workflow.review=<value> to the command.workflow.md instruction printed to stdout.bmad skill's setup, installing bmad first if you do not have it (npx skills add bmad-code-org/BMAD-METHOD --skill bmad), then run the command above once more.uv being unavailable), report the command output and HALT. Do not run any workflow source directly.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 2,682 | 7,875 | +194% | 1 | 1 | 0% | 278 | 494 | +78% | 0 | 0 | — |
case-02 | fail→fail | 17,358 | 7,874 | -55% | 1 | 1 | 0% | 2,547 | 544 | -79% | 0 | 0 | — |
case-03 | fail→fail | 9,157 | 12,581 | +37% | 1 | 1 | 0% | 1,219 | 630 | -48% | 0 | 0 | — |
case-04 | fail→pass | 12,684 | 13,343 | +5% | 1 | 1 | 0% | 1,022 | 1,485 | +45% | 0 | 0 | — |
case-05 | fail→pass | 10,351 | 4,859 | -53% | 1 | 1 | 0% | 1,692 | 902 | -47% | 0 | 0 | — |
case-06 | fail→pass | 5,270 | 4,946 | -6% | 1 | 1 | 0% | 675 | 1,089 | +61% | 0 | 0 | — |
case-07 | fail→fail | 7,075 | 2,559 | -64% | 1 | 1 | 0% | 969 | 518 | -47% | 0 | 0 | — |
case-08 | fail→fail | 6,636 | 13,455 | +103% | 1 | 1 | 0% | 265 | 1,007 | +280% | 0 | 0 | — |
case-09 | fail→fail | 13,717 | 14,197 | +3% | 1 | 1 | 0% | 1,776 | 1,141 | -36% | 0 | 0 | — |
case-10 | fail→pass | 8,461 | 16,788 | +98% | 1 | 1 | 0% | 1,326 | 1,879 | +42% | 0 | 0 | — |
case-11 | fail→fail | 18,266 | 11,536 | -37% | 1 | 1 | 0% | 2,183 | 1,182 | -46% | 0 | 0 | — |
case-12 | fail→pass | 22,459 | 14,110 | -37% | 1 | 1 | 0% | 1,868 | 871 | -53% | 0 | 0 | — |
case-13 | fail→pass | 20,533 | 2,311 | -89% | 1 | 1 | 0% | 3,332 | 517 | -84% | 0 | 0 | — |
case-14 | fail→fail | 9,046 | 12,931 | +43% | 1 | 1 | 0% | 1,242 | 721 | -42% | 0 | 0 | — |
case-15 | pass→fail | 11,888 | 17,974 | +51% | 1 | 1 | 0% | 1,673 | 2,203 | +32% | 0 | 0 | — |
case-16 | fail→pass | 9,654 | 11,161 | +16% | 1 | 1 | 0% | 1,326 | 798 | -40% | 0 | 0 | — |
case-17 | pass→pass | 8,885 | 3,093 | -65% | 1 | 1 | 0% | 1,218 | 679 | -44% | 0 | 0 | — |
case-18 | fail→pass | 9,301 | 2,768 | -70% | 1 | 1 | 0% | 1,729 | 555 | -68% | 0 | 0 | — |
case-19 | pass→pass | 14,582 | 20,082 | +38% | 1 | 1 | 0% | 1,085 | 2,267 | +109% | 0 | 0 | — |
case-20 | pass→pass | 5,706 | 25,754 | +351% | 1 | 1 | 0% | 783 | 4,781 | +511% | 0 | 0 | — |
case-21 | pass→pass | 3,380 | 16,886 | +400% | 1 | 1 | 0% | 404 | 1,644 | +307% | 0 | 0 | — |
case-22 | pass→fail | 11,590 | 22,010 | +90% | 1 | 1 | 0% | 2,003 | 610 | -70% | 0 | 0 | — |
case-23 | pass→pass | 16,085 | 23,611 | +47% | 1 | 1 | 0% | 2,361 | 4,288 | +82% | 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 15 counted toward the lift figure. The other 8 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 +26 percentage points is the difference between those two pass rates over the 15 comparable cases. 2 cases got worse with the skill loaded, and they are 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.
Other measured skills in the registry, with their headline benchmark lift.