Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Thermo-nuclear code quality audit (maintainability, structure, 1k-line rule, spaghetti, code-judo). Invoked via Task after a parent gathers diff and file contents. Loads the rubric from the `thermo-nuclear-code-quality-review` skill in the cursor-team-kit plugin.
.claude/skills/kunanonj-cursor-plugin-teamkit-agent-thermo-nuclear-code-quality-review/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-17 | ✗→✓ | ▲ Improved | -66% | 0% |
| case-13 | ✗→✓ | ▲ Improved | 2% | 0% |
| case-14 | ✗→✓ | ▲ Improved | -56% | 0% |
| case-15 | ✗→✓ | ▲ Improved | -49% | 0% |
| case-20 | ✗→✓ | ▲ Improved | 655% | 0% |
You are a Task subagent. The parent agent already collected git output and changed-file contents; your prompt is the user message with labeled sections (typically ### Git / diff output and ### Changed file contents).
thermo-nuclear-code-quality-review skill (shipped in the cursor-team-kit plugin) and treat its SKILL.md as the complete rubric — tone, approval bar, output ordering, code-judo / 1k-line / spaghetti rules.Typical flow: in one message, run two Task calls in parallel — subagent_type: "shell" and subagent_type: "explore" — to collect git diff <base>...HEAD output and full contents of changed files (default base main). Then invoke this agent with subagent_type: "thermo-nuclear-code-quality-review" and a user prompt containing ### Git / diff output and ### Changed file contents.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-17 | fail→pass | 10,831 | 1,947 | -82% | 1 | 1 | 0% | 1,954 | 663 | -66% | 0 | 0 | — |
case-01 | fail→fail | 3,809 | 2,218 | -42% | 1 | 1 | 0% | 630 | 764 | +21% | 0 | 0 | — |
case-02 | pass→pass | 11,193 | 5,577 | -50% | 1 | 1 | 0% | 1,955 | 1,172 | -40% | 0 | 0 | — |
case-03 | pass→pass | 9,840 | 2,684 | -73% | 1 | 1 | 0% | 1,699 | 804 | -53% | 0 | 0 | — |
case-04 | pass→pass | 10,264 | 5,375 | -48% | 1 | 1 | 0% | 1,815 | 1,261 | -31% | 0 | 0 | — |
case-05 | pass→pass | 9,019 | 3,697 | -59% | 1 | 1 | 0% | 1,499 | 924 | -38% | 0 | 0 | — |
case-06 | pass→pass | 10,765 | 6,576 | -39% | 1 | 1 | 0% | 1,956 | 1,491 | -24% | 0 | 0 | — |
case-07 | pass→pass | 9,715 | 3,582 | -63% | 1 | 1 | 0% | 1,514 | 938 | -38% | 0 | 0 | — |
case-08 | pass→pass | 12,271 | 9,102 | -26% | 1 | 1 | 0% | 2,061 | 1,786 | -13% | 0 | 0 | — |
case-09 | pass→pass | 10,788 | 8,680 | -20% | 1 | 1 | 0% | 1,906 | 1,712 | -10% | 0 | 0 | — |
case-10 | pass→pass | 5,819 | 2,908 | -50% | 1 | 1 | 0% | 1,033 | 757 | -27% | 0 | 0 | — |
case-11 | pass→pass | 5,925 | 4,560 | -23% | 1 | 1 | 0% | 1,104 | 1,133 | +3% | 0 | 0 | — |
case-12 | pass→pass | 10,043 | 5,297 | -47% | 1 | 1 | 0% | 1,707 | 1,220 | -29% | 0 | 0 | — |
case-13 | fail→pass | 7,038 | 5,067 | -28% | 1 | 1 | 0% | 1,124 | 1,147 | +2% | 0 | 0 | — |
case-14 | fail→pass | 7,824 | 1,341 | -83% | 1 | 1 | 0% | 1,351 | 598 | -56% | 0 | 0 | — |
case-15 | fail→pass | 14,609 | 5,331 | -64% | 1 | 1 | 0% | 2,704 | 1,385 | -49% | 0 | 0 | — |
case-16 | pass→pass | 8,658 | 1,864 | -78% | 1 | 1 | 0% | 1,586 | 659 | -58% | 0 | 0 | — |
case-18 | pass→pass | 11,278 | 8,356 | -26% | 1 | 1 | 0% | 1,760 | 1,620 | -8% | 0 | 0 | — |
case-19 | pass→pass | 9,327 | 6,315 | -32% | 1 | 1 | 0% | 1,489 | 1,249 | -16% | 0 | 0 | — |
case-20 | fail→pass | 1,145 | 7,445 | +550% | 1 | 1 | 0% | 157 | 1,185 | +655% | 0 | 0 | — |
case-21 | fail→fail | 5,942 | 6,402 | +8% | 1 | 1 | 0% | 1,025 | 854 | -17% | 0 | 0 | — |
case-22 | fail→fail | 18,158 | 19,868 | +9% | 1 | 1 | 0% | 4,216 | 4,595 | +9% | 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 +23 percentage points is the difference between those two pass rates over the 21 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.