Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Implement application state with Angular Signals, computed derivations, and NgRx Signal Store. Use when implementing reactive state with signal(), computed(), effect(), or @ngrx/signals in Angular.
.claude/skills/hoangnguyen0403-angular-state-management/SKILL.md| Model | Eval pass | Runs |
|---|---|---|
| gemini-3.6-flash | 100% | 6 |
| gemini-3.1-pro-preview | 100% | 1 |
| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | -13% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 1% | 0% |
| case-16 | ✗→✓ | ▲ Improved | -33% | 0% |
| case-18 | ✗→✓ | ▲ Improved | -16% | 0% |
| case-03 | ✓→✓ | = Same ✓ | 2% | 0% |
asReadonly().See signal store pattern for signal-based service and store examples.
computed() for totals, filtered lists, derived values — pure and cached.linkedSignal(() => source()) for dependent writable state that resets when source changes.untracked() to read signal inside computed()/effect() without creating dependency.@ngrx/signals (signalStore) with withState, withComputed, withMethods, and withEntities().effect() only for side effects (logging, localStorage sync, DOM manipulation)..set() or .update(v => ...).BehaviorSubject for state: Use Signals; keep RxJS only for complex event streams.When this skill applies, preserve the following domain terminology or equivalent concrete examples in the answer when relevant:
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 36,702 | 24,252 | -34% | 1 | 1 | 0% | 5,445 | 3,803 | -30% | 0 | 0 | — |
case-02 | fail→pass | 32,957 | 24,543 | -26% | 1 | 1 | 0% | 4,304 | 3,753 | -13% | 0 | 0 | — |
case-03 | pass→pass | 16,941 | 15,642 | -8% | 1 | 1 | 0% | 1,911 | 1,950 | +2% | 0 | 0 | — |
case-04 | pass→pass | 18,039 | 11,588 | -36% | 1 | 1 | 0% | 2,923 | 2,289 | -22% | 0 | 0 | — |
case-05 | fail→pass | 11,489 | 12,186 | +6% | 1 | 1 | 0% | 1,542 | 1,550 | +1% | 0 | 0 | — |
case-06 | pass→pass | 13,459 | 12,588 | -6% | 1 | 1 | 0% | 1,327 | 1,438 | +8% | 0 | 0 | — |
case-07 | pass→pass | 14,989 | 14,395 | -4% | 1 | 1 | 0% | 2,365 | 2,054 | -13% | 0 | 0 | — |
case-08 | pass→pass | 29,199 | 24,018 | -18% | 1 | 1 | 0% | 4,560 | 4,084 | -10% | 0 | 0 | — |
case-09 | pass→pass | 23,220 | 21,819 | -6% | 1 | 1 | 0% | 3,500 | 3,437 | -2% | 0 | 0 | — |
case-10 | pass→pass | 19,139 | 20,407 | +7% | 1 | 1 | 0% | 2,421 | 2,888 | +19% | 0 | 0 | — |
case-11 | pass→pass | 10,856 | 9,772 | -10% | 1 | 1 | 0% | 869 | 1,080 | +24% | 0 | 0 | — |
case-12 | pass→pass | 9,017 | 9,758 | +8% | 1 | 1 | 0% | 1,673 | 2,213 | +32% | 0 | 0 | — |
case-13 | pass→pass | 16,396 | 11,597 | -29% | 1 | 1 | 0% | 1,919 | 2,343 | +22% | 0 | 0 | — |
case-14 | pass→pass | 15,795 | 15,377 | -3% | 1 | 1 | 0% | 1,886 | 2,186 | +16% | 0 | 0 | — |
case-15 | pass→pass | 14,238 | 11,666 | -18% | 1 | 1 | 0% | 1,429 | 1,422 | -0% | 0 | 0 | — |
case-16 | fail→pass | 10,303 | 9,424 | -9% | 1 | 1 | 0% | 1,475 | 984 | -33% | 0 | 0 | — |
case-17 | pass→pass | 9,083 | 13,478 | +48% | 1 | 1 | 0% | 1,604 | 1,806 | +13% | 0 | 0 | — |
case-18 | fail→pass | 17,724 | 12,389 | -30% | 1 | 1 | 0% | 1,906 | 1,609 | -16% | 0 | 0 | — |
case-19 | pass→pass | 21,632 | 13,959 | -35% | 1 | 1 | 0% | 2,475 | 2,748 | +11% | 0 | 0 | — |
case-20 | pass→pass | 12,629 | 11,150 | -12% | 1 | 1 | 0% | 1,405 | 1,243 | -12% | 0 | 0 | — |
case-21 | pass→pass | 9,622 | 12,184 | +27% | 1 | 1 | 0% | 632 | 1,314 | +108% | 0 | 0 | — |
case-22 | pass→pass | 5,928 | 12,033 | +103% | 1 | 1 | 0% | 902 | 1,324 | +47% | 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 +18 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.