Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Orchestrate the breakdown of experiment design into sequenced, estimated, and formatted task plan
.claude/skills/yogsoth-ai-task-decomposition/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 76% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -5% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -13% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -19% | 0% |
| case-13 | ✗→✓ | ▲ Improved | -7% | 0% |
pseudocodeFUNCTION task_decomposition(experiment_design): // Phase 1: Enumerate activities = SPAWN activity-listing(experiment_design) VALIDATE activities.count > 0 // Phase 2: Sequence sequenced = SPAWN dependency-sequencing(activities) VALIDATE no_circular_dependencies(sequenced) // Phase 3: Estimate estimated = SPAWN duration-estimation(sequenced) VALIDATE all_tasks_have_estimates(estimated) // Phase 4: Format plan = SPAWN plan-formatting(estimated) VALIDATE no_placeholders(plan) VALIDATE all_paths_absolute(plan) RETURN plan END
| Condition | Action | |-----------|--------| | Activity list is empty | ABORT — experiment design is incomplete | | Circular dependency detected | Return to dependency-sequencing with conflict details | | Duration estimate has high variance (σ > tM) | Flag as risky, add extra buffer | | Plan contains TBD/TODO | REJECT — return to plan-formatting | | Task count > 50 | Consider grouping into phases | | Task count < 3 | Verify decomposition isn't too coarse |
<!-- BEGIN available-tables (generated) -->
Optional, no fixed order; the final leaf is always a sop.
| SOP | When to use | | --- | --- | | activity-listing | Enumerate all implementation activities from an experiment design | | dependency-sequencing | Determine task dependencies and execution order | | duration-estimation | Three-point PERT estimation for implementation activities | | plan-formatting | Format task plan as bite-sized executable tasks following superpowers:writing-plans conventions |
<!-- END available-tables (generated) -->
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-08 | pass→pass | 27,475 | 29,087 | +6% | 1 | 1 | 0% | 3,660 | 6,047 | +65% | 0 | 0 | — |
case-01 | fail→pass | 15,179 | 28,984 | +91% | 1 | 1 | 0% | 2,776 | 4,877 | +76% | 0 | 0 | — |
case-02 | pass→fail | 11,187 | 32,389 | +190% | 1 | 1 | 0% | 959 | 6,925 | +622% | 0 | 0 | — |
case-07 | pass→fail | 22,242 | 26,872 | +21% | 1 | 1 | 0% | 2,816 | 4,869 | +73% | 0 | 0 | — |
case-03 | fail→pass | 46,529 | 45,187 | -3% | 1 | 1 | 0% | 7,499 | 7,148 | -5% | 0 | 0 | — |
case-04 | pass→pass | 22,099 | 24,191 | +9% | 1 | 1 | 0% | 2,809 | 3,661 | +30% | 0 | 0 | — |
case-05 | pass→pass | 16,348 | 20,016 | +22% | 1 | 1 | 0% | 2,037 | 3,112 | +53% | 0 | 0 | — |
case-06 | pass→pass | 19,300 | 25,842 | +34% | 1 | 1 | 0% | 3,113 | 4,977 | +60% | 0 | 0 | — |
case-09 | fail→pass | 20,853 | 14,640 | -30% | 1 | 1 | 0% | 2,542 | 2,214 | -13% | 0 | 0 | — |
case-10 | fail→pass | 15,485 | 10,533 | -32% | 1 | 1 | 0% | 1,560 | 1,263 | -19% | 0 | 0 | — |
case-11 | pass→pass | 24,112 | 17,546 | -27% | 1 | 1 | 0% | 2,923 | 3,885 | +33% | 0 | 0 | — |
case-12 | pass→pass | 15,613 | 8,547 | -45% | 1 | 1 | 0% | 2,377 | 1,933 | -19% | 0 | 0 | — |
case-13 | fail→pass | 31,372 | 7,915 | -75% | 1 | 1 | 0% | 1,035 | 963 | -7% | 0 | 0 | — |
case-14 | fail→pass | 18,324 | 53,898 | +194% | 1 | 1 | 0% | 2,914 | 8,723 | +199% | 0 | 0 | — |
case-15 | pass→pass | 10,019 | 8,594 | -14% | 1 | 1 | 0% | 937 | 1,903 | +103% | 0 | 0 | — |
case-16 | pass→pass | 8,296 | 15,843 | +91% | 1 | 1 | 0% | 489 | 3,100 | +534% | 0 | 0 | — |
case-17 | pass→pass | 9,110 | 3,283 | -64% | 1 | 1 | 0% | 585 | 1,029 | +76% | 0 | 0 | — |
case-18 | pass→pass | 15,352 | 9,147 | -40% | 1 | 1 | 0% | 1,458 | 1,137 | -22% | 0 | 0 | — |
case-19 | fail→pass | 19,327 | 21,895 | +13% | 1 | 1 | 0% | 2,451 | 3,793 | +55% | 0 | 0 | — |
case-20 | fail→pass | 14,525 | 12,884 | -11% | 1 | 1 | 0% | 2,228 | 2,175 | -2% | 0 | 0 | — |
case-21 | fail→pass | 17,037 | 2,604 | -85% | 1 | 1 | 0% | 2,438 | 978 | -60% | 0 | 0 | — |
case-22 | pass→pass | 21,157 | 7,302 | -65% | 1 | 1 | 0% | 2,626 | 880 | -66% | 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 +32 percentage points is the difference between those two pass rates over the 21 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.