Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Format critical path and prerequisites into bite-sized executable plan following superpowers:writing-plans conventions
.claude/skills/yogsoth-ai-plan-writing/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -68% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 3% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 5% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -48% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -1% | 0% |
Key Question: How to write it as an executable plan?
实现 spec 与 plan 的产出直接交给 superpowers 现成链路，不再自造 bite-sized 规则。
排程层（critical-path-planning / prerequisite-planning / task-decomposition）产出 任务 DAG、关键路径、工期后，本策略：
Skill load superpowers:brainstorming —— 把 ③ experiment-design 的科学实验设计当输入，完整跑一趟（clarify → approaches → 实现 spec → self-review → 用户 gate）， 产出实现 spec。
TDD 结构的可执行 plan。
排程层的任务 DAG 作为 brainstorming/writing-plans 的输入上下文传入。
[排程层: 关键路径 + IO 序列]
→ Skill load superpowers:brainstorming (科学设计 → 实现 spec, 自带用户 gate)
→ 终态自动 → superpowers:writing-plans (出 bite-sized plan)| Step | Max Budget | Output | |------|-----------|--------| | Plan formatting | 5% | Complete task plan | | Validation pass | 2% | Confirmed no placeholders |
A valid plan has:
<!-- BEGIN available-tables (generated) --> <!-- external rows hand-maintained; do not regenerate this file -->
Optional, no fixed order; the final leaf is always a sop.
| Tactic | When to use | | --- | --- | | task-decomposition | Orchestrate the breakdown of experiment design into sequenced, estimated, and formatted task plan |
Optional, no fixed order; the final leaf is always a sop.
| SOP | When to use | | --- | --- | | plan-formatting | Format task plan as bite-sized executable tasks following superpowers:writing-plans conventions | | superpowers:brainstorming | Turn the experiment design into an implementation spec (clarify -> approaches -> spec -> self-review -> user gate) | | superpowers:writing-plans | Produce a bite-sized, TDD-structured implementation plan from the spec |
<!-- 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-01 | fail→pass | 29,296 | 33,562 | +15% | 1 | 1 | 0% | 4,396 | 1,413 | -68% | 0 | 0 | — |
case-02 | fail→pass | 59,768 | 69,586 | +16% | 1 | 1 | 0% | 8,236 | 8,487 | +3% | 0 | 0 | — |
case-03 | fail→fail | 32,845 | 52,092 | +59% | 1 | 1 | 0% | 5,545 | 7,532 | +36% | 0 | 0 | — |
case-04 | pass→pass | 10,980 | 10,083 | -8% | 1 | 1 | 0% | 1,201 | 1,498 | +25% | 0 | 0 | — |
case-05 | pass→pass | 4,720 | 8,797 | +86% | 1 | 1 | 0% | 789 | 1,265 | +60% | 0 | 0 | — |
case-06 | pass→pass | 21,027 | 38,589 | +84% | 1 | 1 | 0% | 3,161 | 7,514 | +138% | 0 | 0 | — |
case-07 | fail→pass | 29,193 | 6,501 | -78% | 1 | 1 | 0% | 808 | 845 | +5% | 0 | 0 | — |
case-08 | fail→pass | 16,331 | 6,810 | -58% | 1 | 1 | 0% | 1,564 | 808 | -48% | 0 | 0 | — |
case-09 | fail→pass | 15,343 | 20,814 | +36% | 1 | 1 | 0% | 1,604 | 1,589 | -1% | 0 | 0 | — |
case-10 | fail→pass | 16,257 | 9,124 | -44% | 1 | 1 | 0% | 1,778 | 1,111 | -38% | 0 | 0 | — |
case-11 | fail→pass | 20,120 | 9,928 | -51% | 1 | 1 | 0% | 2,387 | 1,375 | -42% | 0 | 0 | — |
case-12 | fail→pass | 13,115 | 10,698 | -18% | 1 | 1 | 0% | 1,739 | 1,344 | -23% | 0 | 0 | — |
case-13 | pass→pass | 19,386 | 15,422 | -20% | 1 | 1 | 0% | 2,482 | 2,259 | -9% | 0 | 0 | — |
case-14 | fail→pass | 18,741 | 12,448 | -34% | 1 | 1 | 0% | 2,058 | 1,862 | -10% | 0 | 0 | — |
case-15 | fail→pass | 20,762 | 14,597 | -30% | 1 | 1 | 0% | 2,198 | 2,099 | -5% | 0 | 0 | — |
case-16 | pass→pass | 20,115 | 7,583 | -62% | 1 | 1 | 0% | 2,557 | 1,868 | -27% | 0 | 0 | — |
case-17 | pass→pass | 12,387 | 17,220 | +39% | 1 | 1 | 0% | 1,983 | 2,434 | +23% | 0 | 0 | — |
case-18 | pass→pass | 15,161 | 7,009 | -54% | 1 | 1 | 0% | 1,618 | 841 | -48% | 0 | 0 | — |
case-19 | pass→pass | 13,034 | 7,273 | -44% | 1 | 1 | 0% | 1,274 | 989 | -22% | 0 | 0 | — |
case-20 | pass→pass | 18,844 | 23,443 | +24% | 1 | 1 | 0% | 2,190 | 2,916 | +33% | 0 | 0 | — |
case-21 | fail→pass | 15,335 | 12,022 | -22% | 1 | 1 | 0% | 1,715 | 1,472 | -14% | 0 | 0 | — |
case-22 | pass→pass | 12,325 | 9,758 | -21% | 1 | 1 | 0% | 1,526 | 1,514 | -1% | 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 +50 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.