Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Apply to any non-trivial work, not just bulk work: edits, migrations, analyses, checks. Build the tool that does it or proves it (codemod, script, generator, or a skill your subagents follow) instead of working by hand. The tool is the artifact a reviewer can rerun.
.claude/skills/kunanonj-cursor-plugin-pstack-principle-build-the-lever/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 4% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -25% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -22% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -39% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 11% | 0% |
When the work isn't trivial, build the tool that does it instead of doing it by hand.
Why: Two payoffs. Throughput: a codemod, generator, or script does the work the same way every time and reruns for free. Confidence: the tool is one artifact a reviewer can read and rerun to check the work. Hand-done changes can only be re-verified by redoing them. A deterministic script turns "trust me" into "run this".
Pattern: Default to building the lever. Skip it only when the task is genuinely trivial, a couple of obvious edits you can see at a glance.
Balance: The bar is triviality, not repetition. A one-off still earns a lever when the lever is what makes the work checkable. Per the Laziness Protocol, build the smallest script that does or proves the job, never a framework.
Distinct from Encode Lessons in Structure, which makes a recurring instruction a durable guardrail. This is throughput and reviewability on the work in front of you. For scripting the verification itself, see Prove It Works.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-09 | pass→pass | 13,295 | 8,139 | -39% | 1 | 1 | 0% | 2,247 | 1,893 | -16% | 0 | 0 | — |
case-01 | fail→pass | 10,221 | 7,485 | -27% | 1 | 1 | 0% | 1,977 | 2,057 | +4% | 0 | 0 | — |
case-02 | fail→pass | 11,738 | 6,234 | -47% | 1 | 1 | 0% | 2,351 | 1,766 | -25% | 0 | 0 | — |
case-03 | pass→pass | 11,802 | 9,466 | -20% | 1 | 1 | 0% | 2,538 | 2,523 | -1% | 0 | 0 | — |
case-04 | fail→pass | 16,389 | 7,297 | -55% | 1 | 1 | 0% | 2,544 | 1,977 | -22% | 0 | 0 | — |
case-05 | pass→pass | 5,811 | 2,938 | -49% | 1 | 1 | 0% | 958 | 1,143 | +19% | 0 | 0 | — |
case-06 | fail→pass | 11,682 | 4,479 | -62% | 1 | 1 | 0% | 2,166 | 1,324 | -39% | 0 | 0 | — |
case-07 | pass→pass | 10,347 | 7,337 | -29% | 1 | 1 | 0% | 1,929 | 1,909 | -1% | 0 | 0 | — |
case-08 | pass→pass | 9,360 | 5,228 | -44% | 1 | 1 | 0% | 1,684 | 1,436 | -15% | 0 | 0 | — |
case-10 | pass→pass | 6,082 | 2,633 | -57% | 1 | 1 | 0% | 1,211 | 1,004 | -17% | 0 | 0 | — |
case-11 | pass→pass | 8,768 | 5,346 | -39% | 1 | 1 | 0% | 1,451 | 1,513 | +4% | 0 | 0 | — |
case-12 | fail→pass | 8,516 | 6,727 | -21% | 1 | 1 | 0% | 1,511 | 1,671 | +11% | 0 | 0 | — |
case-13 | pass→pass | 10,239 | 5,817 | -43% | 1 | 1 | 0% | 1,776 | 1,498 | -16% | 0 | 0 | — |
case-14 | pass→fail | 10,528 | 4,490 | -57% | 1 | 1 | 0% | 1,910 | 1,336 | -30% | 0 | 0 | — |
case-15 | pass→pass | 10,080 | 8,588 | -15% | 1 | 1 | 0% | 1,753 | 2,054 | +17% | 0 | 0 | — |
case-16 | pass→pass | 5,105 | 4,676 | -8% | 1 | 1 | 0% | 925 | 1,411 | +53% | 0 | 0 | — |
case-17 | pass→pass | 14,027 | 5,494 | -61% | 1 | 1 | 0% | 1,621 | 1,471 | -9% | 0 | 0 | — |
case-18 | pass→pass | 6,180 | 3,187 | -48% | 1 | 1 | 0% | 1,109 | 1,061 | -4% | 0 | 0 | — |
case-19 | pass→pass | 10,228 | 6,735 | -34% | 1 | 1 | 0% | 1,863 | 1,724 | -7% | 0 | 0 | — |
case-20 | pass→pass | 11,528 | 7,137 | -38% | 1 | 1 | 0% | 2,086 | 1,825 | -13% | 0 | 0 | — |
case-21 | pass→pass | 9,851 | 4,883 | -50% | 1 | 1 | 0% | 1,650 | 1,546 | -6% | 0 | 0 | — |
case-22 | pass→pass | 12,720 | 9,779 | -23% | 1 | 1 | 0% | 2,470 | 2,207 | -11% | 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. 1 case got worse with the skill loaded, and it is 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.