Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when you have a pre-registered analysis plan to execute inline in this session with review checkpoints, on a platform without subagents
.claude/skills/k-dense-ai-executing-analysis/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -52% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 64% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 2% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -18% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 30% | 0% |
Load the pre-registered plan, review it critically, execute all steps exactly as registered, report when complete.
Announce at start: "I'm using the executing-analysis skill to run this analysis."
Note: Tell your human partner that Science Superpowers works much better with access to subagents — the protocol-compliance and rigor reviews catch more when run by fresh agents. If subagents are available, use science-superpowers:subagent-driven-analysis instead.
The plan MUST be pre-registered and frozen (science-superpowers:preregistering-analysis) and the workspace set up (science-superpowers:setting-up-reproducible-analysis). If either is missing, stop and do it first. Executing before freezing turns the analysis exploratory.
For each step:
science-superpowers:verifying-results-before-claiming before marking done.After each natural group of steps (e.g., data prep, then primary model), pause and report results to your human partner before continuing. Keep confirmatory and exploratory results clearly separated in every report.
After all steps complete and verified:
science-superpowers:requesting-red-team-review on the whole result.science-superpowers:reporting-and-archiving-findings.STOP immediately when:
science-superpowers:investigating-anomalous-results (do NOT quietly drop data or tweak parameters)Ask rather than guess. A guess that changes the analysis silently destroys the result's credibility.
Required workflow skills:
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 12,771 | 3,834 | -70% | 1 | 1 | 0% | 2,272 | 1,096 | -52% | 0 | 0 | — |
case-02 | fail→pass | 3,971 | 2,036 | -49% | 1 | 1 | 0% | 678 | 1,113 | +64% | 0 | 0 | — |
case-03 | fail→pass | 6,815 | 2,329 | -66% | 1 | 1 | 0% | 1,051 | 1,068 | +2% | 0 | 0 | — |
case-04 | fail→pass | 10,750 | 4,339 | -60% | 1 | 1 | 0% | 1,971 | 1,625 | -18% | 0 | 0 | — |
case-05 | pass→pass | 9,535 | 4,199 | -56% | 1 | 1 | 0% | 1,650 | 1,560 | -5% | 0 | 0 | — |
case-06 | pass→pass | 13,005 | 6,091 | -53% | 1 | 1 | 0% | 2,039 | 1,781 | -13% | 0 | 0 | — |
case-07 | pass→pass | 10,600 | 10,350 | -2% | 1 | 1 | 0% | 1,850 | 2,514 | +36% | 0 | 0 | — |
case-08 | fail→fail | 12,746 | 6,863 | -46% | 1 | 1 | 0% | 2,002 | 1,878 | -6% | 0 | 0 | — |
case-09 | fail→pass | 8,981 | 6,529 | -27% | 1 | 1 | 0% | 1,488 | 1,936 | +30% | 0 | 0 | — |
case-10 | fail→fail | 8,049 | 8,038 | -0% | 1 | 1 | 0% | 1,202 | 2,103 | +75% | 0 | 0 | — |
case-11 | fail→pass | 10,449 | 2,142 | -80% | 1 | 1 | 0% | 1,446 | 1,115 | -23% | 0 | 0 | — |
case-12 | pass→pass | 11,398 | 12,497 | +10% | 1 | 1 | 0% | 1,847 | 2,812 | +52% | 0 | 0 | — |
case-13 | pass→pass | 8,992 | 4,281 | -52% | 1 | 1 | 0% | 1,354 | 1,438 | +6% | 0 | 0 | — |
case-14 | pass→pass | 9,884 | 6,750 | -32% | 1 | 1 | 0% | 1,515 | 1,777 | +17% | 0 | 0 | — |
case-15 | fail→pass | 17,784 | 9,621 | -46% | 1 | 1 | 0% | 1,783 | 2,185 | +23% | 0 | 0 | — |
case-16 | fail→pass | 7,854 | 2,204 | -72% | 1 | 1 | 0% | 1,164 | 1,132 | -3% | 0 | 0 | — |
case-17 | pass→pass | 10,759 | 5,577 | -48% | 1 | 1 | 0% | 1,587 | 1,639 | +3% | 0 | 0 | — |
case-18 | pass→pass | 11,034 | 7,281 | -34% | 1 | 1 | 0% | 1,729 | 1,879 | +9% | 0 | 0 | — |
case-19 | pass→pass | 17,651 | 18,963 | +7% | 1 | 1 | 0% | 2,774 | 3,551 | +28% | 0 | 0 | — |
case-20 | pass→pass | 6,591 | 9,521 | +44% | 1 | 1 | 0% | 892 | 2,166 | +143% | 0 | 0 | — |
case-21 | pass→fail | 17,317 | 18,293 | +6% | 1 | 1 | 0% | 2,509 | 3,854 | +54% | 0 | 0 | — |
case-22 | fail→pass | 11,085 | 9,376 | -15% | 1 | 1 | 0% | 1,606 | 2,162 | +35% | 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 +36 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.