Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when workflow components are inconsistent, naming conventions vary, or a new team member's work needs alignment to project standards.
.claude/skills/sharpdeveye-calibrate/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-07 | ✗→✓ | ▲ Improved | 3% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 5% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -10% | 0% |
| case-12 | ✗→✓ | ▲ Improved | -3% | 0% |
| case-15 | ✗→✓ | ▲ Improved | -24% | 0% |
Invoke /agent-workflow — it contains workflow principles, anti-patterns, and the Context Gathering Protocol. Follow the protocol before proceeding — if no workflow context exists yet, you MUST run /teach-maestro first. Consult the prompt-engineering reference in the agent-workflow skill for naming and style consistency patterns.
Ensure consistency across all workflow components. Inconsistency creates confusion — for the model, for developers, and for users.
Naming Conventions
Prompt Style
Error Handling
Logging
.maestro.md with the established conventions.| Dimension | Standard | Deviations Found | Priority | |-----------|----------|------------------|----------| | Tool naming | ? | ? of ? tools | High/Med/Low | | Prompt structure | ? | ? of ? prompts | High/Med/Low | | Error format | ? | ? of ? tools | High/Med/Low | | Log format | ? | ? of ? entries | High/Med/Low |
.maestro.md with established conventionsAfter calibration, run /refine for a final polish pass, or /evaluate to verify consistency improvements.
NEVER:
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 28,113 | 2,405 | -91% | 1 | 1 | 0% | 3,361 | 913 | -73% | 0 | 0 | — |
case-02 | fail→fail | 5,475 | 4,877 | -11% | 1 | 1 | 0% | 233 | 914 | +292% | 0 | 0 | — |
case-03 | fail→fail | 10,327 | 3,763 | -64% | 1 | 1 | 0% | 1,717 | 856 | -50% | 0 | 0 | — |
case-04 | pass→fail | 12,001 | 4,952 | -59% | 1 | 1 | 0% | 2,181 | 901 | -59% | 0 | 0 | — |
case-05 | pass→fail | 7,248 | 14,733 | +103% | 1 | 1 | 0% | 1,186 | 2,687 | +127% | 0 | 0 | — |
case-06 | pass→fail | 27,860 | 31,292 | +12% | 1 | 1 | 0% | 5,597 | 6,806 | +22% | 0 | 0 | — |
case-07 | fail→pass | 29,124 | 11,518 | -60% | 1 | 1 | 0% | 2,343 | 2,424 | +3% | 0 | 0 | — |
case-08 | fail→pass | 9,911 | 7,760 | -22% | 1 | 1 | 0% | 1,450 | 1,522 | +5% | 0 | 0 | — |
case-09 | fail→pass | 11,264 | 5,025 | -55% | 1 | 1 | 0% | 1,535 | 1,380 | -10% | 0 | 0 | — |
case-10 | pass→pass | 9,922 | 9,744 | -2% | 1 | 1 | 0% | 1,550 | 1,732 | +12% | 0 | 0 | — |
case-11 | pass→pass | 11,010 | 10,693 | -3% | 1 | 1 | 0% | 1,633 | 2,297 | +41% | 0 | 0 | — |
case-12 | fail→pass | 14,487 | 20,873 | +44% | 1 | 1 | 0% | 2,371 | 2,296 | -3% | 0 | 0 | — |
case-13 | pass→pass | 13,221 | 9,829 | -26% | 1 | 1 | 0% | 2,184 | 2,304 | +5% | 0 | 0 | — |
case-14 | pass→pass | 15,938 | 17,064 | +7% | 1 | 1 | 0% | 2,527 | 3,249 | +29% | 0 | 0 | — |
case-15 | fail→pass | 13,156 | 5,267 | -60% | 1 | 1 | 0% | 1,947 | 1,473 | -24% | 0 | 0 | — |
case-16 | fail→pass | 7,005 | 4,110 | -41% | 1 | 1 | 0% | 1,173 | 1,262 | +8% | 0 | 0 | — |
case-17 | fail→pass | 9,890 | 2,761 | -72% | 1 | 1 | 0% | 1,623 | 982 | -39% | 0 | 0 | — |
case-18 | fail→pass | 10,647 | 2,279 | -79% | 1 | 1 | 0% | 1,467 | 962 | -34% | 0 | 0 | — |
case-19 | pass→pass | 30,183 | 8,297 | -73% | 1 | 1 | 0% | 2,411 | 1,940 | -20% | 0 | 0 | — |
case-20 | pass→pass | 10,260 | 6,389 | -38% | 1 | 1 | 0% | 1,473 | 1,597 | +8% | 0 | 0 | — |
case-21 | fail→pass | 13,308 | 9,602 | -28% | 1 | 1 | 0% | 1,942 | 2,140 | +10% | 0 | 0 | — |
case-22 | pass→pass | 11,789 | 9,142 | -22% | 1 | 1 | 0% | 1,899 | 1,966 | +4% | 0 | 0 | — |
case-23 | fail→fail | 17,101 | 15,967 | -7% | 1 | 1 | 0% | 2,804 | 3,287 | +17% | 0 | 0 | — |
case-24 | fail→fail | 16,452 | 15,276 | -7% | 1 | 1 | 0% | 2,442 | 3,003 | +23% | 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. 24 cases were attempted, and 21 counted toward the lift figure. The other 3 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 +25 percentage points is the difference between those two pass rates over the 21 comparable cases. 3 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.