Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Analyzes completed work to extract reusable patterns: code patterns, debugging approaches, architectural decisions, process improvements. Stores them in the pattern library indexed by domain and task type, with usage count and success rate tracking.
.claude/skills/miosa-osa-pattern-capture/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-05 | ✗→✓ | ▲ Improved | 1% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -50% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 444% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -6% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -1% | 0% |
> Extract and store reusable patterns from completed tasks.
/pattern-capture [--domain <domain>] [--from-session] [--list]Analyzes completed work to extract reusable patterns: code patterns, debugging approaches, architectural decisions, process improvements. Stores them in the pattern library indexed by domain and task type, with usage count and success rate tracking.
Pattern format:
json{ "id": "pattern-id", "domain": "backend", "type": "code_pattern", "title": "Pattern name", "description": "When and how to use", "confidence": 0.85, "usage_count": 5 }
bash# Capture patterns from current session /pattern-capture --from-session # Capture patterns for a specific domain /pattern-capture --domain backend # List all stored patterns /pattern-capture --list
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-05 | fail→pass | 9,119 | 5,974 | -34% | 1 | 1 | 0% | 1,401 | 1,414 | +1% | 0 | 0 | — |
case-01 | fail→fail | 18,802 | 10,964 | -42% | 1 | 1 | 0% | 2,886 | 2,237 | -22% | 0 | 0 | — |
case-02 | fail→fail | 8,929 | 4,119 | -54% | 1 | 1 | 0% | 1,476 | 615 | -58% | 0 | 0 | — |
case-03 | fail→fail | 6,268 | 9,747 | +56% | 1 | 1 | 0% | 952 | 1,942 | +104% | 0 | 0 | — |
case-04 | fail→pass | 31,356 | 12,019 | -62% | 1 | 1 | 0% | 5,439 | 2,717 | -50% | 0 | 0 | — |
case-06 | fail→pass | 4,437 | 5,418 | +22% | 1 | 1 | 0% | 237 | 1,290 | +444% | 0 | 0 | — |
case-07 | fail→pass | 5,925 | 2,769 | -53% | 1 | 1 | 0% | 859 | 809 | -6% | 0 | 0 | — |
case-08 | pass→pass | 10,397 | 3,589 | -65% | 1 | 1 | 0% | 1,247 | 1,019 | -18% | 0 | 0 | — |
case-09 | pass→pass | 7,504 | 3,417 | -54% | 1 | 1 | 0% | 1,214 | 932 | -23% | 0 | 0 | — |
case-10 | fail→pass | 16,598 | 13,563 | -18% | 1 | 1 | 0% | 2,702 | 2,667 | -1% | 0 | 0 | — |
case-11 | fail→pass | 18,043 | 1,747 | -90% | 1 | 1 | 0% | 2,827 | 581 | -79% | 0 | 0 | — |
case-12 | fail→pass | 21,565 | 6,665 | -69% | 1 | 1 | 0% | 1,820 | 519 | -71% | 0 | 0 | — |
case-13 | pass→pass | 25,303 | 2,383 | -91% | 1 | 1 | 0% | 4,727 | 613 | -87% | 0 | 0 | — |
case-14 | pass→pass | 8,175 | 2,723 | -67% | 1 | 1 | 0% | 1,197 | 723 | -40% | 0 | 0 | — |
case-15 | pass→pass | 14,137 | 12,831 | -9% | 1 | 1 | 0% | 2,293 | 2,528 | +10% | 0 | 0 | — |
case-16 | fail→pass | 13,967 | 2,172 | -84% | 1 | 1 | 0% | 2,334 | 687 | -71% | 0 | 0 | — |
case-17 | pass→pass | 11,582 | 3,289 | -72% | 1 | 1 | 0% | 1,644 | 848 | -48% | 0 | 0 | — |
case-18 | fail→fail | 12,112 | 5,990 | -51% | 1 | 1 | 0% | 1,764 | 1,275 | -28% | 0 | 0 | — |
case-19 | fail→pass | 11,813 | 6,401 | -46% | 1 | 1 | 0% | 2,272 | 1,374 | -40% | 0 | 0 | — |
case-20 | fail→pass | 10,490 | 1,786 | -83% | 1 | 1 | 0% | 1,667 | 552 | -67% | 0 | 0 | — |
case-21 | fail→pass | 8,679 | 3,212 | -63% | 1 | 1 | 0% | 1,317 | 935 | -29% | 0 | 0 | — |
case-22 | pass→pass | 9,972 | 9,534 | -4% | 1 | 1 | 0% | 1,609 | 1,738 | +8% | 0 | 0 | — |
case-23 | pass→pass | 14,606 | 17,790 | +22% | 1 | 1 | 0% | 2,556 | 4,120 | +61% | 0 | 0 | — |
case-24 | pass→pass | 8,084 | 5,917 | -27% | 1 | 1 | 0% | 1,450 | 1,232 | -15% | 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 22 counted toward the lift figure. The other 2 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 +46 percentage points is the difference between those two pass rates over the 22 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.