Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Queries the OptimalOS knowledge base using FTS5 full-text search. Returns L0 abstracts (~100 tokens each) ranked by relevance. Use before answering questions about past context -- search first, don't guess which file to read.
.claude/skills/miosa-osa-search/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 20% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 54% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 100% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 88% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -27% | 0% |
> Search the operation's knowledge base for existing information.
/search <query> [--node <node>] [--type <signal|document|note>] [--limit <n>]Full-text search across all reference files, domain knowledge, and stored signals. Returns results ranked by relevance with L0 abstracts. Use before creating new content to avoid duplication.
| Param | Default | Description | |-------|---------|-------------| | --node | all | Restrict search to a specific node | | --type | all | Filter by resource type | | --limit | 5 | Maximum results to return |
bash# Broad search /search "pricing strategy" # Scoped search /search "client feedback" --type signal --limit 10 # Search specific node /search "architecture decision" --node product
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 6,754 | 5,652 | -16% | 1 | 1 | 0% | 998 | 1,193 | +20% | 0 | 0 | — |
case-02 | fail→pass | 3,265 | 3,389 | +4% | 1 | 1 | 0% | 426 | 656 | +54% | 0 | 0 | — |
case-03 | fail→pass | 6,320 | 6,047 | -4% | 1 | 1 | 0% | 635 | 1,272 | +100% | 0 | 0 | — |
case-04 | fail→pass | 14,361 | 2,176 | -85% | 1 | 1 | 0% | 292 | 549 | +88% | 0 | 0 | — |
case-05 | fail→pass | 11,083 | 1,889 | -83% | 1 | 1 | 0% | 734 | 539 | -27% | 0 | 0 | — |
case-06 | fail→pass | 9,416 | 2,652 | -72% | 1 | 1 | 0% | 1,639 | 535 | -67% | 0 | 0 | — |
case-07 | fail→pass | 4,007 | 2,992 | -25% | 1 | 1 | 0% | 588 | 476 | -19% | 0 | 0 | — |
case-08 | fail→pass | 15,388 | 6,192 | -60% | 1 | 1 | 0% | 1,036 | 1,294 | +25% | 0 | 0 | — |
case-09 | fail→pass | 5,068 | 1,930 | -62% | 1 | 1 | 0% | 719 | 455 | -37% | 0 | 0 | — |
case-10 | fail→pass | 8,533 | 1,843 | -78% | 1 | 1 | 0% | 210 | 518 | +147% | 0 | 0 | — |
case-11 | fail→pass | 26,084 | 2,016 | -92% | 1 | 1 | 0% | 3,576 | 532 | -85% | 0 | 0 | — |
case-12 | fail→fail | 9,849 | 4,868 | -51% | 1 | 1 | 0% | 1,555 | 948 | -39% | 0 | 0 | — |
case-13 | fail→pass | 12,107 | 4,615 | -62% | 1 | 1 | 0% | 2,117 | 970 | -54% | 0 | 0 | — |
case-14 | pass→pass | 12,592 | 5,095 | -60% | 1 | 1 | 0% | 2,380 | 1,014 | -57% | 0 | 0 | — |
case-15 | fail→pass | 28,128 | 4,286 | -85% | 1 | 1 | 0% | 737 | 856 | +16% | 0 | 0 | — |
case-16 | fail→pass | 6,342 | 6,545 | +3% | 1 | 1 | 0% | 921 | 1,163 | +26% | 0 | 0 | — |
case-17 | fail→pass | 9,937 | 4,530 | -54% | 1 | 1 | 0% | 1,584 | 916 | -42% | 0 | 0 | — |
case-18 | pass→pass | 11,480 | 3,853 | -66% | 1 | 1 | 0% | 1,616 | 834 | -48% | 0 | 0 | — |
case-19 | fail→pass | 9,316 | 3,484 | -63% | 1 | 1 | 0% | 438 | 906 | +107% | 0 | 0 | — |
case-20 | fail→fail | 5,455 | 15,987 | +193% | 1 | 1 | 0% | 325 | 2,656 | +717% | 0 | 0 | — |
case-21 | pass→fail | 2,436 | 12,604 | +417% | 1 | 1 | 0% | 208 | 2,110 | +914% | 0 | 0 | — |
case-22 | pass→fail | 7,298 | 24,309 | +233% | 1 | 1 | 0% | 1,064 | 4,378 | +311% | 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 17 counted toward the lift figure. The other 5 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 +64 percentage points is the difference between those two pass rates over the 17 comparable cases. 2 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.