Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Scans all context files mentioning a topic, checks when they were last updated, compares against recent signals, and identifies stale or contradictory information. Suggests specific updates to bring contexts current.
.claude/skills/miosa-osa-reweave/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -25% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -54% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 52% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -82% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -75% | 0% |
Backward-pass analysis to find stale contexts and suggest updates.
Given a topic: searches related contexts → traverses entity graph → scores staleness → generates update suggestions. With Ollama: specific diff suggestions. Without: staleness flags with newer context titles.
/reweave "Ed Honour"
/reweave "AI Masters pricing" --max-results 20
/reweave "platform architecture" --staleness-days 14| Flag | Description | Default | |------|-------------|---------| | --max-results | Max contexts to analyze | 10 | | --staleness-days | Threshold for "stale" | 30 |
bashcd engine && mix optimal.reweave "topic" cd engine && mix optimal.reweave "Ed Honour" --max-results 20
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 8,418 | 4,147 | -51% | 1 | 1 | 0% | 1,060 | 791 | -25% | 0 | 0 | — |
case-02 | fail→pass | 10,860 | 3,950 | -64% | 1 | 1 | 0% | 1,585 | 734 | -54% | 0 | 0 | — |
case-03 | fail→pass | 6,323 | 7,432 | +18% | 1 | 1 | 0% | 854 | 1,299 | +52% | 0 | 0 | — |
case-04 | fail→pass | 33,355 | 1,671 | -95% | 1 | 1 | 0% | 2,515 | 453 | -82% | 0 | 0 | — |
case-05 | fail→pass | 16,255 | 2,251 | -86% | 1 | 1 | 0% | 2,474 | 622 | -75% | 0 | 0 | — |
case-06 | fail→pass | 11,427 | 2,216 | -81% | 1 | 1 | 0% | 1,504 | 496 | -67% | 0 | 0 | — |
case-07 | fail→pass | 6,904 | 1,543 | -78% | 1 | 1 | 0% | 1,047 | 424 | -60% | 0 | 0 | — |
case-08 | pass→pass | 18,845 | 1,821 | -90% | 1 | 1 | 0% | 2,685 | 472 | -82% | 0 | 0 | — |
case-09 | fail→pass | 16,326 | 2,054 | -87% | 1 | 1 | 0% | 3,249 | 474 | -85% | 0 | 0 | — |
case-10 | fail→pass | 10,408 | 2,547 | -76% | 1 | 1 | 0% | 1,634 | 568 | -65% | 0 | 0 | — |
case-11 | pass→pass | 15,626 | 3,035 | -81% | 1 | 1 | 0% | 2,554 | 689 | -73% | 0 | 0 | — |
case-12 | fail→pass | 17,943 | 3,738 | -79% | 1 | 1 | 0% | 1,538 | 737 | -52% | 0 | 0 | — |
case-13 | fail→pass | 14,654 | 8,052 | -45% | 1 | 1 | 0% | 2,281 | 960 | -58% | 0 | 0 | — |
case-14 | fail→pass | 18,701 | 2,965 | -84% | 1 | 1 | 0% | 1,914 | 710 | -63% | 0 | 0 | — |
case-15 | fail→pass | 6,040 | 2,828 | -53% | 1 | 1 | 0% | 1,007 | 688 | -32% | 0 | 0 | — |
case-16 | fail→pass | 7,611 | 2,256 | -70% | 1 | 1 | 0% | 1,057 | 417 | -61% | 0 | 0 | — |
case-17 | fail→pass | 8,890 | 2,247 | -75% | 1 | 1 | 0% | 1,278 | 476 | -63% | 0 | 0 | — |
case-18 | fail→pass | 16,809 | 17,235 | +3% | 1 | 1 | 0% | 2,525 | 2,819 | +12% | 0 | 0 | — |
case-19 | fail→pass | 21,001 | 2,144 | -90% | 1 | 1 | 0% | 2,944 | 449 | -85% | 0 | 0 | — |
case-20 | pass→pass | 13,184 | 7,964 | -40% | 1 | 1 | 0% | 2,039 | 1,547 | -24% | 0 | 0 | — |
case-21 | pass→pass | 8,463 | 2,941 | -65% | 1 | 1 | 0% | 1,253 | 718 | -43% | 0 | 0 | — |
case-22 | pass→pass | 19,826 | 37,927 | +91% | 1 | 1 | 0% | 3,946 | 8,456 | +114% | 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 +77 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.