Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Produces content using the Signal Theory framework: resolves all 5 dimensions (Mode, Genre, Type, Format, Structure), applies the correct genre skeleton, and matches the receiver's decoding capacity. Looks up the person's preferred genre from the people registry.
.claude/skills/miosa-osa-write/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-10 | ✗→✓ | ▲ Improved | -31% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 61% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -54% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -25% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 90% | 0% |
/write <brief_title> --format blog|guide|case-study|newsletter]
Produce content from brief to polished draft with SEO optimization and metadata.
| Arg | Type | Required | Description | |-----|------|----------|-------------| | brief_title | string | Yes | Title or ID of the content brief | | --format | string | No | Content format. Default: blog |
Genre: article Format: Markdown draft with metadata
Produces:
1. Load content brief from editor-in-chief
2. Writer researches and produces first draft
3. Writer self-edits (cut 20%, check readability, verify keywords)
4. SEO specialist reviews keyword placement and optimization
5. Submit to editor-in-chief for editorial review/write "7 SEO Mistakes That Kill Rankings"
/write "Q1 Customer Success Story" --format case-study
/write "Weekly Insights #42" --format newsletter| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-10 | fail→pass | 14,844 | 7,685 | -48% | 1 | 1 | 0% | 2,400 | 1,654 | -31% | 0 | 0 | — |
case-01 | pass→pass | 17,418 | 22,088 | +27% | 1 | 1 | 0% | 2,286 | 3,296 | +44% | 0 | 0 | — |
case-02 | pass→pass | 19,662 | 18,005 | -8% | 1 | 1 | 0% | 2,645 | 3,176 | +20% | 0 | 0 | — |
case-03 | pass→pass | 14,354 | 17,897 | +25% | 1 | 1 | 0% | 2,377 | 3,208 | +35% | 0 | 0 | — |
case-04 | fail→pass | 17,554 | 26,320 | +50% | 1 | 1 | 0% | 2,469 | 3,974 | +61% | 0 | 0 | — |
case-05 | fail→fail | 23,381 | 26,955 | +15% | 1 | 1 | 0% | 3,450 | 4,104 | +19% | 0 | 0 | — |
case-06 | pass→pass | 15,522 | 17,643 | +14% | 1 | 1 | 0% | 2,426 | 3,338 | +38% | 0 | 0 | — |
case-07 | pass→pass | 13,265 | 24,378 | +84% | 1 | 1 | 0% | 2,393 | 4,129 | +73% | 0 | 0 | — |
case-08 | pass→pass | 10,798 | 16,391 | +52% | 1 | 1 | 0% | 1,596 | 2,872 | +80% | 0 | 0 | — |
case-09 | fail→pass | 8,357 | 1,774 | -79% | 1 | 1 | 0% | 1,280 | 594 | -54% | 0 | 0 | — |
case-11 | fail→pass | 6,046 | 2,284 | -62% | 1 | 1 | 0% | 888 | 668 | -25% | 0 | 0 | — |
case-12 | fail→pass | 13,365 | 25,246 | +89% | 1 | 1 | 0% | 2,461 | 4,680 | +90% | 0 | 0 | — |
case-13 | fail→pass | 13,739 | 3,685 | -73% | 1 | 1 | 0% | 2,081 | 890 | -57% | 0 | 0 | — |
case-14 | fail→pass | 2,727 | 1,703 | -38% | 1 | 1 | 0% | 336 | 503 | +50% | 0 | 0 | — |
case-15 | pass→pass | 8,929 | 1,405 | -84% | 1 | 1 | 0% | 1,308 | 492 | -62% | 0 | 0 | — |
case-16 | pass→pass | 12,582 | 3,194 | -75% | 1 | 1 | 0% | 1,810 | 826 | -54% | 0 | 0 | — |
case-17 | fail→pass | 4,578 | 2,992 | -35% | 1 | 1 | 0% | 422 | 694 | +64% | 0 | 0 | — |
case-18 | fail→pass | 7,073 | 1,691 | -76% | 1 | 1 | 0% | 1,173 | 602 | -49% | 0 | 0 | — |
case-19 | fail→pass | 5,169 | 1,631 | -68% | 1 | 1 | 0% | 818 | 530 | -35% | 0 | 0 | — |
case-20 | fail→fail | 3,481 | 19,212 | +452% | 1 | 1 | 0% | 550 | 3,378 | +514% | 0 | 0 | — |
case-21 | fail→pass | 10,797 | 13,048 | +21% | 1 | 1 | 0% | 2,038 | 2,746 | +35% | 0 | 0 | — |
case-22 | pass→pass | 8,301 | 6,164 | -26% | 1 | 1 | 0% | 1,382 | 1,283 | -7% | 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 +50 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.