Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Create a personal athlete 81-cell MandalArt grid from an Ohtani Shohei-style 64+8+1 model. Use when the user asks for 大谷翔平 81 宮格, 個人運動員81宮格, sports skill maps, athlete training Mandala charts, badminton 81 grids, or editable JSON/SVG/PNG-ready athlete development templates with Ohtani-style colors.
.claude/skills/twhsi-personal-athlete-81-grid/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 18% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -8% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 33% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 51% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -51% | 0% |
Turn an athlete's core goal into a square 9x9 MandalArt grid:
1 center goal: the athlete's north star.8 domains: the center 3x3 ring around the goal.64 extension cells: each domain expands into eight concrete actions.Default visual style is the Ohtani Shohei reference: thick black 3x3 section lines, short bold text, red center, pink domain centers/core layer, white hard skills, cyan mental/social/recovery/long-term layer.
5 hard or sport-skill domains and 3 long-term domains.For badminton, a good default domain set is:
text後場攻防、切吊變化、網前手感、重心步法、發接前三、恢復保養、球友情場、長壽榜樣
Use this compact shape:
json{ "title": "永錫羽毛球81宮格", "center": "越老越健康\n越久越快樂\n重心強", "visual_style": "ohtani", "domains": [ { "name": "後場攻防", "type": "hard_skill", "items": ["側身準備", "高遠拉開", "殺球角度", "一殺一抽", "抽球拍面", "殺後銜接", "切吊變化", "教練回饋"] } ] }
Rules:
domains.items.items; the renderer places it in each outer 3x3 center.center only when needed.Use these colors unless the user asks for another palette:
| Layer | Default color | Meaning | |---|---|---| | Center goal | red #ff260f | final target, identity-level athletic goal | | Domain cells | pink #fb8aa0 | the eight main domains and outer domain centers | | Hard skills | white #ffffff | sport technique, body mechanics, concrete drills | | Mental/social/long-term | cyan #65eadb | psychology, character, recovery, relationships, longevity | | Section lines | black #050505 | 3x3 blocks | | Cell lines | gray #b9b9b9 | individual cells |
Map domain types:
hard_skill: white outer cells, pink domain center.body: white outer cells, pink domain center.recovery: cyan outer cells, pink domain center.mental: cyan outer cells, pink domain center.social: cyan outer cells, pink domain center.longevity: cyan outer cells, pink domain center.Use the bundled script when a deterministic SVG is useful:
bashpython3 scripts/render_ohtani_81_grid.py assets/yongxi-badminton.json output.svg
Then convert SVG to PNG with a local tool if available, such as:
bashqlmanage -t -s 2400 -o . output.svg
The script expects the JSON schema above and outputs a square 2400x2400 SVG.
When the user gives only a sport and goal, use this pattern internally:
textCreate an Ohtani-style personal athlete 81 grid for SPORT. Center goal: GOAL. Return 8 domains, each with 8 short action cells. Use 5 sport-skill/body domains and 3 recovery/mental/social/longevity domains. Keep labels concise enough for a square 9x9 grid. Use JSON as the editable source.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 18,534 | 13,564 | -27% | 1 | 1 | 0% | 2,998 | 3,531 | +18% | 0 | 0 | — |
case-02 | fail→pass | 29,245 | 18,772 | -36% | 1 | 1 | 0% | 4,803 | 4,439 | -8% | 0 | 0 | — |
case-03 | fail→pass | 13,190 | 12,783 | -3% | 1 | 1 | 0% | 2,492 | 3,307 | +33% | 0 | 0 | — |
case-04 | pass→pass | 17,666 | 17,523 | -1% | 1 | 1 | 0% | 3,063 | 4,122 | +35% | 0 | 0 | — |
case-05 | pass→pass | 21,865 | 18,910 | -14% | 1 | 1 | 0% | 3,206 | 3,932 | +23% | 0 | 0 | — |
case-06 | pass→pass | 11,907 | 10,318 | -13% | 1 | 1 | 0% | 1,719 | 2,349 | +37% | 0 | 0 | — |
case-07 | fail→pass | 11,905 | 11,553 | -3% | 1 | 1 | 0% | 2,030 | 3,060 | +51% | 0 | 0 | — |
case-08 | pass→pass | 16,823 | 13,237 | -21% | 1 | 1 | 0% | 2,804 | 3,348 | +19% | 0 | 0 | — |
case-09 | fail→pass | 16,682 | 1,732 | -90% | 1 | 1 | 0% | 2,493 | 1,225 | -51% | 0 | 0 | — |
case-10 | fail→pass | 15,539 | 2,509 | -84% | 1 | 1 | 0% | 2,509 | 1,428 | -43% | 0 | 0 | — |
case-11 | fail→pass | 12,952 | 2,601 | -80% | 1 | 1 | 0% | 2,096 | 1,405 | -33% | 0 | 0 | — |
case-12 | fail→pass | 10,540 | 2,254 | -79% | 1 | 1 | 0% | 1,691 | 1,334 | -21% | 0 | 0 | — |
case-13 | fail→pass | 12,052 | 2,576 | -79% | 1 | 1 | 0% | 1,887 | 1,300 | -31% | 0 | 0 | — |
case-14 | pass→pass | 22,751 | 12,139 | -47% | 1 | 1 | 0% | 3,689 | 3,235 | -12% | 0 | 0 | — |
case-15 | fail→pass | 26,079 | 7,863 | -70% | 1 | 1 | 0% | 1,112 | 2,276 | +105% | 0 | 0 | — |
case-16 | pass→pass | 8,094 | 1,993 | -75% | 1 | 1 | 0% | 1,127 | 1,259 | +12% | 0 | 0 | — |
case-17 | fail→pass | 22,351 | 11,359 | -49% | 1 | 1 | 0% | 3,633 | 3,198 | -12% | 0 | 0 | — |
case-18 | fail→pass | 10,866 | 1,661 | -85% | 1 | 1 | 0% | 1,972 | 1,266 | -36% | 0 | 0 | — |
case-19 | fail→fail | 9,617 | 3,102 | -68% | 1 | 1 | 0% | 1,511 | 1,474 | -2% | 0 | 0 | — |
case-20 | pass→pass | 10,805 | 6,187 | -43% | 1 | 1 | 0% | 1,555 | 2,029 | +30% | 0 | 0 | — |
case-21 | pass→pass | 13,444 | 8,451 | -37% | 1 | 1 | 0% | 2,235 | 2,390 | +7% | 0 | 0 | — |
case-22 | pass→pass | 7,391 | 6,854 | -7% | 1 | 1 | 0% | 1,275 | 2,182 | +71% | 0 | 0 | — |
case-23 | pass→pass | 18,402 | 13,911 | -24% | 1 | 1 | 0% | 2,863 | 3,382 | +18% | 0 | 0 | — |
case-24 | fail→pass | 8,710 | 1,738 | -80% | 1 | 1 | 0% | 1,410 | 1,226 | -13% | 0 | 0 | — |
case-25 | fail→pass | 13,339 | 2,045 | -85% | 1 | 1 | 0% | 1,898 | 1,301 | -31% | 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. 25 cases were attempted, and 24 counted toward the lift figure. The other 1 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 +56 percentage points is the difference between those two pass rates over the 24 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.