Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →当 ai-shifu 以受控方式接入 MarkdownFlowEditor，并出现变量中间态、临时标签或本地输入回灌问题时使用本技能。
.claude/skills/ai-shifu-markdownflow-controlled-sync/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -19% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -8% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -13% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 22% | 0% |
| case-13 | ✗→✓ | ▲ Improved | -40% | 0% |
MarkdownFlowEditor 在 ai-shifu 中作为受控组件使用，父层会在 onChange 后立刻更新 content。content 的自动提取逻辑，导致本地输入中的中间态被错误保留，例如临时变量名进入变量列表。content 回灌给编辑器。ref 标记最近一次本地输入，只跳过对应的本地 echo；真正的外部更新仍然要同步到编辑器。123、1234 这类中间态变量名。| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 24,733 | 19,254 | -22% | 1 | 1 | 0% | 4,307 | 3,502 | -19% | 0 | 0 | — |
case-02 | fail→pass | 29,138 | 23,028 | -21% | 1 | 1 | 0% | 4,938 | 4,537 | -8% | 0 | 0 | — |
case-03 | fail→pass | 21,743 | 16,169 | -26% | 1 | 1 | 0% | 3,249 | 2,817 | -13% | 0 | 0 | — |
case-04 | pass→pass | 37,266 | 18,770 | -50% | 1 | 1 | 0% | 3,665 | 3,128 | -15% | 0 | 0 | — |
case-05 | pass→pass | 18,032 | 21,301 | +18% | 1 | 1 | 0% | 2,995 | 3,797 | +27% | 0 | 0 | — |
case-06 | pass→pass | 18,096 | 16,817 | -7% | 1 | 1 | 0% | 2,901 | 2,786 | -4% | 0 | 0 | — |
case-07 | pass→pass | 16,810 | 15,478 | -8% | 1 | 1 | 0% | 2,554 | 2,866 | +12% | 0 | 0 | — |
case-08 | pass→pass | 19,705 | 17,443 | -11% | 1 | 1 | 0% | 2,974 | 2,886 | -3% | 0 | 0 | — |
case-09 | fail→pass | 24,564 | 22,578 | -8% | 1 | 1 | 0% | 3,330 | 4,046 | +22% | 0 | 0 | — |
case-10 | pass→pass | 22,311 | 35,591 | +60% | 1 | 1 | 0% | 3,127 | 3,051 | -2% | 0 | 0 | — |
case-11 | pass→pass | 17,536 | 16,135 | -8% | 1 | 1 | 0% | 2,617 | 2,758 | +5% | 0 | 0 | — |
case-12 | pass→pass | 16,067 | 15,929 | -1% | 1 | 1 | 0% | 2,638 | 2,683 | +2% | 0 | 0 | — |
case-13 | fail→pass | 18,695 | 8,935 | -52% | 1 | 1 | 0% | 2,861 | 1,703 | -40% | 0 | 0 | — |
case-14 | fail→pass | 21,064 | 13,464 | -36% | 1 | 1 | 0% | 2,862 | 2,264 | -21% | 0 | 0 | — |
case-15 | fail→pass | 12,022 | 9,542 | -21% | 1 | 1 | 0% | 1,820 | 1,682 | -8% | 0 | 0 | — |
case-16 | pass→pass | 20,200 | 17,077 | -15% | 1 | 1 | 0% | 3,122 | 3,015 | -3% | 0 | 0 | — |
case-17 | pass→pass | 22,097 | 17,306 | -22% | 1 | 1 | 0% | 2,947 | 2,707 | -8% | 0 | 0 | — |
case-18 | fail→pass | 20,878 | 17,332 | -17% | 1 | 1 | 0% | 3,279 | 2,904 | -11% | 0 | 0 | — |
case-19 | pass→pass | 19,804 | 17,978 | -9% | 1 | 1 | 0% | 2,961 | 3,126 | +6% | 0 | 0 | — |
case-20 | pass→pass | 16,521 | 14,664 | -11% | 1 | 1 | 0% | 2,722 | 2,688 | -1% | 0 | 0 | — |
case-21 | pass→pass | 23,082 | 22,509 | -2% | 1 | 1 | 0% | 4,144 | 4,242 | +2% | 0 | 0 | — |
case-22 | pass→pass | 20,397 | 19,862 | -3% | 1 | 1 | 0% | 3,920 | 3,685 | -6% | 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 +36 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.