Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Fill a slot-based question-framing schema (PICO, PECO, or SPIDER) from a paper's stated research question. Use this whenever the user wants a paper's research question structured into one of these standard clinical/qualitative-research question frames; this frames what question is being asked, it does not read or evaluate the paper's content otherwise.
.claude/skills/yogsoth-ai-question-framing/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-10 | ✗→✓ | ▲ Improved | 0% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 49% | 0% |
| case-17 | ✗→✓ | ▲ Improved | -19% | 0% |
| case-22 | ✗→✓ | ▲ Improved | 14% | 0% |
| case-01 | ✓→✗ | ▼ Worse | 61% | 0% |
Slot-filling into PICO/PECO/SPIDER, parameterized on which schema — defines the question being asked, doesn't evaluate paper content.
Subagent — spawned via spawn-agent skill.
This SOP fills slots to describe what question is asked. research-question-appraisal (FINER) instead judges whether a research question is good — different structure, different SOP, not a parameterization of this one (see graph correction M15).
<!-- BEGIN available-tables (generated) -->
| SOP | When to use | | --- | --- | | spawn-agent | Spawn a customized CC subagent with full MCP tool access. |
<!-- END available-tables (generated) -->
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | pass→fail | 21,251 | 45,620 | +115% | 1 | 1 | 0% | 2,689 | 4,322 | +61% | 0 | 0 | — |
case-02 | pass→pass | 31,308 | 30,202 | -4% | 1 | 1 | 0% | 3,913 | 4,543 | +16% | 0 | 0 | — |
case-03 | fail→fail | 12,487 | 9,508 | -24% | 1 | 1 | 0% | 1,159 | 859 | -26% | 0 | 0 | — |
case-04 | fail→fail | 18,376 | 18,609 | +1% | 1 | 1 | 0% | 2,215 | 2,436 | +10% | 0 | 0 | — |
case-05 | pass→pass | 16,643 | 13,533 | -19% | 1 | 1 | 0% | 1,736 | 1,612 | -7% | 0 | 0 | — |
case-06 | pass→pass | 20,069 | 20,349 | +1% | 1 | 1 | 0% | 2,434 | 2,454 | +1% | 0 | 0 | — |
case-07 | pass→pass | 18,742 | 20,483 | +9% | 1 | 1 | 0% | 2,282 | 2,741 | +20% | 0 | 0 | — |
case-08 | pass→pass | 19,953 | 19,815 | -1% | 1 | 1 | 0% | 2,398 | 2,496 | +4% | 0 | 0 | — |
case-09 | fail→fail | 14,177 | 17,656 | +25% | 1 | 1 | 0% | 1,448 | 2,316 | +60% | 0 | 0 | — |
case-10 | fail→pass | 16,008 | 29,872 | +87% | 1 | 1 | 0% | 1,860 | 1,852 | -0% | 0 | 0 | — |
case-11 | pass→pass | 14,104 | 12,775 | -9% | 1 | 1 | 0% | 1,772 | 1,490 | -16% | 0 | 0 | — |
case-12 | fail→pass | 20,857 | 24,210 | +16% | 1 | 1 | 0% | 2,500 | 3,729 | +49% | 0 | 0 | — |
case-13 | fail→fail | 23,883 | 17,350 | -27% | 1 | 1 | 0% | 3,072 | 2,281 | -26% | 0 | 0 | — |
case-14 | pass→pass | 12,223 | 9,786 | -20% | 1 | 1 | 0% | 1,199 | 992 | -17% | 0 | 0 | — |
case-15 | fail→fail | 18,948 | 15,856 | -16% | 1 | 1 | 0% | 2,122 | 1,968 | -7% | 0 | 0 | — |
case-16 | pass→pass | 16,145 | 14,952 | -7% | 1 | 1 | 0% | 1,825 | 1,927 | +6% | 0 | 0 | — |
case-17 | fail→pass | 20,964 | 26,122 | +25% | 1 | 1 | 0% | 2,532 | 2,043 | -19% | 0 | 0 | — |
case-18 | fail→fail | 26,496 | 16,061 | -39% | 1 | 1 | 0% | 3,334 | 2,121 | -36% | 0 | 0 | — |
case-19 | fail→fail | 18,575 | 21,986 | +18% | 1 | 1 | 0% | 2,200 | 3,057 | +39% | 0 | 0 | — |
case-20 | pass→pass | 13,827 | 13,943 | +1% | 1 | 1 | 0% | 1,466 | 1,589 | +8% | 0 | 0 | — |
case-21 | fail→fail | 19,641 | 18,999 | -3% | 1 | 1 | 0% | 2,232 | 2,400 | +8% | 0 | 0 | — |
case-22 | fail→pass | 20,827 | 21,280 | +2% | 1 | 1 | 0% | 2,598 | 2,960 | +14% | 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 +14 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.