Install any skill in seconds. Free to start, no credit card required.
Get Started Free →SOP: Identify theoretical frameworks relevant to a research gap
.claude/skills/yogsoth-ai-theory-identification/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -12% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 8% | 0% |
| case-18 | ✗→✓ | ▲ Improved | -53% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -42% | 0% |
| case-08 | ✓→✗ | ▼ Worse | -70% | 0% |
Identify theoretical frameworks relevant to a research gap, providing a theoretical basis for mechanism extraction.
<HARD-GATE> Preconditions (all must hold before starting):
Not satisfied → stop, return error: missing gap description or domain tags. </HARD-GATE>
json[ { "name": "Theory Name", "source": "Author (Year) or canonical reference", "core_claim": "One-sentence summary of what the theory claims", "relevance": "Why this theory relates to the gap", "applicability": "high | medium | low" } ]
Minimum 3 entries, maximum 8 (sorted by applicability descending).
<!-- BEGIN available-tables (generated) -->
Optional, no fixed order; the final leaf is always a sop.
| SOP | When to use | | --- | --- | | hypothesis-formation-paper-overview | Import SOP: Quick paper scan, returns abstract and metadata (from literature-engine) | | hypothesis-formation-paper-search | Import SOP: Medium-depth literature search, AI summary report (from literature-engine) | | hypothesis-formation-web-search | Import SOP: quick web scan, discover URLs and snippets (from web-browsing) |
<!-- 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 | fail→pass | 23,404 | 21,485 | -8% | 1 | 1 | 0% | 3,997 | 3,531 | -12% | 0 | 0 | — |
case-02 | fail→pass | 21,654 | 30,155 | +39% | 1 | 1 | 0% | 3,345 | 3,625 | +8% | 0 | 0 | — |
case-03 | fail→fail | 22,119 | 39,407 | +78% | 1 | 1 | 0% | 3,046 | 950 | -69% | 0 | 0 | — |
case-18 | fail→pass | 14,891 | 7,018 | -53% | 1 | 1 | 0% | 1,634 | 762 | -53% | 0 | 0 | — |
case-04 | fail→fail | 18,742 | 23,787 | +27% | 1 | 1 | 0% | 2,999 | 3,973 | +32% | 0 | 0 | — |
case-05 | fail→fail | 22,504 | 11,615 | -48% | 1 | 1 | 0% | 2,610 | 830 | -68% | 0 | 0 | — |
case-06 | fail→pass | 14,059 | 7,887 | -44% | 1 | 1 | 0% | 1,525 | 885 | -42% | 0 | 0 | — |
case-07 | fail→fail | 30,123 | 23,111 | -23% | 1 | 1 | 0% | 2,375 | 1,113 | -53% | 0 | 0 | — |
case-08 | pass→fail | 23,506 | 25,600 | +9% | 1 | 1 | 0% | 2,618 | 797 | -70% | 0 | 0 | — |
case-09 | pass→fail | 18,168 | 17,493 | -4% | 1 | 1 | 0% | 2,640 | 1,076 | -59% | 0 | 0 | — |
case-10 | fail→fail | 17,249 | 29,351 | +70% | 1 | 1 | 0% | 2,292 | 1,049 | -54% | 0 | 0 | — |
case-11 | pass→pass | 16,237 | 18,691 | +15% | 1 | 1 | 0% | 2,050 | 2,835 | +38% | 0 | 0 | — |
case-12 | fail→fail | 19,243 | 15,515 | -19% | 1 | 1 | 0% | 2,557 | 763 | -70% | 0 | 0 | — |
case-13 | fail→fail | 20,039 | 16,918 | -16% | 1 | 1 | 0% | 2,512 | 855 | -66% | 0 | 0 | — |
case-14 | pass→fail | 18,950 | 17,104 | -10% | 1 | 1 | 0% | 2,197 | 862 | -61% | 0 | 0 | — |
case-15 | fail→fail | 16,984 | 12,802 | -25% | 1 | 1 | 0% | 1,998 | 829 | -59% | 0 | 0 | — |
case-16 | pass→fail | 24,001 | 12,679 | -47% | 1 | 1 | 0% | 2,976 | 810 | -73% | 0 | 0 | — |
case-17 | fail→fail | 17,118 | 6,705 | -61% | 1 | 1 | 0% | 2,268 | 857 | -62% | 0 | 0 | — |
case-19 | pass→fail | 16,128 | 26,691 | +65% | 1 | 1 | 0% | 2,543 | 790 | -69% | 0 | 0 | — |
case-20 | pass→fail | 19,550 | 21,674 | +11% | 1 | 1 | 0% | 2,413 | 3,568 | +48% | 0 | 0 | — |
case-21 | pass→fail | 18,276 | 11,072 | -39% | 1 | 1 | 0% | 2,202 | 1,564 | -29% | 0 | 0 | — |
case-22 | pass→fail | 18,275 | 11,221 | -39% | 1 | 1 | 0% | 2,497 | 1,623 | -35% | 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, and 9 counted toward the lift figure. The other 13 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 -18 percentage points is the difference between those two pass rates over the 9 comparable cases. 10 cases got worse with the skill loaded, and they are 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.