Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Theory-driven reading tactic — define a theoretical framework first, then guide reading to fill it with evidence. Five stages (theme identification, argument construction, evidence collection, counter-evidence, synthesis). The most intellectually demanding tactic.
.claude/skills/yogsoth-ai-narrative-framing/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | -10% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 14% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -50% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -32% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -13% | 0% |
Define a theoretical framework first, then guide reading to fill it with evidence.
paper-search (import) — targeted reading for evidencepaper-research (import) — deep reading of key supporting/opposing papersthematic-coding (subagent) — identify patterns across papersweb-research (import) — blogs, opinion pieces, industry perspectives<!-- BEGIN available-tables (generated) -->
Optional, no fixed order; the final leaf is always a sop.
| SOP | When to use | | --- | --- | | knowledge-acquisition-paper-research | Full-depth paper reading with raw text extraction. Import of literature-engine/literature-research skill. Must read fullText (true) — equations, hyperparameters, specific claims extracted. | | knowledge-acquisition-paper-search | AI-summarized paper reading for intermediate depth. Import of literature-engine/literature-search skill. Must call get_paper_content for every analyzed paper. | | knowledge-acquisition-web-research | Full-page web reading for non-academic perspectives — blogs, tech reports, product pages, industry analysis. Import of web-browsing/web-research skill. Must fetch full page via apify for every analyzed page. | | thematic-coding | Identify recurring themes across papers using qualitative coding methodology. Produces a codebook with theme definitions, supporting evidence, and frequency counts. Used by narrative-review. |
<!-- 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→fail | 17,320 | 29,654 | +71% | 1 | 1 | 0% | 2,029 | 3,550 | +75% | 0 | 0 | — |
case-02 | pass→pass | 21,228 | 35,414 | +67% | 1 | 1 | 0% | 2,425 | 3,857 | +59% | 0 | 0 | — |
case-03 | fail→pass | 21,839 | 17,851 | -18% | 1 | 1 | 0% | 2,731 | 2,453 | -10% | 0 | 0 | — |
case-04 | fail→pass | 29,312 | 19,891 | -32% | 1 | 1 | 0% | 2,433 | 2,773 | +14% | 0 | 0 | — |
case-05 | pass→pass | 15,378 | 3,575 | -77% | 1 | 1 | 0% | 2,565 | 1,185 | -54% | 0 | 0 | — |
case-06 | fail→pass | 20,007 | 8,845 | -56% | 1 | 1 | 0% | 2,254 | 1,135 | -50% | 0 | 0 | — |
case-07 | fail→pass | 18,102 | 10,281 | -43% | 1 | 1 | 0% | 1,959 | 1,339 | -32% | 0 | 0 | — |
case-08 | fail→fail | 16,286 | 3,132 | -81% | 1 | 1 | 0% | 1,834 | 1,059 | -42% | 0 | 0 | — |
case-09 | fail→fail | 23,882 | 21,061 | -12% | 1 | 1 | 0% | 2,920 | 1,190 | -59% | 0 | 0 | — |
case-10 | fail→fail | 17,061 | 15,884 | -7% | 1 | 1 | 0% | 1,910 | 2,158 | +13% | 0 | 0 | — |
case-11 | fail→pass | 17,483 | 11,244 | -36% | 1 | 1 | 0% | 1,908 | 1,658 | -13% | 0 | 0 | — |
case-12 | pass→pass | 17,696 | 8,999 | -49% | 1 | 1 | 0% | 2,138 | 1,969 | -8% | 0 | 0 | — |
case-13 | fail→pass | 64,133 | 35,399 | -45% | 1 | 1 | 0% | 4,222 | 5,221 | +24% | 0 | 0 | — |
case-14 | fail→pass | 22,869 | 16,716 | -27% | 1 | 1 | 0% | 2,708 | 3,466 | +28% | 0 | 0 | — |
case-15 | fail→pass | 26,887 | 32,958 | +23% | 1 | 1 | 0% | 2,743 | 4,924 | +80% | 0 | 0 | — |
case-16 | fail→pass | 7,266 | 4,592 | -37% | 1 | 1 | 0% | 1,124 | 1,274 | +13% | 0 | 0 | — |
case-17 | pass→pass | 20,768 | 3,751 | -82% | 1 | 1 | 0% | 3,212 | 1,216 | -62% | 0 | 0 | — |
case-18 | pass→pass | 13,133 | 13,571 | +3% | 1 | 1 | 0% | 1,961 | 2,546 | +30% | 0 | 0 | — |
case-19 | pass→pass | 19,305 | 14,265 | -26% | 1 | 1 | 0% | 2,952 | 2,729 | -8% | 0 | 0 | — |
case-20 | pass→pass | 14,023 | 17,780 | +27% | 1 | 1 | 0% | 2,217 | 2,766 | +25% | 0 | 0 | — |
case-21 | pass→pass | 4,510 | 5,863 | +30% | 1 | 1 | 0% | 760 | 1,451 | +91% | 0 | 0 | — |
case-22 | pass→pass | 8,651 | 6,804 | -21% | 1 | 1 | 0% | 1,274 | 1,576 | +24% | 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 21 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 +41 percentage points is the difference between those two pass rates over the 21 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.