Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Autonomous Literature Survey Campaign — 5 research paradigms (scoping, systematic, deep, narrative, snowball) with quantitative budget enforcement. Selects and executes the right survey paradigm based on research intent.
.claude/skills/yogsoth-ai-literature-survey/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-06 | ✗→✓ | ▲ Improved | 16% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 36% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -21% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 29% | 0% |
| case-14 | ✗→✓ | ▲ Improved | 17% | 0% |
Autonomous literature survey engine. Five research paradigms, each a self-contained playbook with quantitative budget enforcement. You provide a research question — it searches, screens, reads, categorizes, identifies gaps, and produces a structured survey output.
| Signal | Strategy | |--------|----------| | panoramic mapping of a new field, broad overview, field mapping | → scoping-survey | | exhaustive coverage, PRISMA, systematic review | → systematic-survey | | precise sub-question, specific mechanism, deep dive | → deep-survey | | theory-driven, argument building, narrative | → narrative-review | | seed papers, citation chain, lineage tracing | → snowball |
| Strategy | web-search | web-research | paper-overview | paper-search | paper-research | |----------|-----------|-------------|---------------|-------------|---------------| | scoping-survey | 100 | 10 | 100 | 20 | 0 | | systematic-survey | 50 | 5 | 60 | 40 | 30 | | deep-survey | 30 | 5 | 40 | 40 | 20 | | narrative-review | 80 | 15 | 50 | 40 | 20 | | snowball | 20 | 3 | 30 | 30 | 20 |
All values ±10% flexibility. Deviations require explicit reasoning.
| MCP Server | Tools | |------------|-------| | brave-search | brave_web_search, brave_news_search, brave_llm_context | | apify | rag-web-browser, google-scholar-scraper | | alphaxiv | discover_papers, get_paper_content, answer_pdf_queries, read_files_from_github_repository | | semantic-scholar | ss_paper, ss_paper_batch, ss_references, ss_citations, ss_recommendations, ss_relevance_search, ss_author, ss_author_papers |
<!-- BEGIN available-tables (generated) -->
Optional, no fixed order; the final leaf is always a sop.
| Strategy | When to use | | --- | --- | | deep-survey | Precise, targeted investigation of a specific sub-problem — few papers, all read in full depth. High paper-research ratio (50% deep-read rate). Use when the user knows exactly what they need to understand and requires detailed technical analysis with equations, hyperparameters, and specific claims extracted. | | narrative-review | Theory-driven literature review for building arguments and frameworks. Flexible, subjective, and narrative-focused — selects evidence strategically to support a thesis. High web-research budget for blogs, opinion pieces, and industry perspectives. Use when the user is writing a position paper, survey introduction, or constructing a coherent narrative around a research theme. | | scoping-survey | Broad landscape mapping strategy — quickly understand what exists in a field. Prioritizes breadth over depth with high paper-overview volume and minimal deep reading. Use when entering a new field or needing orientation before committing to deeper investigation. | | snowball | Citation-chain-driven literature survey starting from seed papers. Traces research lineage in both forward (who cited this) and backward (what this cited) directions until saturation. High deep-read ratio (67%). Use when the user already has key papers and wants to find everything connected to them — ancestors, descendants, and branch points. | | systematic-survey | Exhaustive PRISMA-style literature survey — comprehensive coverage of all related work on a specific question. Multi-stage screening, citation chaining, quality assessment, and structured data extraction. Use when the user needs to demonstrate complete literature coverage or conduct rigorous gap analysis. |
Optional, no fixed order; the final leaf is always a sop.
| SOP | When to use | | --- | --- | | context-checkpoint | Append research process and results to the current Phase's context file. Covers both process and results with genuine substance. Use this skill at plan-designated checkpoint points — typically after each strategy completes or at key decision nodes within a research Phase. | | context-init | Create a new context file for a research Phase. Called once at Phase start to initialize the file that subsequent context-checkpoint calls will append to. Use this skill whenever a new research Phase begins and a fresh context file is needed. |
<!-- 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 | 72,253 | 73,804 | +2% | 1 | 1 | 0% | 8,250 | 2,345 | -72% | 0 | 0 | — |
case-02 | fail→fail | 44,338 | 39,719 | -10% | 1 | 1 | 0% | 8,246 | 2,510 | -70% | 0 | 0 | — |
case-03 | fail→fail | 45,685 | 12,274 | -73% | 1 | 1 | 0% | 8,266 | 1,998 | -76% | 0 | 0 | — |
case-04 | fail→fail | 45,999 | 29,815 | -35% | 1 | 1 | 0% | 8,251 | 2,376 | -71% | 0 | 0 | — |
case-05 | fail→fail | 30,768 | 30,902 | +0% | 1 | 1 | 0% | 3,364 | 3,088 | -8% | 0 | 0 | — |
case-06 | fail→pass | 14,661 | 15,066 | +3% | 1 | 1 | 0% | 2,461 | 2,852 | +16% | 0 | 0 | — |
case-07 | fail→pass | 6,778 | 9,261 | +37% | 1 | 1 | 0% | 1,400 | 1,909 | +36% | 0 | 0 | — |
case-08 | pass→pass | 12,161 | 3,934 | -68% | 1 | 1 | 0% | 1,348 | 1,720 | +28% | 0 | 0 | — |
case-09 | pass→pass | 35,792 | 18,713 | -48% | 1 | 1 | 0% | 3,058 | 3,939 | +29% | 0 | 0 | — |
case-10 | fail→pass | 14,712 | 10,279 | -30% | 1 | 1 | 0% | 2,499 | 1,976 | -21% | 0 | 0 | — |
case-11 | fail→pass | 9,092 | 11,261 | +24% | 1 | 1 | 0% | 1,493 | 1,925 | +29% | 0 | 0 | — |
case-12 | pass→pass | 12,837 | 5,855 | -54% | 1 | 1 | 0% | 1,420 | 1,694 | +19% | 0 | 0 | — |
case-13 | pass→pass | 13,061 | 11,000 | -16% | 1 | 1 | 0% | 1,513 | 1,888 | +25% | 0 | 0 | — |
case-14 | fail→pass | 21,832 | 5,191 | -76% | 1 | 1 | 0% | 1,553 | 1,814 | +17% | 0 | 0 | — |
case-24 | pass→pass | 33,649 | 33,053 | -2% | 1 | 1 | 0% | 6,481 | 8,414 | +30% | 0 | 0 | — |
case-15 | pass→pass | 16,842 | 8,250 | -51% | 1 | 1 | 0% | 2,364 | 2,010 | -15% | 0 | 0 | — |
case-16 | fail→fail | 18,251 | 27,233 | +49% | 1 | 1 | 0% | 2,866 | 4,732 | +65% | 0 | 0 | — |
case-17 | fail→pass | 25,221 | 5,528 | -78% | 1 | 1 | 0% | 4,127 | 2,235 | -46% | 0 | 0 | — |
case-18 | pass→pass | 12,086 | 3,004 | -75% | 1 | 1 | 0% | 1,636 | 1,728 | +6% | 0 | 0 | — |
case-19 | fail→pass | 12,539 | 3,001 | -76% | 1 | 1 | 0% | 2,382 | 1,694 | -29% | 0 | 0 | — |
case-20 | fail→pass | 19,041 | 2,421 | -87% | 1 | 1 | 0% | 3,421 | 1,622 | -53% | 0 | 0 | — |
case-21 | fail→pass | 9,058 | 4,199 | -54% | 1 | 1 | 0% | 1,285 | 1,988 | +55% | 0 | 0 | — |
case-22 | fail→fail | 17,294 | 39,303 | +127% | 1 | 1 | 0% | 3,053 | 9,470 | +210% | 0 | 0 | — |
case-23 | pass→pass | 24,437 | 25,176 | +3% | 1 | 1 | 0% | 4,800 | 6,543 | +36% | 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. 24 cases were attempted, and 22 counted toward the lift figure. The other 2 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 +38 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.