技能创建器
skill-creator
创建、改进和评估自定义技能,把专业流程转化为可重复使用的 AI 能力。
survey-theoretical-literature【数据与研究】Survey and synthesize theoretical literature on a specific topic, identifying seminal papers, key results, open problems, and cross-domain connections. Use when starting research on an unfamiliar theoretical topic, writing a literature review for a paper or thesis, identifying open problems and research gaps, finding cross-domain connections, or eval…
独立 GitHub 快照 · 自动分类 · 以原作者文档为准
於某題行構之理論文獻察,生綜合——映種開之文、追要念之時序、識開題與活前線、顯跨域之連。
搜前精界察:
## Survey Scope
- **Core topic**: [one-sentence definition]
- **Primary search terms**: [list]
- **Secondary search terms**: [list]
- **Exclusion terms**: [list]
- **Time window**: [start year] to [end year]
- **In scope**: [subfields]
- **Out of scope**: [subfields]
得:範緊足以使二究者獨立同某文之屬察否。
敗則:若範太廣(200 以上之或關之文),加子域之限或緊時窗以縮之。若太窄(10 以下),廣次搜詞或延時窗。
由最影響之貢建察之骨:
## Seminal Papers
| # | Authors (Year) | Title | Main Result | Impact |
|---|---------------|-------|-------------|--------|
| 1 | [authors] ([year]) | [title] | [one-sentence result] | [influence on field] |
| 2 | ... | ... | ... | ... |
得:5-15 種開文之表為題之智之骨,各文之主果與影響明陳。
敗則:若搜無清種開文,題或太新或太隘。識最早文與最被引文為錨,注域之典參未現。
追域自源至今之化:
## Chronological Development
### Origin ([decade])
- [event/paper]: [description of foundational contribution]
### Key Developments
- **[year]**: [milestone and its significance]
- **[year]**: [milestone and its significance]
- ...
### Branching Points
- **[year]**: Field splits into [branch A] and [branch B]
- Branch A focuses on [topic]
- Branch B focuses on [topic]
### Current State ([year])
- **Activity level**: [mature / active / emerging / stagnant]
- **Dominant approach**: [current mainstream methodology]
- **Recent trend**: [direction of latest work]
得:新者可讀以解域如何至今之敘時線,含要念之智脈。
敗則:若時序不明(如多獨發、爭優先),書其歧而非強為線敘。並列時線可也。
列尚未知或未解者:
## Open Problems and Frontiers
### Explicitly Open
| # | Problem | Status | Barrier | Potential Impact |
|---|---------|--------|---------|-----------------|
| 1 | [statement] | [conjecture / partial / open] | [why hard] | [incremental / significant / transformative] |
| 2 | ... | ... | ... | ... |
### Active Frontiers
- **[frontier topic]**: [what is happening and why it matters]
- ...
### Implicit Gaps
- [observation without theoretical explanation]
- [conjecture without proof]
- ...
得:構之開題列,至少 3-5 題並難之評,並最活前線之述。
敗則:若無開題顯,察範或太隘(子題已解)或文獻搜失關之評文。廣範或專搜「open problems in [topic]」與「future directions in [topic]」。
連察域於鄰域而合終出:
跨域之連:識察題與他域連於:
連質之評:各連評其為:
缺析:識當有而未探之連。乃潛察機。
察之合:合一至五步之出為構文:
## Cross-Domain Connections
| # | Connected Field | Type of Connection | Depth | Key Reference |
|---|----------------|-------------------|-------|---------------|
| 1 | [field] | [shared math / analogy / method import] | [deep / promising / superficial] | [paper] |
| 2 | ... | ... | ... | ... |
## Unexplored Connections (Research Opportunities)
- [potential connection]: [why it might exist and what it could yield]
- ...
得:完之構之察文,自源至今前線圖題,跨域之連已識而評。
敗則:若察感散,回時線(第三步)為組之脊。每種開文、開題、跨域之連皆當位於時線。
formulate-quantum-problem — 立文獻察中所識之具題derive-theoretical-result — 推或重推察文中所現之要果review-research — 評察中所遇之個文