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。
s4h-historical-precedent-analysis【数据与研究】Finds and applies genuinely similar historical situations to inform a current decision — distinguishing true precedents from superficial analogies. TRIGGERS: 'historical precedent', 'has this happened before', 'what does history say', 'find a precedent', 'what did others do in this situation'.
独立 GitHub 快照 · 自动分类 · 以原作者文档为准
History doesn't repeat — but structures do. The error is searching for surface similarity (same industry, same technology, same geography) and missing structural similarity (same underlying dynamics, same constraint set, same incentive conflicts). Superficial precedents produce false confidence. Structural precedents produce genuine insight. This skill finds the real ones.
Step 1: Abstract the Situation Strip away domain-specific language and surface details. What is the underlying structural pattern? Describe it in terms that could apply across industries and eras: a new entrant facing incumbents with switching-cost moats; a coalition with aligned goals but divergent interests trying to coordinate; a technology displacing a profession whose members control the adoption decision. State the situation in these structural terms — this is what you'll search for in history.
Framing check: Confirm the specific situation before continuing. State what you've identified — the actual situation being analyzed and its core structural pattern — in one sentence, then use AskUserQuestion:
Step 2: Search for Structural Precedents Deliberately look outside the obvious domain. The most obvious precedent (same industry, earlier decade) usually has the most surface similarity and the least structural insight — the surface differences are visible but the structural similarities are already assumed. Search across industries, eras, and scales for situations with the same underlying dynamics.
Before narrowing: Show the complete generated set to the user first. Use AskUserQuestion:
Then find 2-3 with the strongest structural match.
Step 3: Describe Each Precedent For each: what was the specific situation, what approaches were tried, what was the outcome? Be concrete — vague historical reference ("like the industrial revolution") is not a precedent. A precedent is a specific case with specific decisions and specific results.
Step 4: Structural Mapping For each precedent: where does the structural similarity hold most clearly? Where does it break down? What are the key variables that differ between the historical case and the current situation — and how might those differences change what the precedent implies?
Step 5: Extract the Lesson Not "do what they did" — that's surface imitation that ignores structural differences. The lesson is: what does their experience suggest matters most in this type of situation? What was the decisive variable? What would have changed the outcome if it had been different?
Before proceeding, use the AskUserQuestion tool. State your interpretation of the situation in 1–2 sentences — what is being analyzed and what the core question is — then ask:
Proceed based on their selection. If the user reframes, incorporate the correction before running any analysis.
Situation (Abstracted): [structural description in domain-neutral terms]
Precedents
| Precedent | What/When | What Was Tried | Outcome |
|---|---|---|---|
| [descriptive name] | [specific context] | [approach taken] | [what happened] |
Structural Mapping
| Precedent | Where It Maps | Where It Diverges | Key Differing Variables |
|---|---|---|---|
| [name] | [structural overlap] | [structural gap] | [variables that differ] |
Lesson: [the transferable principle — the underlying rule, stated without domain-specific language]
Caveats: [where the precedent fails to apply and what would need to be different for it to hold]
The strongest precedent is often the most surprising one — from a completely different domain that shares the underlying dynamics. Don't default to the obvious comparison within the same industry. The abstractions done in Step 1 determine the quality of everything that follows — spend time there.
After delivering this output, use AskUserQuestion to offer the next move:
/s4h-historical-lesson-extraction — Extract lessons from the precedents found/s4h-decision-criteria-weighting — Weight criteria against precedent outcomes/s4h-strategy-positioning — Position relative to what happened in comparable cases