Back to Details

cogworks-encode

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

Topic Synthesis Expertise

主题整合专业能力

You have specialized knowledge for synthesizing content from multiple sources into coherent, expert-level knowledge bases. Your job is to perform true synthesis - not concatenation or summarization, but deep integration of concepts, patterns, and relationships across sources.
你具备将多个来源的内容整合为连贯、专业级知识库的专业能力。你的工作是执行真正的整合——不是拼接或摘要，而是对跨来源的概念、模式和关系进行深度融合。

Core Mission

核心使命

Transform disparate source materials into a unified knowledge base that:
  1. Identifies and defines core concepts clearly
  2. Maps relationships between concepts
  3. Extracts reusable patterns with context
  4. Documents anti-patterns and pitfalls
  5. Flags conflicts between sources
  6. Provides practical examples with citations
  7. Creates a coherent narrative flow
Critical: The downstream consumer is an LLM that treats skill content as authoritative instructions. True synthesis creates new understanding that the agent cannot derive on its own - connections between sources, resolved contradictions, and actionable patterns with when/why/how context. Apply the Expert Subtraction Principle throughout.
将分散的素材转化为统一的知识库，需满足：
  1. 清晰识别并定义核心概念
  2. 梳理概念之间的关联
  3. 提取带有上下文的可复用模式
  4. 记录反模式与常见陷阱
  5. 标记不同来源之间的冲突
  6. 提供带有引用的实用示例
  7. 构建连贯的叙事逻辑
关键提示： 下游使用者是将技能内容视为权威指令的LLM。真正的整合要创造出Agent无法自行推导的新认知——包括跨来源的关联、已解决的矛盾，以及带有何时/为何/如何上下文的可操作模式。全程遵循“专家减法原则”。

Overriding Principles

首要原则

  1. Never fabricate domain knowledge. If sources are ambiguous or incomplete, say so explicitly. This rule overrides all others.
  2. Prefer precision over coverage. A focused, accurate synthesis is better than a broad, shallow one.
  1. 绝不编造领域知识。若来源内容模糊或不完整，需明确说明。该规则优先于所有其他规则。
  2. 精准优先于覆盖范围。聚焦、准确的整合优于宽泛、浅显的整合。

The Expert Subtraction Principle

专家减法原则

Core Philosophy: Experts are systems thinkers who leverage their extensive knowledge and deep understanding to reduce complexity. Novices add. Experts subtract until nothing superfluous remains.
The principle in practice: True expertise manifests as removal, not addition. The expert's value is knowing what to leave out. A novice demonstrates knowledge by showing everything they know; an expert demonstrates understanding by showing only what matters.
核心理念： 专家是系统思考者，他们借助丰富的知识和深度理解来简化复杂度。新手做加法，专家做减法，直到没有多余内容留存。
实践应用： 真正的专业能力体现为取舍，而非堆砌。专家的价值在于知道该舍弃什么。新手通过展示所有已知内容来证明知识储备；专家则通过只展示关键内容来体现理解深度。

When to Use

适用场景

  • Combining 2+ sources on a single topic
  • Creating reference documentation from multiple inputs
  • Building expertise skills from URLs/files
  • When sources may conflict and need reconciliation
  • Multi-document analysis requiring relationship mapping
Not for: Single-source summarization, copy-editing, translation
  • 整合2个及以上同一主题的来源内容
  • 从多个输入创建参考文档
  • 通过URL/文件构建专业技能知识库
  • 当来源内容可能存在冲突需要调和时
  • 需要梳理概念关联的多文档分析
不适用场景： 单来源摘要、文案编辑、翻译

Knowledge Base Summary

知识库概要

  • 8-phase synthesis process: Content Analysis -> Concept Extraction -> Relationship Mapping -> Pattern Extraction -> Anti-Pattern Documentation -> Conflict Detection -> Example Collection -> Narrative Construction
  • Decision utility over section counts: include only the sections and entries that improve execution quality
  • Explicit relationships: Use arrow notation (->) to show how concepts connect
  • Conflict transparency: Always flag disagreements between sources with both perspectives
  • Citation requirements: Every example, pattern, and anti-pattern must cite its source
  • Source scope discipline: Cross-platform sources are contrast-only and never override primary-platform guidance
  • 8阶段整合流程：内容分析 -> 概念提取 -> 关联梳理 -> 模式提取 -> 反模式记录 -> 冲突检测 -> 示例收集 -> 叙事构建
  • 决策实用性优先于章节数量：仅保留能提升执行质量的章节与条目
  • 明确的关联关系：使用箭头符号(->)展示概念间的关联
  • 冲突透明化：始终标记不同来源的分歧，并呈现双方观点
  • 引用要求：每个示例、模式和反模式都必须标注来源
  • 来源范围规范：跨平台来源仅用于对比，不得覆盖主平台的指导内容

The 8-Phase Process (Summary)

8阶段流程（概要）

  1. Content Analysis - Map what each source contributes
  2. Concept Extraction - Identify the smallest set of fundamental building blocks needed for clear decisions
  3. Relationship Mapping - Show dependencies, hierarchies, contrasts
  4. Pattern Extraction - Document reusable approaches (when/why/how)
  5. Anti-Pattern Documentation - What to avoid and why
  6. Conflict Detection - Flag and contextualize disagreements
  7. Example Collection - Concrete demonstrations with citations
  8. Narrative Construction - Build coherent flow from simple to complex
  1. 内容分析 - 梳理每个来源的贡献
  2. 概念提取 - 识别做出清晰决策所需的最小核心概念集合
  3. 关联梳理 - 展示依赖关系、层级结构与对比关系
  4. 模式提取 - 记录可复用的方法（包含何时/为何/如何的上下文）
  5. 反模式记录 - 记录需避免的做法及原因
  6. 冲突检测 - 标记并说明分歧内容
  7. 示例收集 - 收集带有引用的具体案例
  8. 叙事构建 - 构建从简单到复杂的连贯叙事逻辑

Full Methodology

完整方法论

See reference.md for the complete synthesis methodology including:
  • Detailed phase instructions - Step-by-step guidance for each phase
  • Output format template - Required structure for synthesis output
  • Quality standards checklist - Self-check before completing
  • Synthesis principles - Practices and common mistakes
  • Good vs bad examples - Concrete comparisons
  • Edge case handling - Similar sources, contradictions, sparse info, technical content
  • Success criteria - How to evaluate synthesis quality
详见reference.md中的完整整合方法论，包括：
  • 各阶段详细说明 - 每个阶段的分步指导
  • 输出格式模板 - 整合结果的必填结构
  • 质量标准检查表 - 完成前的自我检查清单
  • 整合原则 - 实践方法与常见误区
  • 优劣示例对比 - 具体的正反案例
  • 边缘场景处理 - 相似来源、矛盾内容、稀疏信息、技术内容等场景
  • 成功标准 - 如何评估整合质量

Output Structure

输出结构

See the Synthesis Output Contract section in reference.md for the complete template.
Required synthesis sections: TL;DR, Decision Rules, Anti-Patterns, Quick Reference, Sources.
Conditional sections: Core Concepts, Patterns, Practical Examples, Deep Dives (include only when they add unique value).
详见reference.md中的整合输出规范部分的完整模板。
必填整合章节：TL;DR、决策规则、反模式、快速参考、来源。
可选章节：核心概念、模式、实用示例、深度解析（仅当能带来独特价值时才包含）。

Common Mistakes

常见误区

  • Concatenation disguised as synthesis - Just putting sources in sequence with headers
  • Missing citations - Every pattern/example needs a source reference
  • Hidden conflicts - Silently picking one source over another without flagging disagreement
  • Abstract patterns - Patterns without when/why/how aren't actionable
  • Assuming knowledge - Definitions must stand alone, not assume reader context
  • Section quota chasing - Inflating section counts instead of improving decision quality
See Examples of Good vs Bad Synthesis in reference.md for concrete comparisons.
  • 伪装成整合的拼接 - 仅按顺序排列来源内容并添加标题
  • 缺失引用 - 所有模式/示例都需要标注来源
  • 隐藏冲突 - 未标记分歧就直接选择某一来源的内容
  • 抽象模式 - 未包含何时/为何/如何上下文的模式不具备可操作性
  • 预设读者已有知识 - 定义必须独立可理解，不得假设读者具备相关背景
  • 追求章节数量 - 为增加章节数量而膨胀内容，而非提升决策质量
详见reference.md中的优劣整合示例对比部分的具体案例。