AgentSkillsCN

learning-logic_consistency

对学习资料进行逻辑梳理，确保概念连贯性，并采用“先问为什么”的讲解方式，帮助用户更轻松地理解知识，避免“神秘跳跃”的疑惑。

SKILL.md
--- frontmatter
name: learning-logic_consistency
description: Review learning materials for logical flow, conceptual continuity, and "Why-First" explanations. Use to ensure tutorials are beginner-accessible and free of "magic" leaps.

Learning Logic Consistency

Objectives

Ensure that learning materials (Notes, Tutorials, Notebooks) follow a linear, logical path where every concept is justified, explained before used, and flows naturally.

Review Criteria

1. The "Zero Leap" Rule

  • •Check: Does the content introduce a technical term, acronym, or code library without a prior plain-English explanation?
  • •Correction: If "SVM" appears, ensure "Support Vector Machine" and the concept of "finding the widest path" were explained in the preceding paragraph.

2. The "IO & Parameter" Transparency Rule

  • •Check (Input): Is it clear what data is being "fed" into the model? (e.g., "We are giving the model the X coordinates and the Y labels").
  • •Check (Output): Is the result of the code explained? What does the return value actually mean in the physical world?
  • •Check (Parameters): Is every "magic" parameter (like C, gamma, random_state) explained?
    • •Bad: SVC(C=1.0)
    • •Good: "We set C=1.0. Think of C as a 'strictness' dial. 1.0 is the default balance."
  • •Correction: Add a "Parameter Table" or explicit "Input/Output" description for every major code block.

3. The "Why-First" Principle

  • •Check: Does the tutorial say "Do X" without explaining why we are doing X?
  • •Check (Code): Is there a "Magic Number" or a parameter (like C=1.0) that isn't explained in terms of its impact on the result?
  • •Correction: Prepend every action with its motivation. "Because real-world data is messy, we need to allow some errors. We do this by adjusting the 'C' parameter..."

3. Conceptual Dependency Chain

  • •Check: Does Step B depend on a concept that was only introduced in Step C?
  • •Correction: Reorder content so that the "Foundation" always precedes the "Application".

4. Code-Theory Synchronization

  • •Check: Does the code use variable names or logic that contradict the terminology used in the text?
  • •Correction: Align names. If the text calls it "Safety Gap", the code comment shouldn't just call it "margin" without linking the two.

5. Transition Verification

  • •Check: Is there a "Cliffhanger" between sections where the reader might get lost? (e.g., jumping from simple lines to complex 3D math).
  • •Correction: Add "Bridge sentences". (e.g., "Now that we know how to handle straight lines, what happens when the data forms a circle? This is where the 'Kernel' comes in.")

Implementation Workflow (Self-Review)

Before finalizing any learning package, perform this 3-point check:

  1. •The "Newbie Test": If I didn't know the title of this page, would I understand the second paragraph?
  2. •The "Why" Audit: Search for every code block. Is the paragraph above it explaining the problem the code is solving?
  3. •The Acronym Scan: Search for all-caps words. Are they all defined at their first occurrence?

Language Tone

  • •Encouraging, pedagogical, and clear.
  • •Use analogies (e.g., "修路", "魔法转换", "惩罚旋钮") to ground abstract math.