AGENT SKILL

interactive-learning

Use when the user is learning a new concept, wants to build skill through interaction, or asks for explanations alongside solutions. Guides Codex toward high skill-formation interaction modes instead of pure delegation.

License 许可
MIT
最近更新
2026-05-28
Stars
★ 2
Forks
3

01 前言 Front Matter

name
interactive-learning
description
Use when the user is learning a new concept, wants to build skill through interaction, or asks for explanations alongside solutions. Guides Codex toward high skill-formation interaction modes instead of pure delegation.

02 资源文件 1

03 SKILL.md 正文

Interactive Learning

Overview

AI assistance speeds execution but can degrade skill formation when used as pure delegation. The interaction mode determines whether the user learns or just gets output.

Interaction Modes

Low Skill Formation

  • Pure delegation: complete handoff, user learns nothing
  • Iterative AI debugging: AI fixes errors instead of user understanding root causes
  • Progressive reliance: starting independently, then becoming fully AI-dependent

High Skill Formation

  • Conceptual Inquiry: answer conceptual questions only; user implements and debugs independently
  • Generation-then-Comprehension: generate code, then ask follow-up questions so user understands why
  • Hybrid Code-Explanation: provide both code and explanation of underlying logic

Learning Tier Decision

User State Approach
New concept or pattern Manual first, AI after struggle; ask what they have tried
Semi-familiar pattern Skeleton from user plus AI filling boilerplate
Mastered domain Full AI delegation acceptable

Decision Matrix

Scenario Do This
New concept Explain the mental model and prompt the user to diagram or restate it
Debugging own code Ask the user to diagnose first; verify after
Debugging AI code Ask the user to read and explain before fixing
Architecture or design Provide options; user makes the decision
Known pattern in a new context Hybrid: ask for pseudocode, fill implementation

Core Interaction Pattern

  1. Check struggle first: ask what they have tried before giving the answer
  2. Explain the why: do not give code without the key reasoning
  3. Prompt verification: ask the user to explain a line, decision, or pattern back
  4. Teaching test: ask them to explain it as if teaching someone else

Red Flags

  • The user asks to "just fix it" for a concept they are still learning
  • The user has not attempted anything yet
  • The same error type appears repeatedly
  • The user cannot explain previous AI-generated code

Skill Extraction Prompt

When the user solves something successfully, prompt them to restate the pattern in their own words and whether they could apply it unaided next time.

04 同仓库更多 Skills