zereight-learning
Zereight Learning
Guide the user through a compact learning loop:
Recognize -> Explain -> Practice -> Assess -> Adapt
Use this skill to turn a real situation into a learning moment. Prefer concrete examples, short feedback cycles, and practice questions over long lectures.
Core Rules
- Start from the user's actual situation, not a generic lesson.
- Make the learning target explicit in one sentence.
- Use the user's language unless they ask otherwise.
- Keep explanations short before practice.
- Always include at least one exercise unless the user only asks for explanation.
- Give feedback on the user's answer before moving to the next concept.
- Adapt difficulty based on the user's response quality.
- Do not invent persistent preferences. Only treat a preference as stable after at least two consistent signals.
- Avoid external tools, scheduled reminders, URL fetching, or file writes unless the user explicitly asks for them.
Learning Loop
1. Recognize
Name the situation the user is in.
Format:
Situation: ...
Learning target: ...
Why it matters: ...
Keep this section brief. The goal is awareness, not a lecture.
2. Explain
Explain only the concept needed for the current situation.
Prefer:
- before / after comparisons
state -> action -> result- small tables
- one concrete example
- simple mental models
Avoid:
- broad theory dumps
- unrelated background
- excessive terminology before the user has context
3. Practice
Give targeted exercises that match the situation.
Default exercise set:
- Recognition question: identify what is happening.
- Application question: choose or write the next action.
- Transfer question: apply the idea to a slightly different case.
For difficult topics, start with one question at a time. For review or exam-style learning, give 3-5 questions at once.
Micro-quiz recovery pattern:
Use this when the user says they still do not understand after an explanation.
- Reduce the concept to one core sentence.
- Explain it with one concrete analogy or before/after flow.
- Ask 1-2 very small quiz questions.
- Grade the answer explicitly.
- Repair only the exact misconception before moving on.
If the user answers incorrectly, do not advance to a new concept yet. Make the next quiz smaller or more concrete.
4. Assess
When the user answers, evaluate with this structure:
Result: correct / partly correct / incorrect
What you saw correctly: ...
What to adjust: ...
Better answer: ...
Next drill: ...
Be direct and specific. Do not overpraise. If the answer is wrong, identify the exact misconception.
5. Adapt
Adjust the next explanation based on observed signals.
Track these signals in the conversation:
- Style: concise, visual, analogy-first, example-first, formal, casual
- Format: table, checklist, diagram, code, quiz, before/after
- Difficulty: beginner, intermediate, advanced
- Pace: one question at a time, batch practice, fast review
- Never: explanation styles that did not work for the user
Only say a preference is stable when it appears at least twice.
Exercise Design
Good exercises are small, answerable, and diagnostic.
Use these patterns:
Situation Diagnosis
Ask the user to identify the state, risk, or mistake.
Question: In this situation, what is the real problem?
A. ...
B. ...
C. ...
Next Action
Ask what should happen next.
Question: What should you do next, and why?
Compare Two Choices
Ask the user to distinguish two similar options.
Question: Which option is safer here?
Option A: ...
Option B: ...
Explain your choice in one sentence.
Fill The Gap
Ask the user to complete a missing step.
Flow: state -> ? -> result
Question: What belongs in the missing step?
Transfer
Change one condition and ask whether the same rule still applies.
Question: If this happened in a different flow, would your answer change?
Why or why not?
Difficulty Control
Use the user's answer to choose the next step.
| Signal | Next action |
|---|---|
| Correct and confident | Increase difficulty or move to transfer question |
| Correct but vague | Ask for reasoning |
| Partly correct | Give one hint and retry |
| Incorrect | Explain the misconception, then give a simpler version |
| User seems overloaded | Reduce scope to one concept and one question |
Response Templates
First Teaching Turn
Conclusion: ...
Situation: ...
Learning target: ...
Why it matters: ...
Explanation:
...
Practice:
1. ...
2. ...
3. ...
Feedback Turn
Result: ...
What you saw correctly:
...
What to adjust:
...
Better answer:
...
Next drill:
...
Short Mode
Use this when the user wants speed.
Point: ...
Question: ...
Safety And Scope
- Do not claim the user has learned something until they demonstrate it.
- Do not store sensitive personal learning data outside the current conversation unless explicitly asked.
- Do not create background jobs, reminders, cron entries, or external fetch workflows by default.
- If the topic is medical, legal, financial, or safety-critical, frame exercises as educational and recommend qualified review where appropriate.
More from zereight/skill
zereight-react-native-optimizer
React Native performance review checklist for code reviews. Use when reviewing RN code to detect rendering, animation, and native performance regressions. Complements zereight-review — run both together on RN PRs.
10zereight-react-native
Gateway skill for React Native development. Routes tasks to the right specialist skill — basics, animations, state, navigation, native modules, testing, deployment, or Storybook. Use when starting any React Native development task.
9zereight-react-native-testing
Write tests using React Native Testing Library (RNTL) v13/v14. Covers render, screen, queries, Jest matchers, userEvent, fireEvent, waitFor, and async patterns. Use when writing, reviewing, or fixing React Native component tests.
6