Topic Analysis: Chapter 2 - What Good Context Looks Like

Metadata

• Syllabus Reference: Part 1, Chapter 2
• Primary Sources: Article 2 (What Good Context Looks Like), Article 7 (Teams)
• Secondary Sources: Interview Q3-4, Presentation
• Analysis Date: 2025-11-28
• Status: Complete

---

1. Source Materials

1.1 Primary Sources

From Article 2: "What Good Context Looks Like"

Perplexity workflow:

"My process looks like this: 1. Give Perplexity a short question with context (5-10 seconds), 2. Perplexity returns summary + links, 3. Take only relevant parts (not everything!), 4. Insert as context into the next tool"

Real task template example:

## User Report
User can't open the edit form for their record in the table even though they're an admin

## Support Verification
User is indeed admin of that record, we verified their profile, they should have access

## Context
Table the user mentions is from <table code file> specifically lines 100-200

## Goal
If user is administrator, they must have access to edit records from the table, not just from the detail page

## Possible Solutions
1. Check if the edit button for the record is functional
2. Check if it properly validates permissions
3. Create a test that would catch this bug before reaching users

Key insight on context size:

"If information about the problem can't be held in your head, I break it into smaller steps."

Iteration approach:

"Don't send a correction message ('sorry, I meant just the first form'). Edit the original prompt to be more specific. Continue working with new requirements."

Context contamination warning:

"Starting a new session/chat is often better - it avoids context contamination where old, wrong information misleads AI even after corrections."

Good context test:

"If you can explain an AI task so that you get at least 80% correct result on first try, you have good context."

From Article 7: "Context Engineering for Teams"

Team task template:

## Problem
[What broke / what needs to be done]

## Context
- Files: [specific files and lines]
- History: [relevant previous changes]
- Constraints: [what must not change]

## Goal
[Clear success criterion - when is it done]

## Possible Solutions
1. [First option]
2. [Second option]
3. [Third option]

## Tests
[How we verify it works]

1.2 Secondary Sources

From Interview Q3 (Context Transfer)

• Always write prompt yourself and note source ("this is from internet" or "found on Perplexity")
• Sometimes copy entire response with links
• Sometimes just relevant parts
• Edit to remove unnecessary sources

From Interview Q4 (Task Example) Real example of good task:

• User report from support
• Support verification
• Specific file and line numbers
• Clear goal
• Possible solutions to investigate

1.3 External Citations

None specific to this chapter - focus is on practical patterns.

---

2. Content Extraction

2.1 Key Concepts

1. Five Components of Good Context

   • Task: What you want AI to do
   • Constraints: What AI must not do
   • Background: Why you need this
   • Examples: What good output looks like
   • Success Criteria: How you'll judge the result
   • Source: Derived from Article 2, Article 7 templates

2. Research → Context → AI Workflow

   • Definition: Use Perplexity/research first, then feed into working tool
   • Example: 5-10 second question → summary + links → extract relevant → insert into Claude/ChatGPT
   • Source: Article 2, Interview Q1-3

3. The Task Template

   • Definition: Structured format for AI tasks
   • Components: Problem, Context, Goal, Solutions, Tests
   • Source: Article 2, Article 7

4. Context Contamination

   • Definition: Old wrong information misleading AI even after corrections
   • Solution: Start fresh session with clean context
   • Source: Article 2, Article 3

5. The 80% First Try Test

   • Definition: Good context = 80%+ correct on first try
   • If 2-3+ iterations needed, problem is context not AI
   • Source: Article 2

2.2 Key Examples

1. Bug Fix Task Example

   • Context: Admin user can't edit records
   • Before: "User can't edit profile, fix it"
   • After: Full template with user report, verification, file lines, goal, solutions
   • Source: Article 2, Interview Q4

2. Perplexity Workflow Example

   • Context: Research phase for any task
   • Process: Short question → Get summary with sources → Extract relevant → Feed to next tool
   • Source: Article 2

3. Zed + Claude Example

   • Context: AI finding files based on keywords
   • Works well: When specifying function/module name
   • Works poorly: Similar file names in different directories
   • Source: Article 2

4. Iteration Example - Form Types

   • Context: AI shows 2 implementations, change relates to one
   • Solution: Edit original prompt to specify which form, don't send correction message
   • Source: Article 2

2.3 Key Quotes

1. "Always write the reason - 'I need this for XYZ' helps AI understand context" - Article 2

   • Use for: Practical tip

2. "If you can explain an AI task so that you get at least 80% correct result on first try, you have good context. If you have to iterate more than 2-3 times, the problem isn't AI but your task description." - Article 2

   • Use for: Quality benchmark

3. "Less is sometimes more - rather precise context for small task than lots of information for big one" - Article 2

   • Use for: Counter-intuitive insight

4. "When something doesn't work, start fresh - sometimes it's simpler to open a new session than fix an old one" - Article 2

   • Use for: Practical decision point

2.4 Data/Statistics

• 90% Perplexity vs 10% Google for research (Interview Q12)
• 80% correct on first try = good context benchmark
• 5-10 seconds for initial Perplexity query
• 2-3 iterations max before reassessing context

---

3. Gap Analysis

3.1 Content Gaps

• [x] Task template provided
• [x] Research workflow explained
• [x] Bad vs good examples included
• [ ] Could expand on non-technical task templates

3.2 Clarity Issues

• None - concepts very practical

3.3 Depth Assessment

• Strong on technical examples
• Could use more non-technical templates
• Iteration concept well covered

---

4. Structure Proposal

4.1 Chapter Outline

Chapter 2: What Good Context Looks Like

Section 2.1: The Five Components of Context

• Main point: Every effective AI context includes five elements
• Content from: Article 2, Article 7
• Include: Task, Constraints, Background, Examples, Success Criteria

Section 2.2: The Universal Task Template

• Main point: Structured format that works for any task
• Content from: Article 2, Article 7, Interview Q4
• Include: Problem, Context, Goal, Solutions, Tests template

Section 2.3: Before and After - Real Examples

• Main point: See the difference good context makes
• Content from: Article 2, Article 8
• Include: Bug fix example, article writing example

Section 2.4: The Junior Developer Test

• Main point: If junior can't complete it, AI can't either
• Content from: Article 3, Interview Q24
• Include: "Could a new colleague complete this?" question

4.2 Opening Hook

"In the previous chapter, we saw why prompts fail. Now let's look at exactly what 'good context' looks like - the practical patterns that get results on the first try."

4.3 Key Takeaways

1. Good context has 5 components: Task, Constraints, Background, Examples, Success Criteria
2. The task template works universally: Problem → Context → Goal → Solutions → Tests
3. If you need more than 2-3 iterations, the problem is your context, not AI

4.4 Transition

"Now that you know what good context looks like, let's explore the most common mistakes people make - and how to avoid them."

---

5. Writing Notes

5.1 Tone/Voice

• Practical, pattern-focused
• Show don't tell with templates
• Emphasize repeatability

5.2 Audience Considerations

• Include team task template for developers
• Include simpler template for general audience
• Show how same principles apply across domains

5.3 Potential Visuals

1. Five Components Diagram

   • Visual breakdown of Task, Constraints, Background, Examples, Criteria

2. Task Template

   • Fillable template format

3. Before/After Comparison Table

   • Side-by-side bad vs good context

4. Research Workflow Diagram

   • Perplexity → Extract → Claude/ChatGPT flow

---

6. Prepared Citations

Internal

• [A2] Article "What Good Context Looks Like"
• [A7] Article "Context Engineering for Teams"
• [I3] Interview Q&A, Question 3 (Context transfer)
• [I4] Interview Q&A, Question 4 (Task example)
• [I24] Interview Q&A, Question 24 (Junior test)

External

• None specific to this chapter

---

7. Open Questions

1. Should team template go here or in Chapter 7 (Teams)?

   • Decision: Introduce here, expand in Chapter 7

2. How detailed should the task template be?

   • Decision: Full template with all components

3. Include Zed-specific details or keep tool-agnostic?

   • Decision: Keep principles tool-agnostic, specific tools in Chapter 5