Quick Reference

How to Use This Guide

This isn't a guide you read top to bottom. It's a reference you come back to. Bookmark it. When you're about to start a new project, debug an issue, or review AI code — scroll to the relevant section and grab what you need.

Prompt Templates

Starting a New Project — Discovery

I have an idea for a software application. Before we write 
any code, act as a Senior IT Business Analyst and Systems 
Architect.

Interview me to extract:
- Business requirements and user stories
- User journeys and workflows
- Technical constraints and preferences
- Data structures and relationships
- Integration requirements
- Edge cases and error scenarios

Rules:
1. DO NOT write any code
2. Ask 1-3 questions at a time
3. Wait for my answers before continuing
4. Research what I don't know and recommend solutions
5. Keep going until you have complete understanding

Say "I am ready. Let's begin." if you understand.

Starting a Coding Session

Act as a Senior Full-Stack Developer. Here are the project 
documents:

[attach BRD, Tech Spec, Data Schema]

Rules:
1. Do NOT write the whole app at once
2. Use EXACTLY the field names in the Data Schema
3. Follow the architecture in the Tech Spec
4. Build one feature at a time, wait for my approval
5. Include error handling in every function
6. Never hardcode secrets — use environment variables

Review the docs and say "Read and Understood."

Debugging — Reporting an Issue

I'm getting an error. Here's the context:

**What I expected:** [describe expected behavior]
**What actually happened:** [describe actual behavior]
**Error message:** 
[paste the full error message]

**Relevant code:**
[paste the file or function where the error occurs]

**What I've already tried:**
[list any fixes you attempted]

Please diagnose the issue and provide a fix. Explain what 
caused it so I understand.

Code Review Request

Review this code for:
1. Bugs or logical errors
2. Security vulnerabilities (especially hardcoded secrets)
3. Missing error handling
4. Performance concerns
5. Code that doesn't match the Data Schema

Here's the code:
[paste code]

For each issue found, explain WHY it's a problem and 
provide the corrected version.

Asking for an Explanation

Explain this code to me in plain English:

[paste code]

For each section, tell me:
1. What it does
2. Why it's there
3. What would break if we removed it

Use simple language — I'm managing this project, not writing the code.

Red Flags Checklist

Quick scan for AI-generated code. If you see any of these, stop and address them:

🚩 Immediate Stops

• Hardcoded secrets — API keys, passwords, tokens in the code
• TODO or FIXME comments — Unfinished work the AI skipped
• No error handling — No try/catch, no checking for errors
• console.log with sensitive data — Logging passwords or tokens

⚠️ Yellow Flags

• Generic variable names — data, temp, x, result
• Functions longer than 50 lines — Probably doing too much
• No comments — The AI should explain its code
• Installing many new dependencies — Ask if fewer would work
• "This should work" — Even the AI isn't confident
• Outdated or deprecated libraries — AI loves to suggest packages that were popular in 2020 but are dead today (e.g., request instead of fetch)
• Unfamiliar package names — If you've never heard of it, search the registry before installing. AI hallucinates packages that don't exist.

✅ Good Signs

• Descriptive function names — calculateShippingCost, not calc
• Error handling present — try/catch blocks, error messages
• Environment variables for secrets — process.env.API_KEY
• Comments explaining why, not what — Good practice
• Small, focused functions — Each does one thing

Git Quick Reference

Golden Rule: Commit after every change that works. Each commit is a save point you can return to.

Common Error Translations

Decision Quick-Reference

Should I start a new chat?

Is the conversation longer than ~50 messages?
├── Yes → Start fresh, re-attach your docs
└── No
    ├── Is the AI contradicting earlier responses?
    │   ├── Yes → Start fresh, re-attach your docs
    │   └── No → Stay in this chat
    └── Did you switch to a completely different feature?
        ├── Yes → Consider a new chat with relevant docs
        └── No → Stay in this chat

Should I start over? (The 50% heuristic)

Time to fix current code > 50% of time to rebuild with docs?
├── Yes → Start over with updated requirements docs
└── No → Keep fixing

Is this AI's job or mine?

Writing boilerplate code        → AI
Deciding what to build          → You
Generating documentation        → AI
Reviewing generated code        → You
Choosing the tech stack         → You (AI can recommend)
Testing if it works             → You
Explaining a concept            → AI
Deciding if the explanation     → You
  is accurate enough

💡 This is a living reference. As you develop your own workflow, add your own templates and checklists to this list. The best reference is the one customized to how you work.