ChatGPT · Module 02 of 8

Coding

Generate, debug, and ship code with ChatGPT — from snippets to full features.

3 lessons12 exercises
Topics & techniques

Code generation

Turn requirements into working code.

4 cards →
Best practice
Ask for Tests Alongside the Code
A habit of requesting unit tests in the same turn as the implementation — including edge cases — so you can verify the generated code instead of eyeballing it. When a test does fail, hand the failure back for a focused [debugging](/ai-hub/chatgpt/coding/debugging/) pass.
L2 · Context engineering
Knowledge card
Generate a Function From a Clear Signature and Test
A generation move: hand ChatGPT the function signature and a couple of input/output examples, and let it write the body — instead of describing the behaviour in vague prose.
L2 · Context engineering
Mistake
Generate in Small Units, Not a Whole App at Once
A scoping rule: ask ChatGPT for one testable unit at a time — a function, a component, a module — and verify it before moving on, rather than requesting an entire app in a single prompt.
L2 · Context engineering
Knowledge card
Give ChatGPT Your Stack and Conventions Up Front
A prompting habit: before asking for code, state your language, framework and versions, plus the conventions the code must follow — so what comes back fits your project instead of a generic template.
L2 · Context engineering

Debugging

Diagnose and fix errors instead of guessing.

4 cards →
Concept
Ask for the Cause Before the Fix
A sequencing habit: ask ChatGPT to explain why the bug happens before it proposes a fix, and only accept the fix once the stated cause makes sense.
L2 · Context engineering
Best practice
Give a Minimal Reproduction, Not the Whole File
A debugging discipline: reduce the problem to the smallest snippet that still reproduces the bug — plus the inputs, expected output, and actual output — before asking ChatGPT to diagnose it.
L2 · Context engineering
Mistake
Debug by Pasting the Error, Not the Vibe
A debugging discipline: give ChatGPT the exact error message, the stack trace, and the smallest snippet that reproduces it — rather than describing the symptom in your own words.
L2 · Context engineering
Mistake
When It Loops on a Wrong Fix, Reset the Context
A recovery move: when ChatGPT cycles through variations of the same wrong fix, stop feeding the failing thread and open a fresh chat with a clean, well-framed statement of the problem.
L2 · Context engineering

Data analysis

Use the code interpreter to analyse files and data.

4 cards →
Workflow
Ask for the Cleaning Steps as a Reusable Script
A workflow move: once you've cleaned and analysed a dataset interactively, ask ChatGPT to consolidate every step into a single reusable script you can run on next month's file without redoing the conversation.
L3 · Workflows
Knowledge card
Describe Your Columns Before Asking for Analysis
A setup step: give ChatGPT a short data dictionary — what each column means, its units, and any quirks — before asking your analysis question, so it doesn't guess wrong about your data.
L2 · Context engineering
Best practice
Have ChatGPT Show the Code, Not Just the Chart
A verification habit for the code-interpreter tool: always ask to see the Python it ran, not just the resulting chart or number, so you can check how the result was actually produced.
L2 · Context engineering
Knowledge card
Upload the File, Then Ask One Question at a Time
A pattern for ChatGPT's data analysis (code interpreter): upload the raw file, ask it to first describe the data, then pose one analytical question per turn.
L2 · Context engineering
ChatGPT for coding

Treat ChatGPT as a pair, not an oracle

ChatGPT writes competent code across most languages, but the developers who get real value treat it like a fast pair-programmer that needs a precise brief, not a vending machine. Give it the signature, the types, the surrounding code, and the constraint — 'this runs in a Lambda with a 3-second timeout' — and it produces code that fits your system. Ask it vaguely and you get plausible code that ignores your architecture. For anything with real logic, switch to a reasoning model so it thinks the problem through before writing.

Debugging: paste the error, not the vibe

The single biggest upgrade to debugging with ChatGPT is boring: paste the exact error message and the smallest snippet that reproduces it, rather than describing the symptom in prose. 'It doesn't work' forces guessing; a stack trace plus the failing function gives ChatGPT something concrete to reason from. For data work, its code interpreter can actually run Python on files you upload — so it can analyse, calculate, and chart rather than just suggest.

The coding topics worth drilling into

This section breaks down into code generation (turning requirements into working code), debugging (diagnosing real errors instead of guessing), and data analysis (using the code interpreter on your own files). Each card below is a specific, tested habit — start with whichever matches the problem in front of you right now.

Frequently asked
Can ChatGPT write code?

Yes, across most popular languages. It is strongest when you give it the types, the surrounding code, and the constraints rather than a one-line description. For logic-heavy tasks, use a reasoning model so it works through the problem before generating.

Can ChatGPT actually run code?

Yes — its data-analysis tool (code interpreter) executes Python in a sandbox, so it can process files you upload, run calculations, and produce charts. For pure code generation it writes the code for you to run in your own environment.

How do I get better bug fixes from ChatGPT?

Paste the exact error message and the smallest snippet that reproduces the bug instead of describing it. Concrete inputs let ChatGPT reason about the real cause rather than guessing at a plausible one. The debugging cards below show the full pattern.

Which ChatGPT model is best for coding?

Use the fast default for quick edits and boilerplate, and switch to a reasoning model for algorithms, tricky logic, or debugging where correctness matters more than speed. Matching the model to the task is especially valuable in coding.

Test yourself

How well do you code with ChatGPT?

Six questions on generating code that fits your project, debugging without guessing, and analysing data you can actually trust. No sign-up, instant score, and you can share how you did.

  1. Q1 Before asking ChatGPT for code, what should you state up front?
  2. Q2 What's the most reliable way to specify a function you want written?
  3. Q3 You're building a feature. What scope should each prompt have?
  4. Q4 You've hit a bug. What do you give ChatGPT?
  5. Q5 ChatGPT keeps cycling through variations of the same wrong fix. What now?
  6. Q6 The code interpreter returns a clean chart. What should you always ask for too?
0 / 6 answered