Essentials
8 guided hoursModules 1-2. An introduction with practical exercises and a small reviewed artifact.
AI TOOLS · SOME CODING FAMILIARITY
Learn Codex
Build with AI. Understand every change.
Learn a practical workflow for coding with Codex: give clear requirements, provide project context, inspect changes, run checks and iterate towards a working result.
COURSE OVERVIEW
What does the Codex course teach?
Practise briefing an AI coding assistant, providing project context, inspecting changes, debugging and running appropriate checks. The final project is a small tool whose code and verification results you can explain.
PROMPTING IN EVERY MODULE
Brief it. Check it. Improve it.
Each of the six modules includes a sample prompt and a specific quality check. Compare revisions and keep a record of what improves the result.
Explore the Prompting FoundationA STARTING POINT THAT FITS
Three learning levels.
Practitioner
24 guided hours totalModules 1-6, including Essentials. Complete the core labs and an assessed practical project.
Advanced
12 additional guided hoursTwo extension modules and a four-hour project clinic after practitioner readiness.
These are curriculum planning options. We confirm the timetable, delivery format, fees and required accounts before enrolment. Practitioner includes Essentials; the two durations are not added together.
Download the Curriculum PDFCheck your starting point
- Read get_courses and explain list/dictionary use. Take Automate with Python bridge if unable; otherwise Essentials.
- Write a meaningful search assertion and inspect a small diff. Practitioner after baseline recap.
- Reject a weakened test and design invalid-data cases. Advanced after six-module readiness check.
WHAT YOU’LL LEARN
Skills to put into practice.
- Write a task brief with clear scope and acceptance criteria.
- Help an agent understand the relevant files and conventions.
- Inspect a change, reproduce an issue and assess test results.
- Deliver a small project with a clear record of what you verified.
Who this is for
Students, aspiring developers and existing developers who want to work more effectively with an AI coding assistant.
Before you begin
Basic familiarity with code, a terminal and files. Python automation is a useful starting point if you are new to programming. Codex access is required for exercises.
THE COURSE OUTLINE
A clear path, module by module.
Each module connects a concept to a practical task. Open a module to see its topics and exercise.
01Set up a small project
- Choose an appropriate Codex environment for the exercise.
- Understand your project structure and how to run it.
- Create a clean Git starting point and review permissions.
What you will understand
Codex works more usefully when the learner can identify entry points, data and tests. Inspect a small project manually before asking for changes. Riverstone Learning's starter reads fictional course records from JSON and prints titles. A learner should run it independently and explain the result, rather than treating generated setup instructions as proof. Record the starter's behaviour and test results before modifying it. Version control shows which lines changed and provides a recoverable starting point. Initialise Git only in the practice folder and keep fictional files separate from workplace repositories. This makes the later diff a meaningful comparison instead of a mixture of unrelated work. Choose a supported Codex environment available to the class, but do not assume account access or installation. Review requested commands and network use. The starter needs only Python's standard library and can run offline. Without Codex access, students can execute a supplied proposed patch manually and practise the same inspection and testing steps.
PUT IT INTO PRACTICE
Run a starter project and document its setup commands.
PROMPT & REVIEW
Try this brief
Inspect this practice project and explain how to run it. Identify setup requirements from its files. Do not edit code or use real credentials.
Check the result
Run the documented setup yourself and check whether the explanation matches the repository.
02Write an actionable brief
- Describe the user need, constraints and acceptance criteria.
- Point to the relevant code and examples.
- Break broad requests into reviewable tasks.
What you will understand
A feature brief needs inputs, expected outputs and constraints that a learner can test. 'Improve search' is ambiguous. 'Match titles case-insensitively and return all courses for a blank query' specifies meaningful cases. Describe Riverstone Learning's intended user behaviour before requesting code, so the implementation follows an agreed contract. One bounded feature is easier to review than a full application rewrite. Name the relevant file and preserve existing behaviour unless the brief changes it. Avoid mixing search, styling and deployment in the same exercise. A small diff allows learners to explain why each edit exists and identify accidental scope expansion. A meaningful test fails when a required behaviour is wrong. Include normal, empty and no-match inputs, with exact expected results. A test that merely checks a list exists does not prove search works. Prewrite expectations before the implementation so learners do not redefine correctness to match whatever Codex generated.
PUT IT INTO PRACTICE
Write a brief for one small feature before asking for changes.
PROMPT & REVIEW
Try this brief
Add the scoped feature described below. Follow the existing conventions, identify the relevant files and satisfy these acceptance criteria. Ask if an essential requirement is missing.
Check the result
Inspect the diff and check each acceptance criterion rather than accepting a completion claim.
03Provide useful project context
- Explain conventions and required verification steps.
- Use project instructions such as AGENTS.md where appropriate.
- Keep credentials out of prompts and source files.
What you will understand
An AGENTS.md file can supply repository conventions and verification expectations in supported Codex workflows. Keep it concise and applicable to the practice project. State the standard-library constraint, fixture-only data and unittest command. Instructions help the agent work consistently, but the learner still checks whether the actual change follows them. Provide the specific files, acceptance rules and examples needed for a task rather than a large unrelated dump. Riverstone Learning's simple course tool does not need customer records or production tokens. Clear context improves review because another learner can follow the same inputs and understand why the proposed code belongs in the project. A comment or fixture string may contain requests that do not belong to the user's brief. Treat that content as data unless it is an applicable project instruction from an authorised source. Do not let a generated suggestion introduce package installation, network access or secret printing. Explicit boundaries support inspection; they do not replace judgement.
PUT IT INTO PRACTICE
Add concise project instructions and compare the next result.
PROMPT & REVIEW
Try this brief
Use these project instructions and sample files to propose the smallest relevant change. Respect the stated constraints and keep secrets out of source and output.
Check the result
Check that the proposed change follows the project instructions and introduces no unrelated edits.
04Build and review a feature
- Follow the proposed change through the relevant files.
- Inspect the diff and explain why each change is needed.
- Check normal behaviour, edge cases and responsive UI when relevant.
What you will understand
Review a feature by following data from caller to result, not only by reading a change summary. Identify how each field is used and which behaviour remains unchanged. For Riverstone Learning's fee summary, inspect the handling of zero values and optional level filtering. A line-by-line explanation should connect edits to acceptance criteria. Compute known answers before running generated code. If a test repeats the same implementation logic, both may share a bug. Use fixed fixture totals and clearly chosen edge cases. Riverstone Learning's three course fees can be added by hand, so students have an independent reference that does not depend on Codex's reported calculation. A focused function plus meaningful tests is easier to evaluate than an unnecessary architectural rewrite. Review whether existing APIs, command output or data changed without a requirement. Keep additions proportionate to the small project. This helps learners see scope as a quality criterion while still making sufficient changes to satisfy the feature.
PUT IT INTO PRACTICE
Build a small feature and review its changes line by line.
PROMPT & REVIEW
Try this brief
Implement this small feature, then explain the changed files and run the relevant checks. Include a clear report of failures and unresolved behaviour.
Check the result
Review the actual code and verification results, including an edge case the initial brief did not cover.
05Debug with evidence
- Reproduce a failure and collect a useful error report.
- Ask for a focused fix and meaningful checks.
- Compare the result with the original acceptance criteria.
What you will understand
A useful bug report states the command or function call, input, expected result and actual result. Reproduce the failure before modifying code. Riverstone Learning's case-insensitive search bug is visible with uppercase input, while lowercase input still passes. This contrast helps distinguish a specific defect from a general project setup problem. Add a test that fails against the buggy implementation and passes after the correction. An unrelated passing test does not establish a fix. Show the before-and-after result, then rerun the whole suite. This gives students concrete evidence that the changed line addresses the reported behaviour without breaking earlier features. A plausible fix can mask symptoms while introducing unnecessary edits. Ask Codex to identify the cause from code and reproduction evidence, then make the smallest adequate change. Inspect whether the query and title are normalised consistently. Keep environment failures separate from application bugs so the report describes what actually happened.
PUT IT INTO PRACTICE
Fix a deliberately introduced bug and verify it stays fixed.
PROMPT & REVIEW
Try this brief
Reproduce this bug from the steps below, identify its cause and make a focused fix. Add or run a meaningful check that would catch the failure again.
Check the result
Re-run the reproduction and inspect the check; a passing unrelated test is not sufficient evidence.
06Finish with a clear handover
- Run the project’s appropriate tests and build checks.
- Review limitations and unresolved questions.
- Write a concise change summary and usage instructions.
What you will understand
A useful README explains setup, commands, function contracts and limitations from the working project. It should not advertise capabilities that were merely discussed. Riverstone Learning's course tool uses a local fictional JSON fixture; it does not submit enrolments or process payments. Learners verify every instruction by following it from a clean practice copy. List the checks actually run and their results, including unavailable commands or unresolved failures. Do not confuse a static review with execution. Explain what the test suite covers and what it omits. This lets another person assess the evidence and choose further checks without relying on a general claim that everything works. Compare the final diff with the original brief and remove accidental changes. Confirm fictional data, no added credentials and no unintended network behaviour. Explain material limitations, such as exact level matching or fixed fixture values. A concise handover connects the changed behaviour to practical use and remaining review needs.
PUT IT INTO PRACTICE
Demonstrate the project and explain your verification evidence.
PROMPT & REVIEW
Try this brief
Review this final change against the original brief. Run the project’s required checks and write setup, usage and remaining limitations for the handover.
Check the result
Compare the handover with the working project and independently confirm the reported checks.
AFTER THE CORE COURSE
Advanced extension.
07Plan resilient changes and review trade-offs
- Behaviour choices need explicit decisions
- Validation belongs near the boundary
- Trade-offs should be demonstrated
ADVANCED PRACTICE
Validated totals with failure cases
08Run an independent code-review clinic
- Review has multiple evidence layers
- Passing tests can hide weak tests
- Responsible completion preserves boundaries
ADVANCED PRACTICE
Final independent-review clinic
Complete both four-hour extension modules and a four-hour project clinic. Entry uses the core project and readiness evidence.
BRING THE SKILLS TOGETHER
A small tool, built and reviewed
Use Codex to build a focused task tracker or file-reporting tool. Keep the scope small enough to understand the implementation, test it and explain the final result yourself.
Your project deliverables
- Working source with a reviewable change history
- Acceptance criteria and verification results
- README covering setup, usage and known limitations
A learning project using sample or permitted material. The final scope is agreed for your learning plan.
KEEP EXPLORING
Useful official resources.
Reference material to complement the Hawa Enterprises curriculum. Account features may vary; basic-feature or manual alternatives support the core exercises.
Course outline published by Hawa Enterprises. Updated . Ask about the learning plan.
GOOD TO KNOW
Before you take the next step.
Can I take this without coding knowledge?
You will get more from this course if you can read basic code and use a terminal. Start with Automate with Python or ask us to discuss a suitable foundation first.
Is this an official OpenAI certification?
No. This is an independent Hawa Enterprises course outline. It does not imply OpenAI affiliation, certification or endorsement.
What are the fees, schedule and delivery format?
These details are confirmed by email before you commit. The curriculum gives guided-hour options for Essentials, Practitioner and Advanced. Actual schedules, fees and delivery arrangements are confirmed before enrolment. Ask about the course you want and your preferred timeline.
How do I enrol?
Choose a course and complete the enrollment wizard with your name, email and WhatsApp number. Review your details before sending the enquiry. Hawa Enterprises will discuss your starting level, goals and course arrangements before enrolment is confirmed.
YOUR NEXT SKILL STARTS HERE
Let’s find the right
course for you.
Tell us your goal, starting level and preferred timeline.