Mobile money is a reliability problem
M-Pesa, Airtel Money and Tigo Pesa callbacks can arrive late, twice or not at all. Design reconciliation, idempotency and an audit trail before the happy path.
ADVANCED COURSE · AI-ASSISTED ENGINEERING FOR EAST AFRICAN TEAMS
The Senior Architect
& Engineer
Vibecoding Prompt Playbook.
A free advanced vibe coding course for developers, tech leads and architects in Tanzania, Kenya, Uganda and across East Africa. Build production software with AI — and keep the judgment that makes it safe to ship.
18 modules392 copy-ready promptsAbout 12 hoursFree · No sign-up
Scope before you build.
Turn a WhatsApp brief, a donor document or a founder’s idea into testable requirements, explicit non-goals and a risk register.
Direct the agent like a senior engineer.
Ask for plans, assumptions, small diffs, tests, evidence and rollback — instead of accepting the first confident answer.
Ship for East African reality.
Field notes on mobile money, low bandwidth, mixed Android devices, multi-currency data and handovers to local IT teams.
THE THREE RULES BEHIND EVERY PROMPT
A longer prompt is not a better prompt. A sharper one is.
Across all eighteen modules, the same discipline applies. Keep these three sentences beside you when you open your AI coding tool.
01Name the invariant.
State the rule that must not change. It keeps a refactor from becoming a rewrite and a fix from creating a new bug.
02Ask for the smallest viable change.
Constrain the diff, name the files in scope, and make larger cleanup a separate, reviewable piece of work.
03Ask what could go wrong.
Demand the failure mode, the evidence you would see, and the test that would catch it before a user does.
THE COURSE PATH · 18 MODULES
From the first requirement to the release you can defend.
Work in order or open the module you need today. Every module ends with a prompt library and a practice task for your own project.
Part 1 · Decide & design
2 modules · 40 prompts- 0120 PROMPTS · 40 MIN
Requirements & Scoping
Turn a vague ask into a testable spec, an honest scope, and a decision you can defend.
PRDs, acceptance criteria, MVP scope, risk registersOpen module & prompts - 0220 PROMPTS · 45 MIN
Architecture & System Design
Choose a design by its trade-offs, failure modes and operating cost — not by confidence.
Options, boundaries, interfaces, failure modes, migration pathsOpen module & prompts
Part 2 · Build & refactor
2 modules · 60 prompts- 0320 PROMPTS · 40 MIN
Code Generation & Implementation
Constrain the diff, preserve the invariant, and make the agent show its assumptions.
Small diffs, existing patterns, error paths, assumptions, rollbackOpen module & prompts - 0440 PROMPTS · 40 MIN
Refactoring & Code Quality
Change the shape of the code without changing what users observe.
Behavior invariants, seams, coupling, deletion, green commitsOpen module & prompts
Part 3 · Prove & harden
5 modules · 120 prompts- 0540 PROMPTS · 45 MIN
Debugging & Root-Cause Analysis
Replace guessing with hypotheses, evidence, and the first point of divergence.
Hypotheses, reproduction, timelines, postmortems, instrumentationOpen module & prompts - 0620 PROMPTS · 40 MIN
Testing & QA
Write the tests that would have caught the bug, not tests that confirm the code runs.
Boundaries, contract tests, regression tests, test strategyOpen module & prompts - 0720 PROMPTS · 35 MIN
Code Review & Critique
Make the agent disagree with you — specifically, with evidence, in priority order.
Risk-first review, hidden coupling, reviewable diffs, API critiqueOpen module & prompts - 0820 PROMPTS · 40 MIN
Performance & Optimization
Measure first, name the bottleneck, and quantify the gain against the risk.
Profiling, N+1 queries, caching, tail latency, costOpen module & prompts - 0920 PROMPTS · 45 MIN
Security & Reliability
Think like an attacker at the boundary and an on-call engineer at 3 a.m.
Threat modelling, authorization, secrets, failure handling, recoveryOpen module & prompts
Part 4 · Ship & sustain
4 modules · 80 prompts- 1020 PROMPTS · 40 MIN
DevOps, CI/CD & Deployment
Assume the deploy will go wrong; design the detection, the rollback and the blast radius.
Rollback, migrations, feature flags, alerts, runbooksOpen module & prompts - 1120 PROMPTS · 35 MIN
Documentation & Communication
Write for the reader who will make the decision or fix the system without you.
ADRs, READMEs, runbooks, release notes, onboardingOpen module & prompts - 1220 PROMPTS · 40 MIN
Legacy & Migration
Prefer incremental coexistence to a heroic big-bang rewrite.
Strangler fig, characterization tests, cutover, rollbackOpen module & prompts - 1320 PROMPTS · 45 MIN
Data & Databases
Design the schema around access patterns, then prove it with the query plan.
Schema, indexes, migrations, concurrency, retentionOpen module & prompts
Part 5 · Lead the agent & the interface
5 modules · 92 prompts- 1420 PROMPTS · 35 MIN
AI-Agent Meta-Prompts
Improve every other prompt by changing how you direct the agent.
Plans, assumptions, checkpoints, citations, self-reviewOpen module & prompts - 1520 PROMPTS · 40 MIN
UI/UX & Interface Engineering
Ask for states, constraints and accessibility — not “a nice design”.
States, forms, errors, responsiveness, accessibilityOpen module & prompts - 1620 PROMPTS · 35 MIN
High-Impact Architectural Questions
The questions that prevent systemic regressions after the immediate fix.
Invariants, blast radius, reversibility, ownershipOpen module & prompts - 1712 PROMPTS · 30 MIN
UI/UX Layout Audit & Remediation
Run an audit that produces evidence and a sequenced remediation plan.
Toolbar fragmentation, component palette, inspector stability, silent failuresOpen module & prompts - 1820 PROMPTS · 40 MIN
Feature Fix, Refine & Improve
Improve a live feature without breaking the users already depending on it.
Baseline, invariants, edge cases, incremental deliveryOpen module & prompts
EAST AFRICA FIELD NOTES
Built for the conditions your users actually have.
The prompts are universal. The context is not. Each module includes a field note for teams building in Tanzania, Kenya, Uganda and the wider region.
Low bandwidth is a design constraint
A 2 GB Android phone on 3G is a real user, not an edge case. Measure payload size, design offline states, and test on the devices your users actually hold.
Multi-currency, multi-language
TZS, KES and UGX do not round the same way, and English or Swahili copy changes layout. Put money in integer minor units and test the longest string.
Small teams, real handovers
One maintainer is a bus factor, and the client’s IT officer may run the system after you leave. Write the README, ADR and runbook as deliverables.
THE COMPLETE PROMPT LIBRARY
392 prompts.
Search, copy, use.
Find a prompt by the problem you are facing — rollback, N+1, permissions, race condition, migration, empty state — then copy it with one click. No account needed.
- 01Copy the prompt.
- 02Replace every [bracketed] slot.
- 03Attach the relevant files or context.
- 04Demand evidence, not confidence.
VIBE CODING, AI PROMPTS & EAST AFRICA — FAQ
Questions developers ask before starting.
Straight answers about the course, the prompts and using AI coding agents on real projects in Tanzania, Kenya and Uganda.
What is the Senior Architect & Engineer Vibecoding Prompt Playbook?
It is a free advanced course and prompt library for people who build production software with AI coding agents. The playbook contains 18 modules and 392 prompts covering requirements, architecture, implementation, refactoring, debugging, testing, review, performance, security, deployment, documentation, migration, data, UI/UX and agent direction. Every prompt is copy-ready and includes a short explanation of why it works.
Is this vibe coding course free in Tanzania, Kenya and Uganda?
Yes. The course, the prompt library and the field notes are free to read and use. You do not need an account to copy a prompt. Your AI tool, model usage, hosting and any paid developer tools have their own costs.
Do I need to know how to code to use these AI prompts?
The prompts are written for people who are accountable for real software, so you will get the most from them if you already work with code, APIs, databases or technical projects. If you are starting from zero, begin with our free Vibe Coding course, then come back to this playbook when you have a small project to practise on.
Which AI coding tools do these prompts work with?
They work with ChatGPT, Claude, Gemini, Cursor, Windsurf, GitHub Copilot, Codex, OpenCode and any other assistant that accepts a text brief. The prompts are tool-neutral: they ask for plans, evidence, diffs and trade-offs rather than relying on one product's features.
Are the prompts adapted for East African teams?
The prompt patterns are universal, and each module adds an East African field note. The notes cover mobile money callbacks (M-Pesa, Airtel Money and Tigo Pesa), TZS, KES and UGX amounts, low-bandwidth users, mixed Android devices, WhatsApp-first workflows, multi-tenant data and handovers to local IT teams. They are practical engineering prompts, not legal advice.
Can I use these prompts for client or agency work?
Yes. Copy a prompt, replace the [bracketed] slots with your project context, and keep the invariant-and-evidence discipline the playbook teaches. Do not paste client secrets, personal data or production credentials into a tool your client has not approved.
How do I copy a prompt?
Open any module or the full prompt library, press the Copy prompt button on a card, and paste it into your AI tool. Text in [square brackets] is a slot for your own context. On a module page you can also copy all of that module's prompts at once.
Where can I learn vibe coding in Tanzania, Kenya or Uganda?
You can take this course online from Dar es Salaam, Nairobi, Kampala or anywhere with a browser. It is self-paced, free and mobile-friendly, so you can read a module, copy a prompt and practise in your own editor without travelling to a bootcamp. The field notes are written for teams working in Tanzania, Kenya, Uganda and the wider East African market.
Is this a prompt engineering course?
Yes. It is prompt engineering for software delivery rather than generic chatbot tricks. You will learn how to write prompts that produce specifications, architecture options, small diffs, tests, reviews, runbooks, migrations and security checks — and how to verify the output instead of trusting it.
Je, kozi hii ya vibe coding ni bure Tanzania, Kenya na Uganda?
Ndiyo. Kozi, field notes na prompts zote ni bure kutumia. Huhitaji akaunti ku-copy prompt. Gharama za AI tool, model usage, hosting au zana nyingine za developer zinabaki kwa akaunti yako mwenyewe.
How long does the course take?
There are 18 modules and about 12 hours of reading and practice. You can work through it in order or use it as a reference during a project. Each module ends with a small practice task you can finish in your own repository.
Continue with Vibe Coding: Build with AI, Think Like an Engineer · Your first steps with AI & prompting · AI ni Nini na Kipato cha Nyumbani · AI kwa Wafanyabiashara Wadogo.
KWA KISWAHILI
Kozi hii inafundisha kwa Kiingereza cha kiufundi.
Watengenezaji wa programu Tanzania, Kenya na Uganda wanaweza kutumia prompts zote bure. Kama unaanza kutoka sifuri, anza na kozi ya AI ni Nini au kozi ya Vibe Coding, kisha rudi hapa kwa playbook ya ujenzi wa mifumo halisi.