The Software Engineering System
178 copy-paste AI prompts for engineers — READMEs and API docs, architecture decision records, code review, debugging, refactoring, tech debt, testing, performance, runbooks and release notes.
Key Capabilities & Features:
- •The Codebase Primer: paste your stack, team size, conventions and constraints once — every prompt after it writes in your idiom
- •Code review prompts that separate blocking issues from personal preference, which is most of what makes review culture bad
- •Debugging prompts that force a hypothesis before a fix, rather than changing things until the symptom disappears
- •Refactoring plans sequenced so the work can be paused halfway without leaving the codebase worse
- •178 prompts across 12 sections — readmes & project docs, api documentation, architecture decision records, code review and more
- •Works with ChatGPT (GPT-5), Claude Opus 5 and Gemini 3.7 Flash — fill the bracket, paste, done
One-time payment • Lifetime access & future updates
What's Inside
12 modules178 promptsEvery module below is included in the single purchase.
- 1
READMEs & Project Docs
14 promptsGetting someone running in under ten minutes, prerequisites stated honestly, what the project is and is not for, and docs that survive a refactor.
Replaces a system previously sold separately
- 2
API Documentation
16 promptsReference docs with examples that actually run, error responses documented, auth flows explained once properly, and versioning and deprecation notices.
Replaces a system previously sold separately
- 3
Architecture Decision Records
14 promptsRecording the context and constraints, options genuinely considered, the decision and its consequences, and what would cause you to revisit it.
Replaces a system previously sold separately
- 4
Code Review
16 promptsReview comments that teach rather than assert, separating blocking from preference, reviewing a large PR without rubber-stamping, and receiving criticism well.
Replaces a system previously sold separately
- 5
Debugging & Root Cause
18 promptsNarrowing a bug systematically, forming and testing hypotheses, reading a stack trace properly, reproducing an intermittent failure, and writing a blameless post-mortem.
Replaces a system previously sold separately
- 6
Refactoring
16 promptsSequencing a refactor so it can be paused, characterisation tests before changing behaviour, keeping a PR reviewable, and knowing when not to refactor at all.
Replaces a system previously sold separately
- 7
Tech Debt
14 promptsInventorying debt with its actual cost, distinguishing debt from code you dislike, making the business case for paydown, and a plan that fits alongside feature work.
Replaces a system previously sold separately
- 8
Testing
18 promptsTest cases from a spec, edge cases and boundary conditions, what deserves a test and what does not, flaky test triage, and testing at the right level.
Replaces a system previously sold separately
- 9
Performance
14 promptsMeasuring before optimising, profiling interpretation, common bottleneck patterns, cost against benefit, and the optimisation that makes the code worse for no gain.
Replaces a system previously sold separately
- 10
Runbooks & Incidents
12 promptsRunbooks usable at 3am by someone unfamiliar, severity definitions, incident comms to stakeholders, on-call handover, and post-incident actions that get done.
Replaces a system previously sold separately
- 11
Release Notes & Changelogs
10 promptsNotes written for users rather than from commits, breaking changes made unmissable, migration guides, and versioning that communicates something.
Replaces a system previously sold separately
- 12
Technical Writing
16 promptsKnowledge base articles and FAQs written from real support tickets, how-to guides someone can follow unaided, and knowing when a doc should be a fix in the product instead.
Replaces a system previously sold separately
Test Workflow Execution in Your Browser
Verify actual prompt structure, token speed, and reasoning depth before downloading.
Frequently Bought Together
Technical Specifications
Architecture & System Overview
The code is rarely the problem. It is the README that assumes six things, the review comment that reads as an attack, the bug someone fixed without ever knowing why it happened.
The Software Engineering System is a 178-prompt playbook covering READMEs and project docs, API documentation, architecture decision records, code review, debugging and root cause, refactoring, tech debt, testing, performance, runbooks and incidents, and release notes.
Start with the Codebase Primer — paste in your stack, team size, conventions and real constraints — and every prompt after it writes in your idiom rather than in generic pseudocode.
The review prompts separate blocking issues from preference, because a review that treats a naming opinion the same as a correctness bug is why people dread opening PRs. The debugging prompts insist on a hypothesis before a fix, since changing things until the symptom goes away is how you ship a bug that returns in three weeks. And the refactoring plans are sequenced to be pausable — real refactors get interrupted, and one abandoned halfway should not leave the codebase worse than before.
Compare with Similar AI Templates
| Specification | ||||
|---|---|---|---|---|
| Customer Rating | 4.93(45) | 4.91(43) | 4.91(38) | 4.92(36) |
| AI Models | GPT-5, Claude Opus 5, Gemini 3.7 Flash | GPT-5, Claude Opus 5, Gemini 3.7 Flash | GPT-5, Claude Opus 5, Gemini 3.7 Flash | GPT-5, Claude Opus 5, Gemini 3.7 Flash |
| Formats Included | PDF Prompt Playbook | PDF Prompt Playbook | PDF Prompt Playbook | PDF Prompt Playbook |
| Lifetime Updates | Yes | Yes | Yes | Yes |
| Creator Rating | Marcus Sterling (99.3%) | Dr. Lucas Vance (99.2%) | Elena Rostova (99%) | Dr. Lucas Vance (99.1%) |
Community Reviews & Ratings
Verified buyer ratings & prompt benchmarks
Write a Review
Sign in to leave a review for this system.
Verified User Feedback
Hypothesis-before-fix caught a real cause
Intermittent failure we had been patching for weeks. It ranked four causes and flagged that three of them did not explain why it started when it did. The one that did was correct. Fixed properly rather than suppressed.
Review culture noticeably better
Splitting comments into blocking and preference has changed how our juniors experience review. Two of them have said so unprompted.
Pausable refactor plans
We always abandon refactors halfway when something urgent lands. Sequencing them so each stage leaves the codebase in a valid state is the first version of this that has survived our reality.
Not a coding assistant, as stated
If you want code written, this is not it and the FAQ is upfront about that. Docs, review and incident material is genuinely strong.
Frequently Asked Questions
How is this different from the separate systems you used to sell?
It is all of them, in one purchase. README & Project Overview Builder, API Documentation & Example Builder, Architecture Decision Record Builder, Code Review Comment & Feedback System, Debugging & Root-Cause Analysis, Refactoring Plan & Task List, Tech Debt Inventory & Paydown Plan, Test Case & Scenario Generator, Performance Optimization Plan, Release Note & Changelog Writer, DevOps Runbook & Incident Plan, TypeScript Component & Prop Plan, Python Automation Script Planner, Internal Tooling & Automation Plan, Knowledge Base & FAQ Builder, Tutorial & How-To Guide Builder are now sections inside this toolkit. Bought individually they came to $332.
Does it write code?
That is not what it is for. It covers the surrounding work — documentation, review, debugging method, refactor sequencing, testing strategy, incident practice. Your IDE assistant writes code; this handles the parts that decide whether the codebase stays workable.
Is it tied to a language or stack?
The Codebase Primer asks yours and the prompts adapt terminology and idiom. Most content — review, debugging, ADRs, testing strategy, incidents — is language-independent. Where something is genuinely stack-specific the prompt says so.
Will it hallucinate APIs?
The Primer sets a standing rule against inventing library methods and instructs it to flag uncertainty instead. That reduces the problem substantially but does not eliminate it — verify anything it tells you about a library before relying on it.
How is it delivered?
You receive a PDF with all 178 prompts organized into 12 sections, delivered within 24 hours of purchase. No app or account required — copy a prompt, fill in the bracketed variables, and paste it into ChatGPT, Claude, or Gemini.
Can I use this for client work?
Yes. Your purchase includes a single-user commercial and personal license, so you can use every prompt in your own business or for client deliverables without restriction.