Start with reality.
Describe the people, friction, workflows or non-negotiable constraints you already understand. No technical vocabulary required.
ShoStack is the engineering discipline I use in my everyday work building and maintaining SaaS and business software. Not a prompt pack created for resale. The real working method: inspect, architect, direct agents, verify and deliver.
The public repo shares transferable methods, not private customer code, internal architecture or proprietary prompts.
That's where the ShoStack Rites of Passage begin. Bring an unfinished thought, a painful manual process, or a service you wish existed. A discovery agent asks useful questions one at a time, challenges assumptions, and helps shape what should be built — instead of pretending a single prompt will produce a complete SaaS.
Describe the people, friction, workflows or non-negotiable constraints you already understand. No technical vocabulary required.
Pricing, adoption, features and architectural preferences stay hypotheses until a decision or test supports them.
The agent explains trade-offs, asks the next consequential question and records decisions — including options you've rejected.
Nine revisitable rites connect the idea to real user journeys, feature-versus-product challenges, boundaries, an honest first experiment and eventually verifiable software delivery. Coverage grows over time, not through an overwhelming first-session questionnaire.
A guided discovery method, not automatic market validation or a one-click production app generator. Business viability and release readiness still require evidence.
Different businesses, different engineering demands. These products provide the public context for my work. The method is shareable; each product's private implementation stays private.
A Malay-language learning experience with reading, learning activities and evolving digital content.
An established language-learning platform with exercises and systems evolving while people continue using them.
Customer-facing campervan travel services with digital journeys closely connected to business operations.
AI creative software under active development, with mock-first experiences, architecture work and backend integration planning.
The named products are contexts, not endorsements or performance benchmarks. This does not claim that one identical harness runs in every project, disclose client architecture, or assert commercial valuations.
Real software needs source-of-truth decisions, contracts, regressions, operations and handoffs — not just a productive chat session.
Business outcomes, repository evidence, runtime constraints and acceptance checks before implementation.
Clear boundaries between frontend, backend, agents and integrations, including what happens when things fail.
Relevant tests, negative cases, smoke checks and precise reports of what merged or deployed.
Read the documented workflow and adapt it to the repository you actually maintain.
The original open source methods, reusable starter specifications and free engineering lessons are available now. No imaginary paid product, unverifiable test claims or download gate.
Portable operating skills distilled from real software work.
Open repository ↗Adapt agent instructions, working records, boundary contracts and release evidence.
Get public templates ↗Understand context, state, verification and the layers around an AI agent.
Read the guide →Every product problem is another opportunity to improve the method. Read the lessons behind the harness.