The problem
Without stable context, each interaction may rediscover the project, repeat decisions and produce disconnected changes. A completed task list may still hide a failing flow.
An open model for
software engineering with AI.
I built modelo_projetos to organize work with AI assistants in my projects. It keeps product context, architecture choices, and test and release commands together. You can use and adapt this base.
MITPT / EN guideFramework independent
01 / Context
Understand before changing.
PATHS · PRD · ESCOPO · DECISOES
02 / Execution
Build within clear boundaries.
QUALITY_ROADMAP · DEMANDAS · TELAS
03 / Evidence
Check what works.
TESTES · BUGS · RUNBOOK · CHECKS
The approach
Keep a map of official sources, clear roles and delivery criteria. Read what matters, compare documents with code and connect each requirement to its evidence.
The intended result
A team can explain what exists, what changed, how it was checked and what remains. The model supports that work. Quality still depends on adoption and review.
Guides for using the model.
Start here
Start here ↗
Review the kit, configure the map and make your first delivery.
WorkflowWorkflow ↗
Context, SOLID, packages, batches, review and operations in one cycle.
ReferenceReference ↗
Understand each document’s role and when to read it.
ExamplesExamples ↗
See a requirement become code, tests and evidence.
DownloadsDownloads ↗
Templates, validators, MIT license and bilingual instructions.
You do not need a specific stack.
The base can guide a service, app, legacy system or product monorepo. It does not choose your architecture. You record real modules and commands, keep working conventions and adapt the policies.
The examples are educational. The download includes real validators, but needs setup and review in the adopted project.
Download the project model.
Kit 1.0.0 · 35 files · MIT · 66 KB
Common questions
Is this a library or an app?
It is a repository engineering and governance base. Documents define context and checks. Scripts help with inventory and validation. It does not replace your product or framework.
Can I use it in commercial projects?
Yes. The kit uses the MIT license. Keep the copyright notice and license. You may adapt it to your team’s needs.
Can I use it without AI?
Yes. People can read and use the process. An assistant can help with work, but is not required to maintain documents, decisions and tests.
Does a passing check mean a finished product?
No. Checks verify only their own scope. Release, external integrations, restore, devices and human acceptance need their own evidence.
Do I need access to a private repository?
Not to use the kit. The ZIP includes adoption documents and scripts, without credentials or personal machine paths.