Choosing a backend foundation and collaboration workflow

September 9, 2026 · View on GitHub

Use this guide to decide whether the blueprint fits your project. Evaluate the backend architecture and the repository-specific AI harness together: adopting both means learning the code structure and the process used to change it.

What you gain, and what you take on

NeedWhat this blueprint offersAdoption cost or boundary
Multiple business domainsShared CRUD bases, domain discovery, and a consistent Router → Service → Repository path.Learn the layer boundaries, DTO conventions, and dependency-injector wiring; a small API may not justify the structure.
API, worker, and admin access to business logicFastAPI, Taskiq, and NiceGUI surfaces around domain services.Learn each interface's contracts and authentication requirements. The MCP server is still planned.
A local evaluation before provisioning servicesSQLite, InMemory infrastructure, and deterministic AI stubs.Evaluation does not validate production capacity or model quality. Real deployments need infrastructure, credentials, and operational configuration.
AI features without making them mandatoryOptional embedding, LLM, and RAG adapters and examples.Select and configure provider extras, models, credentials, and storage appropriate to your service.
Several contributors using AI coding toolsShared rules and skills, a plan/execute workflow, checks, and review procedures across tool adapters.Contributors must set up their tools and retain human review. Some workflow controls are reminders, not blocking checks.
Continued architectural consistencyImport checks, documented contracts, review checklists, and guideline synchronization.These controls require maintenance as your application changes; they do not prove every implementation is correct.

Start with the backend demo and the API-change workflow. Try one domain change before deciding whether the conventions fit your team.

When a different starting point may fit better

  • A small, single-purpose API: a minimal FastAPI application can avoid abstractions you do not yet need.
  • A customer frontend included from day one: inspect a full-stack starter against your frontend, authentication, and deployment requirements. This blueprint's admin UI serves operators, not your customer application.
  • An established architecture you want to retain: assess individual patterns before copying shared infrastructure or adopting the full harness.
  • A framework or raw-throughput decision: benchmark your own workload and compare deployment and ecosystem requirements. This repository is not a comparative performance study.
  • A standalone, framework-independent harness: the rules and skills here reference this repository's paths, layers, and commands. Extracting them needs adaptation; a turnkey independent harness package is not provided.

Alternatives to explore

The links below are starting points, not feature rankings. Check the current documentation and code of each candidate for the capabilities you need.

DecisionOfficial sources to inspect
Build a small FastAPI application yourselfFastAPI tutorial
Evaluate another template's structure and setupFull Stack FastAPI Template, s3rius/FastAPI-template, teamhide/fastapi-boilerplate
Compare Python web frameworksLitestar, Robyn
Generate a project from a configurable templateCookiecutter

For every candidate, check: can the team understand one feature end-to-end, run it locally, verify changes, configure deployment, and maintain its chosen conventions? Feature counts alone do not answer those questions.

Adoption paths

  • New project: use the GitHub template, run the evaluation, then follow the first-domain tutorial.
  • Existing project: use the adoption guide to assess a gradual introduction of patterns. Copying _core or the harness requires checking dependencies and compatibility with your application's conventions.
  • Manual or AI-assisted development: both use the same backend structure. The tool guide adds AI collaboration to that foundation; the shared operating model explains workflow responsibilities and exceptions.

Treat future upstream changes as code to review against your own modifications, not as automatically safe template upgrades.