Skip to content
All projects

Developer Tooling · AI-assisted Development

ai-project-factory

A project generator that creates small, runnable Python and web starters with requirements, tests, CI and AI-agent instructions already aligned.

Status
Open source · MIT
Discipline
AI Engineering
Key stack
Python, Copier
Repository
View on GitHub

Overview

Overview

What it does

AI Project Factory is a local, Copier-based generator with three starter profiles: a Python CLI, a web SaaS (Next.js, FastAPI and PostgreSQL) and an AI service (FastAPI with a deterministic mock provider).

Each generated project runs on its own and ships with requirements and architecture documents, tests, CI, and repository instructions, skills and review agents for AI-assisted development. Generation itself makes no AI API calls and needs no paid service.

Problem

Starting a project is easy; keeping its requirements, architecture, tests, AI-assistant instructions and CI aligned is harder. Ad hoc starters drift, and AI coding agents work poorly without clear, consistent project context.

Who it helps

Developers who frequently start small services or tools and want a consistent, tested baseline for requirements-driven, AI-assisted work.

Solution

Solution

  • A Pydantic-validated project schema with compatibility rules, so unsupported or conflicting feature selections are rejected before generation.
  • Optional capabilities, such as OpenAI or Amazon Bedrock adapters, are added only when selected, as tested code or explicit extension seams.
  • Staged generation in a temporary sibling directory, moved into place only after Copier succeeds; existing directories are never overwritten.
  • A data-driven catalog of 22 UI presets with light and dark semantic tokens and accessibility fallbacks, plus a dependency-free gallery.
  • cloud-incident-service was generated from its ai-service profile.

Technology stack

Application
  • Python 3.12+
  • Typer
  • Rich
  • Pydantic
  • Copier
  • Jinja
Generated starters
  • Next.js
  • TypeScript
  • FastAPI
  • PostgreSQL
  • Docker Compose
AI / LLM
  • OpenAI and Amazon Bedrock adapters (optional)
Testing & quality
  • pytest
  • Ruff
  • uv
  • Secret scanning
DevOps
  • GitHub Actions

Infrastructure & architecture

Infrastructure & architecture

A locally run generator with no hosted infrastructure. Its CI verifies both the factory and representative generated output.

  1. CLI
  2. validate
  3. select template
  4. Copier render
  5. staged move
Distribution
Installed with uv tool install; templates are packaged into the wheel. Tagged release v0.1.0 on GitHub.
CI
GitHub Actions runs formatting, linting, tests and a secret scan, then generates a SaaS project and type-checks, tests and builds its frontend.
Containers
Generated web SaaS projects include a Docker Compose setup, which CI validates with docker compose config.
Security
CI runs without secrets, cloud credentials, live models or production databases.
Next project: Website Portfolio