Skip to main content

Quick Start

Get SoftArchitect AI running locally in under 5 minutes with Docker Compose.

Architecture Overview

Understand the Clean Architecture + Hexagonal design behind the platform.

RAG Engine

See how Retrieval-Augmented Generation powers architecture recommendations.

API Reference

Explore the FastAPI backend endpoints for chat, projects, and knowledge.

What is SoftArchitect AI?

SoftArchitect AI is not another “chat with code” tool. It’s your Shadow Software Architect — an AI-assisted work environment that guides you through the most critical and overlooked phase of software development: project conception. Through a Guided Architectural Workflow (Context → Requirements → Architecture → UX/UI → Planning), it acts as a preventive quality gate — ensuring your application design complies with SOLID principles, Clean Architecture, and OWASP security guidelines before you write the first line of code.

Local-First Privacy

Run LLMs entirely on your machine with Ollama. Your data never leaves your network.

RAG Knowledge Base

29 curated architectural documents, 934 vector embeddings powering every recommendation.

Tech Packs

Stack-specific architecture rules for Flutter, Python, Firebase, and more.

Streaming Chat

Real-time streaming responses with optimistic UI and WebSocket support.

Context Factory

Auto-generates AGENTS.md and RULES.md so your AI copilot works better.

Docker Deployment

One command brings up the full stack: API, ChromaDB, and Ollama in under 2 minutes.

The architectural workflow

SoftArchitect AI guides you through five sequential phases, each building on the last:
1

Context gathering

Define your project context: team size, experience level, business domain, and constraints. The AI interviews you to build a complete picture.
2

Requirements analysis

Transform business needs into technical requirements. The RAG engine retrieves relevant architectural patterns from the knowledge base.
3

Architecture design

Get concrete architecture recommendations — Clean Architecture layers, hexagonal ports, SOLID compliance checks, and OWASP security gates.
4

UX/UI planning

Plan the user experience with wireframe descriptions and component hierarchies aligned to your chosen stack.
5

Development planning

Generate a prioritized backlog with user stories, acceptance criteria, and technical tasks ready for your sprint planning tool.

Tech stack

LayerTechnology
FrontendFlutter Desktop (Linux / Windows / macOS)
BackendPython 3.12 · FastAPI · LangChain
AI EngineOllama (local) · Groq · Gemini (cloud)
Vector StoreChromaDB
InfrastructureDocker Compose
Testingpytest · flutter_test · 1,286 tests · ≥80% coverage

Project status

SoftArchitect AI is currently in Pre-Alpha. Core features are functional with 1,286 passing tests and ≥80% coverage. The project is licensed under GPL-3.0.
MetricValue
Tests893 Flutter + 393 Python
Test Coverage≥80%
Knowledge Base29 files · 934 vectors
ArchitectureClean + Hexagonal
Security Gates16/16 passing
Deployment<2 min Docker

Build docs developers (and LLMs) love