Platform concepts

Understand Wiki-driven AI Architecting, the project workspace, and how Stractal bridges your codebase and AI assistants

Context rot

As codebases scale, AI coding assistants lose understanding of architecture, patterns, and business logic. Teams spend 60-80% of their time correcting AI-generated code that misunderstands the existing system. Stractal calls this problem context rot.

Wiki-driven AI Architecting

Stractal places a Functional-Domain Wiki at the center of the development lifecycle:

Repository analysis — reverse-engineer your codebase into a domain hierarchy
Functional-Domain Wiki — a living document describing what your system does and how
Feature specification — describe a feature in natural language
Impact analysis — identify which domains are affected
Blueprint Bundle — an implementation-ready context package for your AI assistant

flowchart LR
    A["GitHub\nrepository"] --> B["Stractal\nanalysis"]
    B --> C["Functional-Domain\nWiki"]
    C --> D["Feature\nspecification"]
    D --> E["Impact\nanalysis"]
    E --> F["Blueprint\nBundle"]
    F --> G["AI coding\nassistant"]

Project workspace

Each project has four tabs that correspond to stages of the workflow:

Tab	Purpose
Wiki	Synthesize your repository, browse the domain hierarchy, view architecture diagrams and DDD insights
Design	Create and manage feature specifications with scope definitions and impact analysis
Blueprints	Select confirmed designs, generate Blueprint Bundles, and export in your preferred format
Knowledge	Add supplementary context (team conventions, deployment notes, business rules) that gets incorporated into blueprints

Business and technical domains

Business domains describe what your system does. They nest hierarchically until each leaf represents a single atomic operation. They carry concrete artifacts: owned tables, API endpoints, page routes, key types, and state management.

Technical domains describe how something is implemented. They are always leaf nodes attached to a parent business domain, with a subtype indicating their role: api, page, component, database, service, worker, config, or test.

Domain categories

Every domain belongs to one of four strategic categories aligned with Domain-Driven Design:

Category	Priority	Description
Core	Highest	Domains providing competitive advantage
Infrastructure	Low	Generic capabilities (auth, database, middleware)
Support	Medium	Necessary but non-distinctive (billing, email)
External	Minimal	Third-party integrations (GitHub API, Stripe)

Designs

A design is a feature specification that moves through a lifecycle: Draft → Proposed → Confirmed → In blueprint → Implemented → Archived. Confirmed designs include a scope definition (what's in scope, what's out of scope, and the technical approach) and can be linked to Wiki domains through impact analysis.

You can create designs manually from the Design tab or through the AI chat assistant in Build mode.

AI chat assistant

The project chat assistant has two modes:

Ask mode — ask questions about your Wiki, codebase architecture, or domain relationships. Available on all plans.
Build mode — describe a feature and have the assistant create a design proposal. You can accept, reject, or edit proposals before they become designs. Available on paid plans.

You can @mention specific Wiki domains to focus the conversation on particular areas of the codebase.

Blueprint Bundles

When you describe a feature, Stractal identifies affected Wiki domains and generates a context package tailored to your AI coding assistant. Bundles include affected domains, architectural patterns, implementation guidance, and file-level references. Exported in formats for Claude Code, Cursor, Gemini, Copilot, Windsurf, or generic markdown.

On this page