carchitect

Generates a machine-readable ARCHITECTURE.md documenting codebase structure and constraints.

65|2|Updated Mar 27, 2026
One-click install
npx skills add https://github.com/joshft/correctless --skill carchitect
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: carchitect
Source: https://github.com/joshft/correctless/tree/main/correctless/skills/carchitect
Command: npx skills add https://github.com/joshft/correctless --skill carchitect

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

It helps engineering teams formalize and communicate system architecture decisions by producing a machine-readable ARCHITECTURE.md, ensuring consistency between code, design, and reviews.

Core Features & Use Cases

  • Generate a structured ARCHITECTURE.md containing an Entrypoints YAML block and prose sections for both reverse-engineered and greenfield designs.
  • Support mode selection (reverse-engineer or greenfield) to capture existing patterns or drive discovery in new projects.
  • Provide a templated documentation workflow that serves as a single source of truth for architecture decisions and constraints.

Quick Start

Run the architecture-definition agent to generate and populate .correctless/ARCHITECTURE.md for this project.

Frequently Asked Questions about carchitect

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I document architecture decisions for an existing codebase?

You document architecture decisions by reverse-engineering existing code to capture entrypoints, patterns, and conventions into a machine-readable ARCHITECTURE.md. This file serves as a single source of truth for downstream tooling and code reviews.

What is a machine-readable ARCHITECTURE.md and when do I need it?

A machine-readable ARCHITECTURE.md is a structured documentation file containing an Entrypoints YAML block and prose sections that validates entries against a stated schema. You need it to ensure consistency between code, design, and reviews, logging decisions in a structured format for downstream tooling.

Can I use this architecture documentation workflow for greenfield projects?

Yes, you can use this workflow for greenfield projects by selecting greenfield mode to drive discovery and capture new project designs. It provides a templated documentation workflow to generate and populate the structured architecture file for new systems.

What's the best way to formalize system architecture constraints for automated validation?

The best way to formalize system architecture constraints for automated validation is producing a machine-readable ARCHITECTURE.md with an Entrypoints YAML block. It validates entries against a stated schema, logging decisions in a structured format for downstream tooling.

How do I reverse-engineer codebase patterns into a structured format?

You reverse-engineer codebase patterns by running the architecture-definition agent in reverse-engineer mode to analyze existing code. It captures entrypoints, patterns, and conventions, validating them against a schema and logging them into a structured ARCHITECTURE.md file.