^ AI-Developer Knowledge Parity System

Caret aims to be more than just a coding tool; it aspires to be an 'AI coding partner' that grows with the user. The core technology that realizes this vision is the 1:1 knowledge parity system, where the AI and the developer share the exact same knowledge.

Why Is This Necessary? For a True Partnership

True partners must view and discuss the same information. This system fundamentally resolves the information imbalance between AI and developers, creating a powerful collaborative environment where the AI accurately understands the developer's intent, and the developer can trust the AI's working methods.

Comparison	Other AI Tools	Caret
Knowledge Source	Separate (AI ↔ Developer)	✅ Synchronized
Transparency	Low (AI's internal process is unclear)	✅ High
Trust Level	Moderate	✅ Very High

Core Structure: .caretrules and caret-docs

This system operates through two separate directories based on their purpose, and their content is always synchronized by the AI agent.

• .caretrules/ (The AI's Brain): A collection of rules and workflows that the AI directly references to perform tasks. It is composed of machine-efficient formats like YAML, JSON, etc.

• caret-docs/ (The Developer's Manual): Human-friendly documents written for developers to easily understand the AI's working methods and the project's rules.

Thanks to this unique dual structure and automatic synchronization mechanism, Caret's AI can function as a true team member that perfectly understands the project's rules and context, rather than just a simple command executor.

You can see a concrete implementation example of this system in the Caret Rules feature document.

Learn More →