Read this document in English | ํ๊ตญ์ด๋ก ์ฝ๊ธฐ | ๆฅๆฌ่ชใง่ชญใ | ้ ่ฏปไธญๆ็
Caret is more than just an AI coding tool; it's a VS Code extension that aims to be an AI companion growing with developers. It preserves the proven advantages of the open-source Cline while 'overlaying' more powerful and flexible features to maximize the development experience.
| Feature | Cline | Cursor | Caret |
|---|---|---|---|
| AI Behavior | Plan/Act (somewhat rigid) | Ask/Agent (natural) | Chatbot/Agent Mode (more intuitive and powerful) |
| AI Transparency | โ Open Source (High) | โ Black Box (Low) | โ Open Source + Improved System Prompt (Very High) |
| AI Efficiency | Basic | Basic | 50% Token Savings through System Prompt Optimization |
| Persona Support | โ Not Supported | โ Not Supported | โ Template & Custom Personas, Profile Image Support |
| Multilingual Support | โ Not Supported | โ Not Supported | โ Full Multilingual Support (i18n Overlay) |
| Architecture | Core Features | Closed | Overlay Structure (Stability + Extensibility) |
Beyond Cline's somewhat rigid Plan/Act mode, Caret offers Chatbot/Agent Modeโflexible like Cursor's Ask/Agent, but more intuitive than 'Ask'. We haven't just changed the interaction style; we've improved the system prompts to enhance both AI response performance and demeanor. Through verified experiments, we've achieved 50% token savings and 20% API cost reduction, enabling smarter and more cost-predictable AI collaboration.
Add joy to your coding with Caret's pre-prepared template personas like K-POP idols, OS-tans, and more. You can register your own AI agent name and profile image to create a visually vibrant development environment.
Default Personas:
Caret: A friendly robot friend who loves coding and helps developers.
Oh Sarang: A K-pop idol and a tsundere engineering girl who helps you navigate between logic and emotion.
Madobe Ichika: A neat and reliable assistant inspired by Windows 11.
Cyan Macin: A concise and efficient helper modeled after macOS.
Tando Ubuntu: A warm collaborator who solves problems with an open-source spirit.
Other AI tools often overlook multilingual support, but Caret solves this. With an i18n-based overlay architecture, developers unfamiliar with English can fully utilize all features in their native languages, including Korean, Japanese, and Chinese.
Caret preserves the core of Cline's proven stability and layers its innovative features on top as an 'overlay'. This allows you to benefit from Cline's stability and transparency while experiencing Caret's powerful extensibility.
- Installation: Search for "Caret" in the VS Code Marketplace and install it. (Coming Soon)
- Select Persona: Choose your favorite AI persona from the sidebar or create your own.
- Start Chatting: Begin coding with your AI companion!
Caret continues to evolve towards becoming the 'ultimate AI companion'.
- Self-Login & Credit System: We are preparing self-login functionality (available within 1 week) and a credit purchase system (available within 2 weeks).
- sLLM & Sovereign Model Support: We will enhance support for local LLMs (sLLM) and specialized sovereign models for security and cost efficiency.
- Community-Driven Feature Expansion: We plan to add features co-created with user feedback and contributions.
Caret is an open-source project that grows with your participation. We welcome any form of collaboration, including bug reports, feature suggestions, and code contributions!
| Contribution Type | Description | Benefits |
|---|---|---|
| ๐ป Code Contribution | Feature development, bug fixes, documentation improvements | Service Credits + GitHub Contributor Listing |
| ๐ Bug Report | Issue reporting, providing reproduction steps | Service Credits |
| ๐ก Idea Suggestion | Proposing new features, improvements | Service Credits |
| ๐ฐ Financial Contribution | Project sponsorship, development support | Service Credits + Special Contributor Listing |
| ๐ Documentation | Writing guides, translations, tutorials | Service Credits + Documentation Contributor Listing |
- Service Usage Credits: Caret service credits provided based on contribution scale
- GitHub Contributor Listing: Name listed in project README and release notes
- Service Page Listing: Profile listed on the official website's contributor page
- Priority Support: Priority access to new features and beta versions
| Profile | GitHub ID | Role |
|---|---|---|
 |
@potakim | Business Planning, Server & Cloud Development, Community Support |
 |
@fstory97 | Business Planning, Client & AI Research Development, Community Support |
 |
@ysyang1973 | Documentation |
 |
@jikime | Community Support, Preview and LangConnect Development |
 |
@parkjongill | Business Planning |
 |
@freelife1191 | Backend Development, ML Ops |
 |
@allipcloud | Educational Content Development, Business & Sales Support |
 |
@jaydenchoe | Educational Material Creation, Educational Content Development, RAG & Memory Development |
 |
@su-record | Client and Service Frontend Development |
 |
@calmglow | Proposed the project name 'Caret' |
- Check Issues: Find issues to contribute to on GitHub Issues
- Join Discussion: Share feature suggestions or questions in Issues or Discussions
- Code Contribution: Contribute code through Fork โ Develop โ Pull Request process
- Documentation Contribution: Improve or translate documents in the
caret-docs/folder
For detailed contribution guidelines, please refer to CONTRIBUTING.md.
All information necessary for Caret project development is systematically organized here.
- Developer Guide (DEVELOPER_GUIDE.md) - Basic information on build, test, and packaging
- Development Guide Overview (development/) - Navigation for all development guides
- Caret Architecture Guide - Fork structure, extension strategy, design principles
- Extension Architecture Diagram - Overall system structure visualization (Mermaid)
- New Developer Guide - Project onboarding and development environment setup
- Testing Guide - TDD, test writing standards, coverage management
- Logging System - Integrated logging, debugging, development/production modes
- Interaction Patterns - Preventing circular messages, Optimistic Update
- Webview Communication - Message types, state management, communication structure
- UI-Storage Flow - Data flow and state management patterns
- AI Message Flow Guide - Complete AI message send/receive flow
- System Prompt Implementation - System prompt design and implementation
- Message Processing Architecture - Message processing system design
- Component Architecture - React component design principles
- Frontend i18n System - Multilingual support implementation (UI)
- Backend i18n System - Multilingual support implementation (System Messages)
- Utilities Guide - Development utilities usage
- File Storage and Image Loading - File processing system
- Link Management Guide - Link management system
- Support Model List - AI model support status
- Documentation Guide - Documentation standards and conventions
- JSON Comment Conventions - JSON file comment writing rules
- AI Work Index Guide - AI Essential Pre-Reading ๐
- AI Work Guide - TDD, architecture review, Phase-based work
- Environment Setup: Developer Guide โ Development Guide Overview
- Project Understanding: New Developer Guide โ Caret Architecture Guide
- AI System Understanding: AI Message Flow Guide โ System Prompt Implementation
- Start Development: AI Work Guide โ Testing Guide
- Advanced Features: Interaction Patterns โ Component Architecture
- Task Documents - Specific implementation task guides
- Strategy Documents - Project vision and roadmap
- User Guide - Usage instructions for end-users
๐ก Mandatory read before starting development: Please familiarize yourself with the TDD-based development process and architectural principles in the AI Work Methodology Guide.
โก Want to understand AI system?: Check the AI Message Flow Guide to see the complete process of how user messages are sent to AI and responses are received!