Hey Guys,I’m Yafei Lee, a full-stack Ruby on Rails engineer with 15 years of coding experience. Today, I’m thrilled to introduce you to a new AI engineer: Clacky.

We’ve always believed that the future of programming would inevitably shift to the cloud. That’s why, starting in 2021, our team spent over 1,000 days developing a Cloud Development Environment（CDE）. Fortunately, the rise of the AI era powered by large language models (LLMs) has massively sped up the adoption of cloud development.

After over 8 months of refining the product internally, I’m thrilled to officially introduce ClackyAI — a truly AI-native, CDE-powered AI Agent.

The idea behind ClackyAI isn’t to replace engineers but to work alongside them, creating a collaborative environment where they can focus more on architecture and business logic in an AI-first development world.

The Problems and Scenarios Clacky Aims to Solve

"All we want to do is code, nothing else." That’s every engineer’s dream — to focus entirely on building core functionality, while leaving everything else to Clacky.

Here’s how Clacky makes that happen:

1. Automated Environment Setup

   
   

   Clacky takes care of setting up your cloud development environment for you. It currently supports languages like Golang, Node.js, Python, Java, and Ruby, whether you're working on front-end, back-end, or full-stack "monolithic" projects. Plus, it comes with built-in support for databases like MySQL, Redis, Postgres, and MongoDB.

   
   

2. Natural Language Requirements

   
   

   With Clacky, you can submit requirements (Issues) in plain natural language. Clacky will fork a new Thread, plan the tasks, and get everything organized. Once you confirm, you can see the AI in action, making real progress. Need to make changes? You can also add new requirements or roll back part of its work anytime.

   
   

3. Seamless Deployment

   
   

   Once the work is done, no more fiddling with Git commands. With just a few clicks, Clacky handles everything for you, from wrapping up the changes to submitting the PR to the main branch.

   
   

AI-Native CDE: Unlocking the Full Potential of AI

In the future, 80% of code will come from AI, with just 20% from humans. Embracing this vision, we’ve built a cutting-edge, AI-first, fast, and API-rich cloud development environment where AI truly thrives. It’s designed to “live” within the environment itself, fully understanding and mastering the entire codebase. Not only can it better plan tasks, but you can even watch it "dance" within the dev environment, giving humans the tools to make smarter decisions about AI’s actions.

Here’s what our AI-Native CDE offers:

• Lightning-Fast Setup: Launch a fully functional development environment for a “monolithic” project in just 2 seconds. With one click, configure persistent layers like MySQL, Redis, and Postgres — all seamlessly driven by the AI Agent. Set it up once, and reuse it endlessly.

  
  

  

• A Great Experience for Both AI and Humans: While the CDE is built to let AI perform at its best, we’ve also designed it to align with human coding habits. Features like Ctrl+P for file navigation, LSP-powered code completion, and support for vim/emacs programming modes ensure a smooth experience for developers. One of our core principles in designing this architecture was to make it open, transparent, and easy to use, with minimal learning curves.

  
  

  

  
  

We believe this is the next generation of AI-driven, cloud-based development environments — a leap akin to moving from gas-powered vehicles to fully electric ones. It’s an exciting shift, and we think you’ll love giving it a try.

AIAgent & A New Human-AI Collaboration — Setting Boundaries for AI-Generated Code

So, how well does Clacky AIAgent perform? That’s likely the question on everyone’s mind. In my view, the real focus shouldn’t be on its current capabilities, as the AIAgent is constantly evolving. Instead, the key is to establish a robust framework that guarantees a reliable baseline for effective, high-quality AI-generated code.

Here’s how we achieve that:

1. Structured Task Planning
   
   

   Tasks are organized into a structured hierarchy: Task → TaskStep → TaskAction. Before starting any work, the AI collaborates with you to confirm its task plan. Each action it performs is clearly presented, giving you full control to regenerate, undo, or adjust the results as needed.
   
   

   

   
   

2. Task Time Machine
   
   

   All AI-generated outputs are displayed in a “time machine” view, allowing you to replay and adjust results at any time. This ensures transparency and puts you in full control of every revision.
   
   

   
   
   

3. Seamless Integration with Familiar Workflows
   
   

   We’ve designed Clacky to work seamlessly with popular L2-level AI coding workflows. For instance, you can quickly select a block of code and tweak it using the shortcut Cmd+K, making the interaction smooth and intuitive.
   
   

   
   
   

AI is undoubtedly the future, but we firmly believe in the importance of human oversight and judgment. By blending the strengths of AI with human expertise, we strive to provide the most reliable and efficient coding experience.

Cursor & VSCode SSH Access - Seamlessly Integrated with Local IDE Workflows

We provide SSH access that works seamlessly with IDEs like Cursor, VSCode, and JetBrains. Once connected, you can keep using your local IDE to edit code, just like you’re used to in your regular workflow.

We treat the local IDE as a client of the cloud development environment, even allowing both local and cloud users to edit the same file at the same time without conflicts (though we don’t recommend doing so).

Issue to PR

Each requirement can have its own independent environment for development and management; you just need to focus on defining clear requirements. Maintenance and updates of Git branches are handled by AI, allowing you to commit and create PRs with just one click.

Note: Currently, only GitHub (public and private) is supported. Let us know if you need support for other platforms!