Alibaba's Qoder: This NEW AI Editor is FULLY FREE w/ Sonnet.. But is it Good?

Key Concepts:

• AI Code Editor (Coder by Alibaba)
• Quest Mode (Asynchronous task execution)
• Byte Rover (Memory layer for AI coders)
• MCP (Memory, Context, and Persistence) Tools
• Codebase Indexing
• Repo Wiki (Automated documentation)

1. Introduction to Coder (Alibaba's AI Code Editor)

• Coder is a new AI code editor developed by Alibaba, similar to Cursor, Trey, Windsurf, or Kiro.
• It's currently free and doesn't have rate limits.
• The AI model used is not selectable but is stated to be either Sonnet or an OpenAI model.
• Key features include intelligent codebase search, repository insights, memory, and tool use.

2. Quest Mode: Asynchronous Task Execution

• Quest mode is a feature similar to Jewels or Codeex, allowing users to assign tasks to the AI and have them run in the background.
• Users can check back later for updates and results.
• This enables asynchronous task execution, allowing developers to continue working on other things while the AI handles specific coding tasks.

3. Codebase Indexing and Repo Wiki

• Coder indexes the codebase for efficient search and understanding.
• It includes a "Repo Wiki" feature that automatically generates documentation for the codebase.
• This documentation feature is designed to improve code maintainability and onboarding for new developers.

4. Integration and Compatibility

• Coder has integrated memory capabilities.
• It is compatible with MCP tools, enhancing its ability to remember context and persist information across sessions.

5. Hands-on Experience and Initial Impressions

• The initial setup involves downloading and signing up for an account.
• The editor's performance is noted to be somewhat sluggish, especially on the first start, due to added animations.
• The interface resembles VS Code, lacking the distinct styling of Kio or Trey.

6. Chat Interface and Agent Options

• The right panel provides access to "Agent" and "Ask" options.
• "Ask" is for general chat without code editing capabilities.
• "Agent" uses tools to make changes to the codebase.
• Users can manually add context by selecting specific files.

7. Configuration Settings

• Coder settings allow customization of languages, response language, and code completion settings.
• Users can adjust autocomplete settings and completion length.
• Auto-run settings are available for MCP, web search, and terminal commands.
• Inline chat completions provide tooltips for editing within the chat interface.
• Quest mode, rules, memories, and indexing can also be configured.

8. Byte Rover: Memory Layer Configuration

• Byte Rover is a memory layer that allows users to create and store memories accessible by the AI coder.
• It can be configured in other coders that support MCP, enabling memory synchronization across different tools.
• Byte Rover facilitates memory sharing among team members.

9. Demo: Autoplay Chess Game

• The AI is tasked with creating an autoplay chess game.
• The AI uses modern coding conventions, including creating a to-do list.
• The inference speed is fast.
• The generated code is functional and runs without issues.

10. Model Inference and Performance Observations

• The AI model used seems to vary, sometimes resembling Sonnet and other times Quen 3 coder.
• Tool calling errors occasionally occur, suggesting the use of Quen 3 coder underneath.
• The performance of the underlying model has been improved with Alibaba's techniques.

11. Demo: Minecraft Clone with Quest Mode

• Quest mode is used to create a Minecraft clone.
• The "write spec first" option is selected, prompting the AI to create a plan for approval.
• A Git repository is required for the task to run correctly.
• The AI generates a to-do list and executes tasks automatically.
• Users can review the task report and changes made after completion.

12. Conclusion and Final Thoughts

• The lack of model selection is a drawback.
• The tool works surprisingly well, especially considering it's free.
• Users are encouraged to try Coder and share their thoughts.

Notable Quotes:

• "Coder is basically what Curser or Trey or Windsurf or Kirro is, but this is much newer and lacks a lot of features as of now."
• "Quest is basically like jewels or codecs. It allows you to give it a task and then let it run in the background."
• "Byte Rover is basically a memory layer that allows you to create memories and those memories can be accessed by your AI coder anytime."

Key Concepts Explained:

• AI Code Editor: A software tool that uses artificial intelligence to assist developers in writing, editing, and debugging code.
• Quest Mode: An asynchronous task execution feature that allows users to assign coding tasks to the AI and have them run in the background.
• Byte Rover: A memory layer that stores and manages memories for AI coders, enabling them to retain context and improve performance over time.
• MCP (Memory, Context, and Persistence) Tools: Tools that provide memory, context, and persistence capabilities to AI coders, allowing them to remember previous interactions and maintain state across sessions.
• Codebase Indexing: The process of creating an index of the codebase to enable efficient search and retrieval of code elements.
• Repo Wiki: An automatically generated documentation system for the codebase, designed to improve code maintainability and onboarding.

Synthesis/Conclusion:

Coder by Alibaba is a new, free AI code editor with features like asynchronous task execution (Quest mode), codebase indexing, and automated documentation (Repo Wiki). While it lacks model selection and has some performance issues, it offers a functional and surprisingly effective coding experience, especially when combined with a memory layer like Byte Rover. Its integration with MCP tools and asynchronous task execution make it a potentially valuable tool for developers.