The new version of Klein introduces the ability to toggle client rules, giving you full control over custom instructions. The new "Client Rules" popover, located right below the chat input, allows you to instantly switch between different rule files, such as the new memory bank feature.

You can easily toggle between rule sets, like the React best practices, without the need for editing prompts or configurations. This feature makes the process of working with your AI assistant even more seamless, allowing you to precisely tailor its behavior to your specific workflow.

Additionally, you can ask Klein to improve your client rules based on feedback, enabling the self-improving AI assistant to create new rules for you if you don't have any set or provide context to define the rules. This flexibility and adaptability ensure that Klein can be precisely aligned with your development needs.

The latest version of Klein introduces several new slash commands that enhance the AI assistant's functionality. One of the standout features is the /new task command, which allows Klein to start a new task while carrying over relevant context. This is particularly useful for branching out new ideas or starting fresh with a clean state.

In addition to the /new task command, Klein now offers a variety of other slash commands that enable quick actions. These commands can be paired with the new "jump to client chat input" feature, which provides a command palette or keybinding to quickly access Klein's responses.

These new slash commands make it even easier to leverage Klein's capabilities, allowing you to streamline your workflow and focus on the task at hand. With the ability to quickly start new tasks, edit previous messages, and access a range of commands, Klein becomes an even more versatile and efficient AI coding assistant.

The latest release of Klein introduces a powerful new feature - message editing and checkpoints. This allows you to edit previous messages by simply double-clicking them. When you do, Klein uses its checkpoint system, which has also received a new and improved UX design, to let you roll your workspace back to a previous state.

This "infinite undo button" enables you to seamlessly explore different approaches without losing any progress. If you take a wrong turn, you can edit, restore, and branch off to try new ideas, all while maintaining the context and history of your previous sessions.

This feature greatly enhances the flexibility and iterative nature of your workflow with Klein. You can now quickly correct mistakes, experiment with alternative solutions, and maintain a cohesive development process, all powered by Klein's advanced checkpoint management system.

The latest release of Klein introduces several new model integrations, further expanding the AI assistant's capabilities. These integrations include:

• OpenAI 3 (GPT-3): Access to the powerful GPT-3 language model, enabling more advanced natural language processing and generation.
• GPT-4 Mini: Integration with the smaller, more efficient GPT-4 model, providing a balance of performance and efficiency.
• Azure Deepseek: Leveraging Microsoft's Deepseek model for enhanced search and retrieval functionality.
• Google Gemini: Seamless integration with Google's Gemini model, allowing for improved URL handling and web-related tasks.
• Quen 3 Models: Support for the latest Quen 3 models, offering increased flexibility and versatility in AI-powered workflows.

These new model integrations empower users to tackle a wider range of development tasks, from natural language processing to web-based operations, all within the Klein AI assistant. By expanding the available AI capabilities, Klein becomes an even more versatile and powerful tool for autonomous coding and task completion.

The standout feature in this release of Klein is the brand new memory bank. This is a structured, markdown-based system that allows Klein to retain and apply context across sessions.

The key benefits of the memory bank are:

1. Persistent Context: The memory bank enables Klein to remember projects by reading documented files in a specific hierarchy. This drastically reduces token expenditure and improves accuracy by maintaining context.

2. Self-Documenting Workflow: The memory bank encourages good documentation habits automatically, without the need for any additional setup.

3. Scalability: The memory bank system scales easily, making it ideal for both simple tasks and complex projects.

To set up the memory bank, you can either use custom instructions or create a new global rule in the Client Rules section. Then, create a "memory bank" folder in your project root and initialize the memory bank using the provided command.

Once set up, you can use commands to load the memory bank for context, initialize it for a new project, and update it based on recent progress. This will significantly improve Klein's performance and maintain consistent context across your coding sessions.