v0.8.0

KaibanJS Release Notes - Version 0.8.0

We're thrilled to announce the release of Kaiban Board v0.8.0! This version brings powerful new features for local development, easy deployment, and intuitive AI agent management.

Key Features

1. Local Development and Execution

• Run your entire Kaiban Board locally with a single command:

npx kaibanjs@latest init

• Automatic setup of environment variables, development tools, and configuration files
• Local Vite server launched automatically, accessible at http://localhost:5173/

2. Automatic Configuration Generation

• Automatic creation of team.kban.js file in your project root for new users
• Sample configuration provided as a starting point for AI agent setup

3. Multi-Team Support

• Create multiple .kban.js files for different team configurations
• CLI automatically detects and sets up all team configurations in your project

4. One-Click Deployment to Vercel

• Deploy your Kaiban Board to the cloud with a simple command:

npm run kaiban:deploy

• Automatic Vercel integration, including installation if needed
• Production build and deployment handled automatically

5. Visual Kanban-Style Interface for AI Workflows

• Real-time visualization of AI agent tasks and progress
• Intuitive task tracking from "To Do" to "Done"
• Simplifies management of complex AI operations

6. Comprehensive Agent and Task Management

• Define and configure AI agents and tasks using JavaScript
• Real-time preview of agent configurations
• Detailed task views including descriptions, progress, and results

7. Enhanced Security Options

• Flexible API key management for different AI service providers
• Recommendations for securing API keys in production environments

Improvements

• Optimized CLI for seamless local development and deployment
• Enhanced error handling and user feedback during setup and execution
• Improved performance for real-time updates in the Kanban board interface

Getting Started

To experience these exciting new features visit our Quick Start Guide

Why It's a Game-Changer

• 100% Local Development: Run everything on your own machine for rapid iteration and testing
• Deploy with Ease: Share your AI projects with the world in just one command
• Visualize AI Workflows: Watch your AI agents work in real-time with our intuitive interface
• Full Control: Manage how your agents interact, what tools they use, and how they're deployed

We can't wait to see what you'll create with the new Kaiban Board! As always, we welcome your feedback and suggestions for future improvements.

Happy AI development!

The Kaiban Team

v0.7.0

KaibanJS Release Notes - Version 0.7.0

We're excited to announce the release of KaibanJS version 0.7.0, which introduces significant improvements in API key management and LLM provider integration. This release focuses on enhancing security and flexibility for both development and production environments.

Key Features

1. Customizable API URL for LLM Providers

• Introduced a standardized way to define custom API URLs for consuming LLM provider services.
• Allows developers to specify a custom apiBaseUrl for each LLM provider.
• Maintains flexibility for both rapid prototyping and secure production setups.

Example usage:

const agent = new Agent({
   name: 'AI Assistant',
   role: 'Helper',
   llmConfig: {
      provider: "openai",
      apiBaseUrl: "http://localhost:3000/llm/openai",
      model: "gpt-4",
   }
});

2. Kaiban LLM Proxy Integration

• Introduced the Kaiban LLM Proxy, a boilerplate for quickly setting up a secure proxy for LLM providers.
• Supports multiple LLM providers including OpenAI, Anthropic, Google Gemini, and Mistral.
• Enables secure API key management by keeping keys server-side.
• Provides a flexible solution adaptable to various environments and frameworks.

Additional Improvements

• Updated documentation with comprehensive guidelines on API key management in browser-based applications.
• Added a new guide: API Key Management

Benefits

• Enhanced security for API key management in frontend applications.
• Simplified development process, especially for prototyping and testing.
• Increased flexibility in configuring LLM provider interactions.
• Improved compatibility with multiple LLM providers.

Upgrading

To take advantage of these new features, update your KaibanJS dependency to version 0.7.0. Refer to the updated documentation for detailed usage instructions and best practices.

We appreciate your continued support and feedback. Happy coding with KaibanJS 0.7.0!

v0.6.0

KaibanJS 0.6.0 Release Notes

New Feature: Custom LLM Integration

We're excited to announce support for custom LLM integration in KaibanJS, allowing users to integrate their own LLMs or additional providers beyond the currently supported models.

New Supported LLM Providers:

In addition to the existing support for OpenAI, Anthropic, Google, and Mistral, KaibanJS now supports:

• Ollama
• Cohere
• Azure OpenAI
• Groq
• Other custom integrations

For detailed information on supported LLMs and their usage, please refer to our LLMs Documentation.

Usage Example:

import { Agent } from 'kaibanjs';
import { ChatOllama } from "@langchain/ollama";

const agent = new Agent({
    name: 'CustomAgent',
    role: 'Custom LLM User',
    goal: 'Use a custom Ollama instance',
    llmInstance: new ChatOllama({
      model: "llama2",
      temperature: 0.7,
      baseUrl: "http://localhost:11434"
    })
});

For more examples and detailed setup instructions, visit our Custom LLM Integration Guide.

Documentation Structure:

1. Built-in Models:

   • OpenAI
   • Anthropic
   • Google
   • Mistral

   These offer streamlined setup and integration within KaibanJS.

2. Custom Integrations:

   • Ollama
   • Cohere
   • Azure OpenAI
   • Groq
   • Other specialized options

   These provide advanced options for users with specific requirements or preferences.

This release significantly expands the capabilities of KaibanJS, allowing for more diverse and powerful AI integrations. We encourage users to explore the new LLM options and provide feedback on their experiences.

v0.5.0

KaibanJS 0.5.0 Release Notes

We're excited to announce the release of KaibanJS 0.5.0, which brings significant improvements to documentation, tool integration, and agent customization capabilities.

Major Features

1. Comprehensive Tools Documentation

We've significantly expanded our documentation with a new section dedicated to tools. This update includes:

• LangChain Tools: Detailed documentation on tested LangChain tools that work seamlessly with KaibanJS.
• Custom Tools: Comprehensive guides on creating and implementing your own tools.
• Examples and Demos: Real-world examples demonstrating the benefits of tool integration, available on the Kaiban Board.
• Community Focus: We encourage community input on tool creation and suggestions for new tools.

Explore the new Tools Documentation

Watch our brief overview video

2. Custom Agent Prompt Customization

Developers can now customize internal prompts used by agents, allowing for more tailored interactions and improved contextual responses. This feature enables:

• Greater flexibility in agent behavior
• More precise control over agent interactions
• Broader applicability across diverse scenarios and industries

Read about custom agent prompts

Improvements

• Expanded testing framework for Langchain Tools integration
• New examples and tutorials in the documentation
• Improved developer experience with clearer guides and API references

Community Contributions

We'd like to thank our community for their valuable feedback and contributions, which have helped shape this release. We encourage you to:

• Explore the new tools documentation
• Try out the examples on the Kaiban Board
• Share your experiences with creating custom tools
• Suggest tools you'd like to see in future updates

Looking Ahead

We're committed to continually improving KaibanJS based on your needs and feedback. Stay tuned for more updates and new features in upcoming releases.

v0.4.0

KaibanJS 0.4.0 Release Notes

We're excited to announce the release of KaibanJS 0.4.0, which includes several improvements to documentation, user experience, and core functionality. This release is part of our 0.4.0 milestone.

Documentation Enhancements

1. Streamlined "Get Started" Section: The documentation now features a more comprehensive and user-friendly "Get Started" section, including detailed examples for both React and Node.js environments. This update aims to reduce the learning curve and improve the overall user experience.

2. Updated Discord Community Link: The documentation now correctly links to the KaibanJS Discord community, ensuring users can easily connect with the right support channels.

3. Mobile-Friendly Documentation Menu: A new mobile-friendly menu has been implemented to improve navigation on smaller screens, making it easier for users to access different sections of the documentation on mobile devices.

4. Search Feature: A new search functionality has been added to the documentation, allowing users to quickly find specific topics or sections.

User Experience Improvements

1. Mobile Playground Message: The mobile version of the Playground now displays a message encouraging users to switch to a desktop for the best experience, setting clear expectations for mobile users.

2. Website Code Examples Fix: All code examples on the main website have been updated to correctly import from 'kaibanjs' instead of 'AgenticJS', ensuring consistency and preventing confusion for new users.

Core Functionality Update

1. Enhanced team.start() Method: The team.start() method now returns comprehensive workflow statistics along with the workflow status and result. This update simplifies the process of retrieving workflow data and ensures consistency between the workflow result and associated statistics.

We'd like to thank all contributors and users who helped identify these issues and suggest improvements. Your feedback is invaluable in making KaibanJS better with each release.

For more details on each change, please refer to the linked issue numbers and our 0.4.0 milestone. As always, we encourage users to update to this latest version to benefit from these improvements and ensure compatibility with future updates.

To get started with the latest version, check out our updated Getting Started guide.

v0.3.0

KaibanJS v0.3.0 Release Notes

We’re excited to announce the release of KaibanJS v0.3.0, which marks a significant update as we officially transition from AgenticJS to KaibanJS. This release also includes enhanced tool integration capabilities in the KaibanJS Playground.

🚨 Name Change to KaibanJS

After careful consideration, we’ve renamed the project from AgenticJS to KaibanJS to avoid potential confusion and better reflect the unique value of our framework—AI agents within a Kanban-inspired system.

What’s New:

• Repository Rename: The repository has been renamed to KaibanJS and transferred to the new Kaiban organization.
• Documentation Updates: All related documentation, links, and resources have been updated to reflect this name change.
• Domain and Branding Changes: Our landing page and associated domains now align with the KaibanJS branding.

🔗 New Links

Here are all the new links associated with KaibanJS:

• 🌐 Website
• 📚 Official Documentation
• 💻 GitHub Repository
• 📦 NPM Package
• 📝 Project Backlog
• 🤝 Join the Community

New Features

Expanded Tool Integration in KaibanJS Playground

• Custom Tool Selection: Users can now select and integrate various tools within the KaibanJS Playground, similar to the flexibility provided in the local setup.
• Improved Playground Experience: This update enhances the playground by allowing broader tool integration, making it more useful for demonstrations and educational purposes.

List of available tools here: https://docs.kaibanjs.com/category/tools-docs

Improvements

• Updated Documentation: The documentation has been fully updated to reflect the new project name and the enhanced tool integration features.
• Branding and Domain Updates: Our website and all associated domains have been rebranded to match the KaibanJS identity.

Breaking Changes

• Repository Transfer: The repository has been transferred to the new Kaiban organization. Please update any references to point to the new location.

Feedback and Contributions

If you encounter any issues or have feedback, please let us know through our GitHub repository or join the discussion in our community channels. We’re here to help and appreciate your input.

v0.2.0

AgenticJS v0.2.0 Release Notes

We're excited to announce the release of AgenticJS v0.2.0, which introduces significant enhancements to our AI agent framework, particularly in the areas of Human-in-the-Loop (HITL) functionality and improved observability.

New Features

Human-in-the-Loop (HITL) Functionality

Human in the Loop (HITL) enables manual intervention and oversight in AI-driven tasks. It provides a mechanism for human operators to review, provide input, or make decisions at critical points in the workflow, ensuring accuracy, reliability, and ethical considerations in complex or sensitive operations.

The HITL feature addresses the need for human judgment in scenarios where:

• Tasks reach a complexity level beyond AI capabilities
• Sensitive decisions require human oversight
• Quality assurance is critical
• Ethical considerations necessitate human involvement

For detailed information on how to implement and utilize these features, please refer to our Human-in-the-Loop documentation.

TypeScript Support

AgenticJS now includes TypeScript support for its public APIs, enhancing type safety and improving the development experience. For a quick start and example usage, check out the NodeJS TypeScript example.

Big thanks to @alienkarma for tirelessly working on this enhancement!

Improvements

• Renamed executeTask to workOnTask for clarity across the codebase
• Added workOnFeedback method to handle feedback processing
• Implemented initialize method for agents to set up API keys and configurations
• Enhanced the Team class with improved workflow control and task management
• Added functionality to derive context dynamically from workflow logs
• Updated the React example to include HITL-related UI elements and functionality
• Improved error handling in workflow processes

Documentation Updates

• Enhanced the overall design of the documentation, including updates to color schemes, fonts, and layout for improved readability and visual appeal
• Updated State Management documentation to reflect new HITL features and observability improvements
• Added a new Use Cases section to the documentation, showcasing practical applications of AgenticJS in real-world scenarios

Breaking Changes

• The executeTask method has been renamed to workOnTask. Please update your code accordingly.

Bug Fixes

• Various minor bug fixes and performance improvements

We're excited about these new features and improvements, and we look forward to seeing how you'll use them in your projects. The introduction of HITL functionality and expanded tool integration capabilities significantly enhances the versatility and reliability of AgenticJS.

For a complete list of issues addressed in this release, please refer to our v0.2.0 milestone on GitHub.

As always, we welcome your feedback and contributions to make AgenticJS even better. Thank you for your continued support!

v0.1.0

AgenticJS Version 0.1.0 - Release Notes

We are thrilled to introduce the first beta release of AgenticJS, version 0.1.0! AgenticJS is crafted exclusively for JavaScript developers, offering a robust framework to build and manage multi-agent AI systems. This release marks a significant step towards empowering JavaScript developers with AI capabilities without the need to shift to other programming languages.

What's New

• Role-Based Agent Design: Agents can now be customized with specific roles and objectives, enhancing their ability to handle tasks effectively.

• Tool Integration: Enable your AI agents to utilize specialized tools to perform tasks with greater accuracy and efficiency.

• Multiple LLMs Support: The framework now supports integration with multiple AI models, broadening its adaptability and use cases.

• Robust State Management: Incorporates a Redux-inspired state management architecture, ensuring consistent and reliable interactions across AI agents.

• Observability and Monitoring: Track every state change with detailed stats and logs, providing full transparency and control.

• Real-Time Agentic Kanban Board: A built-in visualizer is now available, allowing you to observe and manage the workflow of agents in real-time.

Key Resources

• Landing Page: AgenticJS
• Official Documentation: Documentation
• GitHub Repository: Repo
• NPM Package: npm
• Support and Community: Join our Discord

Known Issues

As this is our first beta release, you may encounter some bugs and areas for improvement. We invite the community to help us improve by reporting any issues they find.

How to Contribute

We encourage contributions from the community. Please read the Contribution Guidelines for more details on how to contribute.

Upcoming Features

For a detailed look at what's coming next for AgenticJS, check out our current Backlog on GitHub. We’re constantly updating it with new features and enhancements based on community feedback and our ongoing development efforts.

Thank you for your interest in AgenticJS! Your feedback and contributions are crucial as we continue to develop a leading-edge AI framework for JavaScript developers.