News

Forked Again: AWS's Kiro Is Latest AI Assistant Based on VS Code

• By David Ramel
• 07/21/2025

Amazon Web Services (AWS) last week announced an agentic AI IDE called Kiro, adding to a growing camp of AI coding assistants based on Microsoft's ubiquitous open-source based Visual Studio Code editor.

This marks at least six major offerings in this camp, consisting of: Cursor, Windsurf, Trae, PearAI, Void, and now Kiro. And that's not to mention other tools tightly integrated into VS Code tech, coming in extensions for the editor or in proprietary offerings, including well-known tools like GitHub Copilot, Codeium, Tabnine, Cline, Amazon Q Developer and many more.

Why VS Code for AI Assistants?
The explosion of AI-focused coding assistants built on VS Code's open-source core (Code OSS) is no accident. VS Code has often been named the No. 1 dev tool in many surveys. Its unique position in this new AI-assistant space stems from its permissive MIT license, which invites virtually anyone to use, modify, and even commercialize the codebase with minimal restrictions. This makes it a prime launchpad for innovation, letting companies and independent developers tap into a familiar, cross-platform editor without reinventing the wheel. By leveraging a robust foundation that is both technically sophisticated and freely adaptable, AI toolmakers can focus on delivering cutting-edge features instead of basic editor plumbing--a strategy that explains VS Code's central role in this fast-moving ecosystem.

Looking at Kiro: An Agentic AI IDE from AWS
AWS officially introduced Kiro, an experimental, agentic AI-powered IDE, last week.

Being designed as an agentic IDE means it's capable of autonomous, goal-driven actions beyond simple prompt responses. Developers describe desired outcomes, and Kiro actively investigates, opens, and modifies codebase files to fulfill requests. It connects to the local development environment, interprets intentions, and provides transparent feedback on changes. Kiro leverages advanced AI models, including Anthropic's Claude Sonnet 4.0 and 3.7.

A core aspect of Kiro is its spec-driven development methodology. This structured approach guides the development process through three main phases:

• Requirements: Kiro generates a requirements.md document with user stories and acceptance criteria using EARS (Easy Approach to Requirements Syntax).
• Design: Based on approved requirements, it creates a design.md document, outlining technical architecture with data flow diagrams, TypeScript interfaces, database schemas, and API endpoints.
• Tasks: Kiro breaks down the work into a checklist of coding tasks that build upon each other, providing a clear implementation plan.

Beyond this structured workflow, Kiro introduces Agent Hooks, an event-driven automation framework. These hooks monitor filesystem changes (for example, file creation, saving, deletion) and trigger AI actions in the background, such as security scans, style checks, or full test suites.

Kiro also features Agent Steering, enabling developers to provide persistent project-specific knowledge through markdown files (for example, product.md, tech.md, structure.md) stored in a .kiro/steering/ directory. These files help guide the AI's behavior according to project conventions, architectural decisions, and business context.

The IDE supports multimodal context integration, processing diverse inputs like files, codebases, documentation, images, and terminal output via Model Context Protocol (MCP) servers. This allows Kiro to maintain a comprehensive understanding of the project's current state. It supports capabilities such as writing and refactoring code, explaining code and architecture, and debugging with context. Kiro currently supports Python and JavaScript, with plans for additional languages.

As far as what might be the best AI tool for your situation--in case tools like GitHub Copilot aren't your thing--we dug into the details to compare the six VS Code-forked offerings:

What's Microsoft Doing?
All of the above comes as Microsoft itself is embarked on an effort to turn VS Code into an open-source AI editor.

As reported in June 2025, Microsoft has begun evolving the open-source version of VS Code (Code - OSS) into a transparent AI platform designed specifically to support extensible, local, and cloud-based AI workflows. The company said, "We believe AI experiences can thrive by leveraging the vibrant open-source community--just as VS Code has successfully done over the past decade."

That includes adding structured AI scaffolding to VS Code's open-source project base, making it easier for toolmakers to plug in AI agents, model connectors, and data visualizations directly into the editing surface.

Why It Matters
This effort is especially relevant given the proliferation of AI IDEs built on top of VS Code forks. Rather than compete directly with proprietary offerings like Cursor or Kiro, Microsoft is expanding the foundational capabilities of the editor itself, offering a standard platform that supports both Microsoft's own Copilot tooling and third-party AI systems.

The new AI architecture includes formal support for:

• Agentic interactions and multi-step prompting
• Custom AI pipelines for different model backends
• Pluggable AI workflows using open standards and JSON specs

Microsoft frames this move as a direct response to developer demand for greater transparency and trust in AI tooling. By achieving the first project milestone--opening up the source code for Copilot Chat, including its agent mode logic, prompt engineering, and telemetry handling--Microsoft enables the entire developer community to audit, adapt, and extend how AI interactions are managed within the editor. Developers and extension builders can now see precisely what information is being sent to large language models and how responses are incorporated into coding workflows--addressing long-running concerns about data privacy and black-box automation.

Relation to Kiro and Other Forks
Microsoft's work reinforces the importance of VS Code as the foundation of the modern AI developer stack. While companies like AWS (Kiro), Cursor, and Trae have opted to fork and customize the editor for their agentic workflows, Microsoft's goal is to enhance the open-source core so that AI functionality is first-class without requiring a fork.

That difference matters. Kiro is a purpose-built environment that takes VS Code in a distinct, AWS-integrated direction. Cursor, likewise, uses VS Code OSS as a base but adds proprietary enhancements tied to specific models like GPT-4. In contrast, Microsoft is focusing on extensibility, aiming to serve as a vendor-neutral base layer for any LLM or agent system.

This puts Microsoft in a position to act as infrastructure provider rather than product competitor--a role more similar to Git than GitHub. As the AI IDE landscape continues to fragment into forks and commercial assistants, Microsoft is doubling down on a common platform designed to keep pace with the generative AI era.

Going forward, Microsoft said: "Next, we will carefully refactor the relevant components of the extension into VS Code core. The original GitHub Copilot extension that provides inline completions remains closed source--but in the following months we plan to have that functionality be provided by the open sourced GitHub Copilot Chat extension. Also, we are excited to partner with the open source AI community to make sure our plan covers impactful open source scenarios."

Stay tuned for more in this rapidly evolving space, as the lines between coding, AI, and development environments continue to blur. The future of programming is increasingly about how we interact with intelligent systems, and VS Code remains at the heart of that transformation.