AMAZON INTRODUCED A NEW DEVELOPMENT ENVIRONMENT WITH NEURAL NETWORKS
17.07.2025
On July 14, 2025, the cloud platform Amazon Web Services presented a new environment for software developers — KIRO.
Kiro is an agent-based development environment that introduces new features such as specifications, governance, and hooks.
Specifications are project documents that define what exactly is being developed, how it will be structured, and who will implement it. They help the team understand the development stages and collaborate effectively.
Governance refers to markup files in .md (Markdown) format that set the standards for the project — from variable names to overall architecture — allowing development teams to strictly follow development order and coding standards.
Agent hooks are automation tools that optimize the development process by automatically executing predefined agent actions when certain events occur in the environment. Thanks to hooks, there is no need to manually trigger routine tasks. Hooks help maintain consistency across the entire codebase.
Developers claim that Kiro turns prompts into clear requirements, structured projects, implementation tasks, thoroughly tested results, and code generated by advanced agents. It’s a programming flow powered by neural networks, enhanced with mature engineering practices.
Kiro also uses the Model Context Protocol (MCP) to access specialized knowledge bases and AWS documentation, integrate with external services and APIs, enable new tools focused on specific tasks, and create custom tools for unique workflows.
Kiro features an autopilot mode — a self-operating execution mode that allows the agent to make changes to your codebase and handle complex tasks with minimal user input.
It is currently known that Kiro works with CLAUDE agents, a project in which Amazon is making significant investments.
All major frameworks are confirmed to be supported out of the box, including game development capabilities and a convenient migration path from Visual Studio Code to Kiro.
Video from the developers.
Visit the SAID-Test to practice identifying fake generations.
said correspondent🌐
To discuss, create a topic in the community.