Amazon Web Services Inc. today announced the general availability of Kiro, a new development environment integrated with artificial intelligence agents that’s designed to assist with rapidly building software with natural language prompts.

Kiro sets itself apart from other “vibe coding” options on the market, such as Cursor and Windsurf, by first producing a “spec,” or specifications, from the prompt given by the user. Included in these specifications, Kiro generates clear requirements, structured designs and test validated implementation tasks before getting to work.

According to Amazon, this makes Kiro one of the few agentic coding apps that inherently includes industry best practices in specification planning for software development.

In addition to the software’s general availability, Amazon announced that Kiro is being launched with team support. Software development teams can now sign up with AWS IAM Identity Center, with support for more identity providers coming soon. Admins can manage access, where they can assign subscriptions and turn on overages, monitor costs, set up and control AI tool connectivity, and manage a single bill.

Teams can collaborate on steering files, a method for providing Kiro direction and persistent knowledge about the working environment. Instead of explaining organization coding conventions and personal preferences in every chat, steering files give Kiro a consistent understanding of how to follow established patterns, libraries and standards. They’re stored as easy-to-read markdown files written in natural language, just like any other prompt.

Users can produce steering files that are specific to them, an individual project or team-wide. This enables different layers of personal customization to Kiro’s behavior coupled with additional methods to align with team needs.

Amazon introduced a new integrated development environment, the editing and development software that allows users to view, change and deploy code. The new IDE comes with three new capabilities including property-based tests to measure generated code against the specification, checkpointing allowing developers to “rewind” agent-made modifications and support for multiple base folders.

The company also debuted a command-line interface capability with Kiro CLI. This new offering brings Kiro into the terminal, a line-by-line, text-based way to interact with the development agents. Unlike graphical user interfaces, which rely on visual elements such as icons and menus, a CLI involves typing commands sequentially and receiving text output in a window.

CLIs are highly efficient for advanced users who don’t need the bells and whistles of a fully visual development environment. Working through a terminal also provides a direct way of communicating with the operating system and software — for example, installing necessary components without the need for a mouse.

Google LLC took a similar step earlier this year with the debut of Gemini CLI in June, which brought its AI agent into the developer terminal to streamline coding tasks. The release signaled a broader industry effort to make AI agents accessible from within core developer workflows. Anthropic has since expanded the market with upgrades to Claude Code, while OpenAI advanced its own tooling with updates to the coding-optimized Codex CLI.

Amazon unveiled Kiro in preview during July for download on Apple macOS, Windows and Linux. A free option is available, with subscriptions beginning at $20 per month for Kiro Pro, $40 for Kiro Pro+, and $200 for Kiro Power, each offering successively more AI generation credits.

Alongside today’s general availability announcement, Amazon said startups in most countries may apply to receive one year of free Kiro credits to build their projects.

Image: SiliconANGLE/Microsoft Designer