AWS Kiro powers up software development with dynamic AI expertise

In a move that might change how developers interact with AI coding tools, Amazon Web Services (AWS) has rolled out a powerful new upgrade to its Kiro IDE: “Kiro powers.” This new feature enables software developers to equip Kiro, the AI agent inside the development environment, with deep, context-sensitive knowledge tailored to specific workflows. Whether you're working on user interfaces, APIs, back-end systems, or cloud deployments, Kiro powers bring relevant tools and best practices into the agent’s brain precisely when they’re needed, and not a moment before.

Why Kiro needed powers

As AI agents take on more responsibility in software development, developers face a familiar tension. The more tools you load into the agent's context, the faster you burn through memory and tokens. This becomes expensive, inefficient, and bloa

In a move that might change how developers interact with AI coding tools, Amazon Web Services (AWS) has rolled out a powerful new upgrade to its Kiro IDE: “Kiro powers.” This new feature enables software developers to equip Kiro, the AI agent inside the development environment, with deep, context-sensitive knowledge tailored to specific workflows. Whether you're working on user interfaces, APIs, back-end systems, or cloud deployments, Kiro powers bring relevant tools and best practices into the agent’s brain precisely when they’re needed, and not a moment before.

Why Kiro needed powers

As AI agents take on more responsibility in software development, developers face a familiar tension. The more tools you load into the agent's context, the faster you burn through memory and tokens. This becomes expensive, inefficient, and bloated. On the other hand, limiting tool access makes the AI less capable in real-world tasks.

Kiro powers solve this problem through dynamic loading. Instead of dumping an entire toolbox into the agent at once, Kiro loads only the tools relevant to the task at hand. Need to work with Stripe APIs for payment? Just mention “checkout” and the Stripe power activates. Done with that? The power unloads, making room for the next task. This system keeps the context clean and focused while giving the agent precision memory and sharper reasoning.

Inside a power: what makes Kiro tick smarter

Each Kiro power is essentially a well-organized bundle. At the heart is POWER.md, a steering file that acts as a how-to manual for the AI agent, telling it what it can do, when to do it, and how. Add in the MCP (Model Context Protocol) server configuration, which serves as a universal protocol for tool access, and you have a recipe for contextual intelligence.

Hooks and additional steering files give the agent reactive abilities. It can run specific actions when something changes in the IDE or when a slash command is used. The beauty lies in the modularity. A developer can plug in just what’s needed and unplug it when it’s not.

Expertise on tap, across the development lifecycle

Kiro powers span the full software lifecycle, from design and development to deployment and observability. Today’s selection includes integrations with developer favorites like Datadog, Dynatrace, Figma, Neon, Netlify, Postman, Stripe, Supabase, and AWS itself.

The setup is frictionless. Developers can browse and install powers either from the IDE or via the kiro.dev website. Each installation auto-registers and configures the power without requiring any additional tinkering. For those looking to go custom, there's also the option to build and share their own powers, either privately within a team or publicly through GitHub.

Why it matters: the bigger picture of agentic coding

AI agents are no longer just autocomplete tools. They are moving toward being reliable development companions that understand projects, adapt to tasks, and respond with contextual precision. Kiro powers turn the AI into something closer to a senior engineer: aware of best practices, tool-savvy, and quick to react.

Until now, the promise of agentic coding was hindered by context overload. Traditional AI setups often required developers to decide upfront which tools to load, which workflows to activate, and how much memory to dedicate. This slowed things down and reduced the utility of AI in live development.

By contrast, Kiro’s approach is refreshingly fluid. Mentioning a database automatically activates Neon configurations. Want observability? Pull in Datadog. Need front-end magic? Load Figma. Each request becomes a trigger to summon the right capability. Once done, the memory clears the desk for the next round.

What’s next for Kiro and its powers

Currently, Kiro powers work only within the Kiro IDE. But AWS has plans to broaden their reach across other development tools. Think Kiro CLI, Cline, Cursor, Claude Code, and beyond. The foundation is already there. MCP servers act as the common interface, so once the powers are built, extending them to other environments is largely a matter of compatibility and UI support.

With more partners and open-source contributions expected in the coming months, the Kiro ecosystem is poised to grow. If the vision holds, we might soon see a shift from static IDE configurations to dynamic, AI-enhanced environments that evolve based on real-time developer intent.

A quiet revolution in how code gets built

Kiro powers are not flashy. But under the hood, they represent a major leap forward in making AI agents practical, modular, and intelligent. They turn the Kiro IDE into something more than just a tool. It becomes a living, adaptive workspace where expertise flows in and out as needed. For developers navigating complex builds and demanding timelines, this could be the edge they didn’t know they were missing.