Kiro

Kiro and JFrog are bringing software supply chain governance to spec-driven development. Packaged as a Kiro power, the plugin will bundle JFrog Platform Skills, MCP tools, and steering guidance into a single one-click install that activates when a task calls for it and stands down when it does not.

Kiro Plugin Features

Packaged as a Kiro Power, Installed in One Click

Powers bundle MCP tools, skills, and steering into a single installable package. Installing the JFrog power registers it automatically, with no JSON configuration files and no command-line setup, and Kiro shows a visual indicator when it is active in the editor.

Built on the Agent Plugins Format

Kiro recommends the Agent Plugins format for new powers — a plugin manifest, a skills directory, and an MCP configuration. That is the same standard behind JFrog’s existing agent plugins, so Platform Skills authored once in a shared repository carry across agents rather than being rebuilt per tool.

Loads Only When the Task Calls For It

Kiro activates powers dynamically from conversation context, so JFrog tool definitions are not consuming the context window during unrelated work. Mention a dependency, build, or vulnerability and the power activates; move to unrelated code, and it stands down.

Package Safety Checks and Curation-Aware Resolution

Before a dependency is pulled, the package safety skill checks it against your curation policy and resolves it through Artifactory rather than a public registry — so an implementation plan that Kiro executes autonomously cannot quietly introduce a component your policy would block.

Governed MCP and Power Discovery

Kiro cautions that powers are third-party tools it has not screened, and that users should install only from trusted sources. Routing discovery through the JFrog AI Catalog and Agent Guard restricts agents to servers and skills the platform team has approved, with credentials held by the platform.

About Kiro

Kiro is an agentic IDE and command-line interface that works alongside developers from prototype to production through spec-driven development. It turns prompts into executable specs, validates code correctness, and builds across large codebases using steering files, agent hooks, and powers.