Kiro Release Notes

Note: To get the latest 1.0.x release (1.0.89 or later), download it directly from kiro.dev/downloads. Auto-updates are paused at this time and we are working on re-enabling them.

This release brings performance improvements and Agent Focus Mode polish. Sessions restore automatically on launch, idle resource consumption is lower, prepaid credit packs appear in the account dashboard, and several fixes ensure permissions, hooks, and steering behave correctly across windows and custom agent profiles.

This release adds dedicated commands for managing OAuth authentication on remote MCP servers, so you no longer need to restart sessions or manually clear credentials when auth flows get stuck. The /usage display also gets a refresh to reflect the new prepaid credits model.

MCP Authentication Management

Three new commands give you full control over OAuth credentials for remote MCP servers. /mcp auth forces re-authentication when a token has expired or become invalid, /mcp cancel-auth aborts a pending auth flow that's stuck waiting for browser confirmation, and /mcp logout removes stored credentials entirely. Keyboard shortcuts are also available in the MCP panel status view: ^A to force auth, ^X to abort, and ^R to remove credentials.

Learn more ->

Individual users on paid plans (Pro, Pro+, Pro Max, and Power) can now purchase add-on credits to continue working after exhausting their monthly allocation. Add-on credits replace the previous overage model with a prepaid approach that gives you full control over spending.

Here's how it works:

• $0.04 per credit, billed immediately when you buy a pack
• Pack sizes range from $5 (125 credits) to $100 per pack
• Up to 5 packs at a time - earliest-expiring pack is consumed first
• 12-month expiry - unused credits roll over month to month until used or expired
• Plan credits first - Kiro uses your monthly allocation before touching add-on credits

Purchase from the Kiro portal via the Buy add-on credits button.

Learn more ->

This release gives the sandbox direct access to your AWS resources and streamlines connecting Powers to services like Figma, Stripe, and Supabase.

IAM Role For Sandbox

Configure an IAM role that Kiro Web assumes on your behalf when running tasks. The sandbox receives short-lived credentials that the agent, CLI tools, and MCP servers can all use to interact with your AWS resources. Credentials refresh automatically while the task runs and are removed when it completes. Set up the role in Settings > Agent > Sandbox > AWS Credentials.

Learn more ->

Authorize Powers for Third-Party services

Powers that connect to third-party services now use a standard authorization flow. Authorize once from Manage Powers in your Agent settings, and the agent uses the connection for all future tasks without prompting again. Disconnect at any time to immediately revoke access. Tokens are encrypted at rest and never exposed to the sandbox or the agent.

Learn more ->

Claude Sonnet 5 is now available in the Kiro IDE, CLI, and Web. It's Anthropic's most agentic Sonnet model yet, a substantial upgrade from Sonnet 4.6 with stronger reasoning, tool use, and coding that approaches Opus 4.8 performance at Sonnet-class pricing. Plans before it edits, runs longer without supervision, and checks its own output without being asked. Sonnet 5 is rolling out gradually with experimental support to Kiro Pro, Pro+, Pro Max, and Power customers in the AWS US-East-1 (Northern Virginia) and AWS Europe (Frankfurt) regions, with cross-region inference support. It ships with the full 1M context window, 1.3x credit multiplier, the same as Sonnet 4.6. Restart your IDE or CLI to access it from the model selector.

This release removes friction from MCP iteration and gives custom agent authors tighter control over context. Agent and MCP configs now reconcile live when you save a change, without restarting the session or losing conversation history. A new setting also lets custom agents opt out of inheriting default resources.

MCP and Agent Config Hot-Reload

Agent and MCP configurations now hot-reload when you save changes on disk. Edit an agent config, add a new agent file, or modify mcp.json and the changes take effect without restarting the session. Only affected servers restart, and your conversation context is preserved. The config diff is order-independent, so reordering environment variables won't trigger unnecessary restarts. Learn more ->

Custom Agent Resource Inheritance

The new chat.disableInheritingDefaultResources setting stops custom agents from inheriting default steering, skills, and AGENTS.md. Set it to true when you run only custom agents and don't want global resources pulled in. Since v2.7.0, custom agents inherit default resources automatically. This setting gives you a way to turn that off. Learn more ->

Agent Focus (Experimental)

Direct multiple agents from a chat-first layout designed for developers who spend their day guiding agents. Launch sessions that work independently and in parallel, see status at a glance, and watch file changes appear as inline diffs. Start with structured workflows (Spec, Plan, Bug Fix, Quick Spec) or freeform chat. Toggle between Agent Focus and the traditional IDE from the top-right corner — your work carries over in both directions. Learn more ->

Permissions

Define exactly what the agent can and cannot do. A capability-based permissions system evaluates every file read, command execution, and MCP call against your rules. The IDE prompts for consent on anything you haven't pre-approved, and you can persist decisions as rules scoped to a workspace or globally. No configuration required to start — defaults apply, and your policy grows as you make consent decisions. Learn more ->

Custom Agents

Build specialized agents in minutes. Write a Markdown file, declare tool access with simple tags (read, write, shell, web), embed MCP servers and permission rules inline, and share the file with your team via version control. Your agent appears in the selector the moment you save it. Switch agents mid-session without losing conversation history. Learn more ->

Hooks

Describe the automation you want in natural language and Kiro generates the hook configuration. Hooks use a structured v1 JSON format stored in .kiro/hooks/ with triggers for file events, tool use interception, spec tasks, and prompt submission. Legacy hooks from 0.x can be migrated directly from the Agent Hooks panel. Manual hooks have been replaced by manual steering files. Learn more ->

Dockable Chat

Open any chat session as a full-width editor tab for more space to read and review long conversations. Right-click a chat tab and select Open in Editor to dock it. Split horizontally or vertically alongside source files, drag between editor groups, or use a multi-monitor setup with chat on a separate screen. The editor tab and panel stay in sync — messages sent from either reach the same session. Learn more ->

Agent Selection

Switch between agents within a chat session without losing conversation history. The selector sits in the chat input bar alongside the model picker — go from Default to a custom code reviewer to Spec mode in a single conversation. Set a preferred default agent that persists across restarts so new sessions always start the way you want. Learn more ->

You can now schedule recurring work in Kiro on the web. Create an automation, describe the task, select repositories from GitHub or GitLab, and set a schedule. Kiro runs it on that cadence in its own sandbox and opens a pull request when done.

Go to Automations in Kiro on the web to set up scheduled tasks. Add up to five schedules per automation using built-in options (hourly, daily) or a cron expression. Each run spins up an independent sandbox, clones your repositories, executes the work autonomously, and opens a pull request with the changes. Edit, disable, or delete automations at any time. Changes apply to the next scheduled run. Learn more ->

This release introduces an early access preview of Kiro CLI V3. The new harness runs alongside your existing 2.x install, so you can try it without disrupting your current setup. Run kiro-cli --v3 to opt in.

Kiro CLI 3.0 is built on the same unified agent harness that powers the Kiro IDE and Kiro Web. Every improvement to the harness now ships to all surfaces simultaneously. V3 introduces spec-driven development in the terminal, a capability-based permissions model, enhanced hooks with a standalone file format, and tag-based agent configuration. Your existing 2.x setup is unchanged until you opt in with kiro-cli --v3.

This release gives you more control over the agent's direction and autonomy. Queue steering lets you redirect the agent at tool boundaries without waiting for it to finish, /goal creates iterative loops that verify completion before stopping, and the /rewind picker now shows exactly what happened at each turn.

/goal command

Start a goal-driven loop where the agent works iteratively toward an objective and must verify the task is complete before finishing. Unlike a regular prompt, /goal keeps the agent running through multiple cycles of implementation and self-check (default: 5 iterations, configurable with --max) until the acceptance criteria are met. Useful for multi-step tasks where you want autonomous execution with a built-in quality gate.

Queue Steering

Send messages to the agent while it's working, and it picks them up at the next tool boundary. Instead of cancelling and restarting when you spot a wrong turn, queue a correction and the agent adjusts course mid-turn. Press Ctrl+S to toggle between steer mode (inject mid-turn) and queue mode (buffer until turn ends).

Enriched /rewind preview

The /rewind turn picker now shows a high-level summary of each turn — tool calls, files touched, commands run, and any intermediate messages so you can quickly identify the right branch point without expanding individual messages.

This release expands Kiro Web beyond GitHub with full GitLab support and brings the Specs workflow to the browser. Connect GitLab repos with a personal access token, and plan work as reviewable requirements, design, and task files before the agent writes a line of code.

Connect your GitLab repositories

Work on GitLab repositories the same way you work on GitHub. The agent clones your repos, writes code, and opens merge requests on your behalf. Connect with a personal access token, inspect existing merge requests and review issues without leaving the chat, and mix GitLab and GitHub repos in a single session.

Specs in the browser

Describe what you want to build, fix, or plan, and Kiro generates requirements, design, and task files for you to review before work begins. Edit the plan through chat, run every task or just the ones you select, and download the artifacts when you're done. A single spec session can span multiple repos, with the agent planning across all of them.

This release focuses on portability and getting back to work faster. Export any conversation to share or archive, label terminal windows so you know which session is where, and start a session at the reasoning level you need. Your model and effort choices now stick around on their own, so there's no extra step to make a preference the default.

Transcript Export

Save a conversation to share with a teammate, attach to a ticket, or feed into another tool. Run /transcript save to export the full chat as markdown, plaintext, or JSON. The --plain and --json flags also switch the rendered transcript format inline, so you can read it the way you want without leaving the pager.

Terminal Window Titles

Keep track of which session is running where. Use /title to set or clear the terminal window title so each window has a meaningful label. Turn it on from /settings display → Terminal title.

Set Effort at Launch

Start a session at the reasoning depth the task calls for. The new --effort flag for kiro-cli chat sets the initial effort level (low, medium, high, xhigh, max) when you launch, so quick lookups stay fast and complex work gets the deeper reasoning it needs from the first prompt.

Persistent Model and Effort Preferences

/model and /effort now remember your choices automatically. Switch models or change effort once and Kiro carries that preference into future sessions, so you no longer need to run /model set-current-as-default to make a selection stick.