What is the difference between Kiro and Claude Code?
Kiro is an IDE that turns a feature request into written specs before its agent codes, while Claude Code is an agent you give tasks to from the terminal, your editor, a desktop app or the browser. Kiro is built on Code OSS, the open source base of VS Code, and works through requirements, design and tasks files. Claude Code reads your codebase, edits files and runs commands, and you steer it with plan mode and a CLAUDE.md file.
Both now have a terminal agent, hooks, MCP and Agent Skills, so the real split is workflow and models. Kiro puts a reviewed spec at the center and lets you pick models from several vendors. Claude Code puts the conversation at the center and uses Anthropic's Claude models, directly or through a cloud provider or gateway. Everything below comes from the Kiro docs and the Claude Code docs, read on 27 October 2026.
How does Kiro work?
The heart of Kiro is the spec. Per the specs docs, "Every spec generates three key files that form the foundation of your specification":
- requirements.md (or bugfix.md for bug specs): user stories and acceptance criteria. Requirements use EARS notation, such as "WHEN a user submits a form with invalid data THE SYSTEM SHALL display validation errors next to the relevant fields."
- design.md: architecture, sequence diagrams and implementation notes.
- tasks.md: a trackable plan. When you run all tasks, independent tasks run in waves: "Waves execute sequentially; tasks within a wave execute concurrently."
The files land in .kiro/specs/<name>/, so you can commit and review them like code. A Quick Spec writes all three in one pass with no approval gates.
Around specs, Kiro adds:
- Steering files in
.kiro/steering/with inclusion modesalways,fileMatch,manualandauto. Per the steering docs, Kiro can generateproduct.md,tech.mdandstructure.mdfor you, and "AGENTS.md files do not support inclusion modes and are always included." - Agent hooks that run a shell command or inject an agent prompt on triggers such as
PostFileSave,PreToolUse,PostToolUseandStop. - Kiro CLI, installed with
curl -fsSL https://cli.kiro.dev/install | bashand started withkiro-cli, including a headless mode for CI.
Sign-in uses GitHub, Google, AWS Builder ID or AWS IAM Identity Center, and the FAQ says "You do not need an AWS account to use Kiro." The Kiro guide covers install and setup.
How does Claude Code work?
Anthropic describes Claude Code as "an agentic coding tool that reads your codebase, edits files, runs commands, and integrates with your development tools." It runs in the terminal, VS Code (and Cursor), JetBrains IDEs, a desktop app and the browser at claude.ai/code, and every surface uses the same engine.
curl -fsSL https://claude.ai/install.sh | bash # macOS, Linux, WSL
cd your-project
claude
The closest thing to a Kiro spec is plan mode. Per the permission modes docs, "Plan mode tells Claude to research and propose changes without making them." You enter it with Shift+Tab, a /plan prefix on one prompt, or claude --permission-mode plan. When the plan is ready you approve it, ask for changes, or press Ctrl+G to edit it in your text editor first. The plan mode guide goes deeper.
Beyond that, Claude Code has subagents that work on parts of a task in parallel, hooks that run shell commands before or after its actions, MCP servers, cloud scheduled tasks, and claude -p for scripts and CI, with GitHub Actions and GitLab CI/CD integrations.
Kiro vs Claude Code side by side
| Kiro | Claude Code | |
|---|---|---|
| Maker | AWS | Anthropic |
| What it is | Spec-driven agentic IDE, plus a CLI, web and iOS apps | AI coding agent |
| Main home | Kiro IDE (Code OSS based) | Terminal, plus VS Code, JetBrains, desktop app and web |
| Planning | Specs: requirements.md, design.md, tasks.md in .kiro/specs/ |
Plan mode, approved before edits |
| Models | Auto by default; Claude, GPT-5.6 and open weight models | Claude models, via Anthropic, Bedrock, Agent Platform, Foundry or a gateway |
| Project instructions | .kiro/steering/*.md and AGENTS.md |
CLAUDE.md, or AGENTS.md when there is no CLAUDE.md |
| Agent Skills | .kiro/skills/, global ~/.kiro/skills/ |
.claude/skills/, global ~/.claude/skills/ |
| Hooks | Command or agent prompt on file save, tool use and stop | Shell commands before or after actions |
| Scripting and CI | kiro-cli headless mode |
claude -p, GitHub Actions, GitLab CI/CD |
| Sign-in | GitHub, Google, AWS Builder ID, IAM Identity Center | Claude subscription, Claude Console, or a cloud provider |
| Entry price | Free tier with 50 credits | Claude Pro, $20 a month billed monthly |
How much do Kiro and Claude Code cost?
Kiro. The pricing page lists, as of October 2026:
| Plan | Price | Credits per month |
|---|---|---|
| Free | $0 | 50 (open weight models and Claude Sonnet 4.5) |
| Pro | $20 per user a month | 1,000 |
| Pro+ | $40 per user a month | 2,000 |
| Pro Max | $100 per user a month | 5,000 |
| Power | $200 per user a month | 10,000 |
| Enterprise | Contact sales | Not listed |
Add-on credits on paid plans cost $0.04 each, and monthly plan credits don't roll over.
Claude Code. Claude Code comes with a paid Claude plan. The Claude pricing page lists, as of October 2026:
- Free: no Claude Code.
- Pro: $17 a month billed annually, or $20 billed monthly.
- Max: from $100 a month, in 5x and 20x tiers.
- Team: Standard seats at $20 a month billed annually ($25 monthly), Premium seats at $100 ($125 monthly).
- Enterprise: $20 per seat a month billed annually with a 20 seat minimum, plus usage at API rates.
You can also use Claude Code with a Claude Console account billed per token, or through a supported cloud provider.
How do instructions and skills carry over?
Instructions. Kiro reads AGENTS.md from the workspace root, subfolders and ~/.kiro/steering/, and always includes it. Claude Code reads AGENTS.md in place of a CLAUDE.md when you don't have one. If you do, start it with an import:
@AGENTS.md
## Claude Code only
- Run npm test before saying a task is done.
Keep Kiro-only context, like fileMatch steering for one folder, in .kiro/steering/.
Skills. Both follow the Agent Skills format. Kiro's skills docs require name (matching the folder, lowercase letters, numbers and hyphens, max 64 characters) and description (max 1,024 characters), in .kiro/skills/ or ~/.kiro/skills/. Claude Code's skills docs use .claude/skills/ and ~/.claude/skills/ and say its skills "follow the Agent Skills open standard," with extra fields of its own. A skill that sticks to the standard fields works in both once it sits in each folder. See the Claude Code skills guide.
Specs in Claude Code. If you like Kiro's flow but prefer Claude Code, a skill can recreate it. The spec-driven development skill is one example.
Which should you pick?
Kiro tends to fit when:
- You want requirements and a design reviewed and committed before code is written.
- You want an IDE with the agent built in, and you already know VS Code.
- You want to choose between Claude, GPT and open weight models, or start on a free tier.
- Your team buys through AWS or uses IAM Identity Center.
Claude Code tends to fit when:
- You want to hand off a task and review the diff, with planning as an optional step.
- You work in the terminal, over SSH, or across several editors.
- You want to script the agent in CI or chain it with other command line tools.
- You already pay for a Claude plan.
Many teams use both: Kiro to shape a feature into a spec, Claude Code to work through tasks from the terminal. For Kiro against an AI editor, see Kiro vs Cursor.
Keep your skills and prompts portable
Specs, steering files and skills are worth more than any one tool. Kipt keeps your prompts and Agent Skills as Markdown in a private GitHub repo you own, and installs every skill into Claude Code or Codex with one npx skills add command. For Kiro, copy the folder into .kiro/skills/. The free SKILL.md generator turns a prompt you keep reusing into a valid SKILL.md.