Design, price, and deploy production-ready AWS & GCP infrastructure through conversational AI
Kiro Power •
InsideOut •
Luther Systems •
InsideOut is a Kiro IDE power that brings AI-powered cloud infrastructure design directly into your editor. Describe what you want to build in plain language, and Riley — your AI infrastructure advisor — guides you through selecting services, configuring them, estimating costs, generating Terraform, and deploying to AWS or GCP.
No authentication or API keys required. Install the power and start designing.
- Design infrastructure conversationally — describe your app, get expert recommendations
- Get real-time cost estimates — see monthly costs as components are added
- Generate Terraform — production-ready, modular code with security best practices
- Deploy with one command — deploy directly to AWS or GCP from the conversation
- Inspect deployments — verify what was actually provisioned in your cloud account
- Compare providers — evaluate AWS vs GCP options side-by-side
| AWS | GCP |
|---|---|
| ECS, EKS, Lambda, Fargate | Cloud Run, GKE, Cloud Functions |
| RDS, DynamoDB, ElastiCache | Cloud SQL, Firestore, Memorystore |
| ALB, CloudFront, Route 53 | Cloud Load Balancing, Cloud CDN |
| S3, EFS | Cloud Storage, Filestore |
| VPC, Security Groups | VPC, Firewall Rules |
| CloudWatch, SNS, SQS | Cloud Monitoring, Pub/Sub |
| And 15+ more | And 10+ more |
- Open Kiro IDE
- Open the Powers panel
- Click Add power from GitHub
- Enter:
luthersystems/insideout-power
- Clone this repo:
git clone https://github.com/luthersystems/insideout-power.git - Open Kiro IDE
- Open the Powers panel
- Click Add power from Local Path
- Select the cloned directory
MCP support must be enabled in Kiro:
- Open Settings (
Cmd + ,/Ctrl + ,) - Search for "MCP"
- Enable the MCP support setting
That's it. No API keys, no local binaries, no additional setup.
Kiro asks for approval before each MCP tool call. InsideOut ships with autoApprove configured for its safe, read-only tools, but Kiro does not currently honor the autoApprove field — this is a known Kiro bug. Until it's fixed, you'll need to click "Allow" for each tool call the first time it's used in a session.
We've pre-configured autoApprove in mcp.json so that once Kiro fixes this, the conversational and monitoring tools will auto-approve and only tfgenerate and tfdeploy (which create or modify real cloud infrastructure) will prompt for confirmation.
Once installed, mention anything about infrastructure, cloud, AWS, GCP, Terraform, or deployment in a Kiro chat. The power activates automatically.
You: "I need to set up cloud infrastructure for a web app"
Kiro: [Activates InsideOut power, calls convoopen]
Riley: "Hi! I'm Riley, your infrastructure advisor. Tell me about the app
you're building — what does it do, who uses it, and what scale
are you planning for?"
You: "It's an e-commerce platform expecting 50k monthly users on AWS"
Riley: "Great! I'd recommend ECS Fargate for your containers, RDS PostgreSQL
for your database, ElastiCache Redis for sessions, and an ALB.
Estimated cost: ~$350/month. Want me to adjust anything?"
You: "Looks good, generate the Terraform"
Kiro: [Calls tfgenerate — downloads production-ready Terraform files]
You: "Deploy it"
Kiro: [Calls tfdeploy — deploys to AWS, streams logs via tflogs]
| Tool | Description |
|---|---|
convoopen |
Start a new infrastructure design session |
convoreply |
Continue the design conversation with Riley |
convoawait |
Wait for long-running operations |
convostatus |
View current components, config, and pricing |
tfgenerate |
Generate production-ready Terraform files |
tfdeploy |
Deploy generated Terraform to AWS or GCP |
tfstatus |
Check deployment progress |
tflogs |
Stream real-time deployment logs |
awsinspect |
Inspect deployed AWS resources |
gcpinspect |
Inspect deployed GCP resources |
help |
Get workflow guidance |
insideout-power/
├── POWER.md # Power metadata, onboarding, and agent instructions
├── mcp.json # MCP server configuration (remote HTTP)
├── README.md # This file
├── LICENSE # Apache 2.0
├── assets/
│ ├── banner.svg # GitHub banner
│ └── logo.svg # InsideOut logo
└── steering/ # Workflow-specific guidance for the agent
├── getting-started.md # First-time setup walkthrough
├── aws-design-patterns.md # AWS architecture patterns and prompts
├── gcp-design-patterns.md # GCP architecture patterns and prompts
└── troubleshooting-guide.md
InsideOut uses a multi-agent AI system behind a single MCP server:
| Agent | Role |
|---|---|
| Riley | Infrastructure advisor — leads the design conversation |
| Hippo | Cost estimation and pricing optimization |
| Joy | User experience and requirement gathering |
| Etch | Terraform code generation |
| Core | Architecture validation and best practices |
| Axel | Deployment orchestration |
The conversation flows through these agents automatically. From Kiro's perspective, you're talking to Riley — the other agents work behind the scenes.
- Fork this repository
- Create a feature branch:
git checkout -b feature/my-improvement - Make your changes
- Test locally: Install the power from local path in Kiro
- Submit a pull request
- Test POWER.md changes by reinstalling the power from local path
- Verify MCP connectivity with the
helptool after changes - Steering files are loaded on-demand — test each workflow independently
- InsideOut Platform
- Demo Video
- Standalone Web App — try InsideOut without Kiro
- Luther Systems
- Kiro IDE
- Kiro Powers Documentation
- MCP Protocol
- Discord — chat with the devs and InsideOut users
- General Inquiry Call — talk with us
- Tech Call — talk with the devs