Getting Started

Getting Started

Quick Start

Get a budgeted Claude running on AWS in five commands. Assumes AWS SSO is already set up; see the Operator Guide for full prerequisites.

# Install
go install github.com/whereiskurt/klanker-maker/cmd/km@latest

# 1. One-time platform configuration (domain, account IDs, region)
km configure

# 2. One-time bootstrap: SCP + KMS + artifacts bucket (in management account)
km bootstrap --dry-run=false

# 3. One-time per region: build Lambdas/sidecars, provision shared VPC
km init --region us-east-1

# 4. Health check (20+ checks across all accounts)
km doctor

# 5. Ship a Claude
km create profiles/goose.yaml --alias dev1
km agent run dev1 --prompt "summarize CHANGELOG.md" --wait
km destroy dev1 --yes

Need to manage many at once? km list shows the fleet, km at schedules deferred work, and the Slack #sb-{id} channels give every sandbox a thread you can talk to.

# Spin up 5 worker sandboxes from one profile
for i in 1 2 3 4 5; do km create profiles/goose.yaml --alias worker-$i & done; wait

# Schedule nightly destroys
km at 'every weekday at 11pm' kill worker-1
km at 'every weekday at 11pm' kill worker-2

# Send all of them a prompt at 6am
km at '6am tomorrow' agent run worker-1 --prompt "pull main, run tests, post results" --auto-start

---

Built-in Profiles

Profile	TTL	Network	Budget	Use Case
hardened	4h	eBPF+proxy (both), AWS services only	None	Production-adjacent testing
sealed	1h	Proxy, .anthropic.com + .npmjs.org only	$5 / $10	Minimal egress, short-lived
goose	4h	eBPF+proxy (both), Anthropic + GitHub + npm + PyPI + OpenAI	$0.50 / $1	Goose agent (Block) with Bedrock + MCP
codex	4h	Proxy, OpenAI + GitHub	$2 / $5	OpenAI Codex agent
ao	8h	eBPF+proxy (both), Anthropic + GitHub + npm + OpenAI	$4 / $10	Multi-agent orchestration (Claude + Codex + AO)
learn	2h	eBPF+proxy (both), wide-open TLD suffixes	$2 / $0	Traffic observation for profile generation

---

Substrates

Substrate	How It Works	Cost
EC2 Spot (default)	Shared regional VPC, per-sandbox SG, spot instance, SSM access, sidecar systemd services	~$0.01/hr for t3.medium
EC2 On-Demand	Same as above, guaranteed capacity (required for hibernation)	~$0.04/hr for t3.medium
ECS Fargate Spot	Fargate task with sidecar containers, service discovery	~$0.01/hr for 1 vCPU / 2GB
ECS Fargate	Same as above, guaranteed capacity	~$0.04/hr for 1 vCPU / 2GB
Docker (local)	Docker Compose on local machine, sidecar containers, IAM roles via STS	Free (local compute)

Spot interruption handlers automatically upload artifacts to S3 before instances are reclaimed. EKS substrate is on the roadmap.

---