See your real ECS fleet in
5 minutes. Locally. Read-only.
Nothing leaves your machine.
Download a single .md file — a universal AI agent skill that connects to your AWS account with read-only credentials, scans your ECS clusters, and generates a visual fleet report + ready-to-import Fortem config. Runs in Claude Code, Codex, OpenCode, or any CLI agent. No account required. No data sent anywhere.
How it works
Download
One file — fortem-onboarding.md. About 200 lines of plain text. Open it in your AI agent. Read every line before you run it — we want you to.
Run
Open the skill in Claude Code, Codex, OpenCode, or any CLI agent. It connects to your AWS account with the read-only credentials you configure. Scans ECS clusters, services, task definitions, and tags. Calculates costs from on-demand Fargate pricing. Optionally reads your local Terraform. Asks you ~5 clarifying questions — which environments are production, what timezone.
Get your fleet
Two files land on your machine: your Fortem environment config (ready to review and import) and an HTML fleet report with real costs, scheduling savings estimates, and an environment map. Everything stays local.
Takes 5–15 minutes. Works on any AWS account with ECS clusters. Terraform optional.
Permissions the skill uses
These six read-only permissions are already declared inside the .md file. Shown here so you can inspect before running. Nothing is created, modified, or deleted.
{
"Version": "2012-10-17",
"Statement": [
{
"Effect": "Allow",
"Action": [
"ecs:ListClusters",
"ecs:DescribeClusters",
"ecs:ListServices",
"ecs:DescribeServices",
"ecs:DescribeTaskDefinition",
"ecs:ListTagsForResource"
],
"Resource": "*"
}
]
}Optional: add rds:DescribeDBInstances and elasticache:DescribeCacheClusters for richer discovery.
What you get
The skill generates your environment config locally. Here's the Fortem console you'd be using — with your data — after onboarding. Same interface every day.
Fleet Dashboard
Every environment, color-coded by stage and region. Service health at a glance (green = running, red = stopped). Database pills, CI/CD branches, schedule status, and live cost per environment. Same interface your team uses daily.
| ENV | TEAM | STATUS | SVCS | SCHED | COST |
|---|---|---|---|---|---|
use1-prod-main us-east-1 | Platform | Running | — | $2.4k | |
use1-stag-main us-east-1 | Platform | Running | Mon-Fri 8-22 | $1.9k | |
use1-dev-qa1 us-east-1 | Backend | Running | Mon-Fri 9-19 | $890 | |
usw2-dev-ml1 us-west-2 | ML | Running | Mon-Fri 9-18 | $350 | |
euw2-dev-demo eu-west-2 | Frontend | Scheduled | Mon-Fri 9-19 | — |
AI Ops — diagnostics that fix things
Your fleet comes with real diagnostics. Fortem watches your environments and proposes specific fixes — idle services costing money, task failures with exact root causes, over-provisioned CPU. You approve. It applies.
No deploys, no traffic, no scale events in 8 days. Cost: $11.60/day running 24/7.
ECS task stopped repeatedly. ResourceInitializationError: unable to pull image. ECR repo policy missing cross-account pull.
Task definition: 1024 CPU. CloudWatch p95 over 14 days: 87 CPU. Right-size to 256 saves $54/mo across services.
Environment Config · YAML
Your environments mapped into Fortem's format — ready to import. A starting point. A Fortem engineer adjusts it with you on the onboarding call.
# Generated by Fortem AI Onboarding — review before importing
environments:
- id: use1-prod-main
stage: prod
cluster_arn: "arn:aws:ecs:us-east-1:123456789:cluster/main"
schedule: null # production — never scheduled
estimated_cost_mo: 2400
services: 12
- id: use1-dev-qa1
stage: dev
schedule:
suggested: weekdays-9-19
timezone: America/New_York
estimated_cost_mo: 890
estimated_savings_mo: 623
services: 8
Your data never leaves your machine
Platform engineers look for security before anything else. Here's exactly how the skill handles your infrastructure data.
Read-only access
The skill uses only ecs:List* and ecs:Describe* permissions. It cannot create, modify, or delete any resource. Ever.
Nothing transmitted
The skill runs entirely inside your AI agent environment. No AWS credentials, resource names, account IDs, or cost data go anywhere. Not to Fortem. Not to Anthropic. Not to OpenAI.
Open file — read it yourself
It's a plain text .md file, ~200 lines. Read every line before you run it. No compiled binaries, no obfuscated scripts. We want you to audit it.
You set the scope
The skill works with just 6 ECS read permissions. Want richer output? Optionally add RDS and ElastiCache read access. You decide how much is appropriate.
The generated HTML report contains your infrastructure data. Treat it like internal documentation — share it with the same caution you'd share a Terraform state file.
What happens after
Run the skill
Get your fleet report — your clusters, your costs, your savings estimate. All local.
Book 20 minutes
Bring the report. A Fortem engineer walks through it with you. Not a sales pitch — just 'does this make sense for your setup?'
If it fits
Onboarding takes 7 business days. Your config is already half-written. Your team sees the real console in the first call.
Common questions
Start with your actual infrastructure
Every other tool shows you a demo. This one maps your real ECS fleet — your environments, your costs, your savings — before you talk to anyone. 5 minutes. Locally. No strings attached.
No signup, no credit card. Read-only access. Nothing leaves your machine.