Run an AWS data architecture review pass
Point the AWS Architecture Reviewer at an architecture description or IaC and follow its four-step workflow: map the services to a data flow, check IAM for least privilege, estimate dominant cost drivers, and identify single points of failure. The agent is read-only and never creates resources.
At a glance
| When | A new AWS data architecture needs a cost/IAM/reliability pass before build; An existing platform keeps producing cost surprises or single points of failure |
|---|---|
| Where | AWS |
| Why | Thirty AWS data-platform anti-patterns repeat across teams, and a fixed four-step review pass turns them into findings instead of incidents. |
Learning objectives
- Map an AWS data platform into a single data flow before judging it
- Check IAM roles for least privilege
- Name the dominant cost drivers and single points of failure
What you need
- An architecture description or IaC
- Workload characteristics (volume, freshness, consumers)
- [object Object]
Steps
- Map the described services to a data flow. (from agent-aws-architecture-reviewer )
- Check IAM for least privilege. (from agent-aws-architecture-reviewer )
- Estimate dominant cost drivers. (from agent-aws-architecture-reviewer )
- Identify single points of failure. (from agent-aws-architecture-reviewer )
- Collect the four outputs — architecture diagram (text), IAM findings, cost drivers, reliability risks. (from agent-aws-architecture-reviewer )
- Hold the guardrails — the agent never creates resources and requires human approval for any IAM recommendation. (from agent-aws-architecture-reviewer )
Check yourself
- The review flags a Glue job running with an admin role and no bookmark, causing nightly full reprocessing — the agent's own worked example.
- Every IAM recommendation stops at human approval, and the agent created zero resources during the pass.
Worked example
Bind the AWS MCP server to a named profile — from mcp-aws
mcpServers:
aws-core:
command: uvx
args: [awslabs.core-mcp-server@latest]
env:
AWS_PROFILE: ${AWS_PROFILE}
FASTMCP_LOG_LEVEL: ERROR
Four-step review workflow — from agent-aws-architecture-reviewer
- Map the described services to a data flow.
- Check IAM for least privilege.
- Estimate dominant cost drivers.
- Identify single points of failure.
Outputs and guardrails — from agent-aws-architecture-reviewer
Outputs:
- Architecture diagram (text)
- IAM findings
- Cost drivers
- Reliability risks
Guardrails:
- Never creates resources
- Requires human approval for any IAM recommendation
Sourced from
Security scan
Last sweep 2026-10-24: 0 finding(s), 0 marked as examples.
Common gotchas
- The agent never creates resources — anything beyond findings needs a human.
- IAM recommendations always require human approval; never auto-apply them.
- Bind mcp-aws to a dedicated least-privilege AWS profile with short-lived SSO credentials — never attach an admin role.
Production considerations
- Score each pass against the 30-anti-pattern checklist so reviews are comparable over time.
- Keep the review read-only against the account; resource-creating tools require human confirmation.
Diagram
aws architecture review
|
|- map services --> one data flow
|
|- IAM --> least privilege --> human approval only
|
|- cost drivers --> dominant spend ranked
|
|- single points of failure --> reliability risks
|
--> findings: diagram + IAM + cost + reliability
Registry stats
| Verification | community · Oct 21, 2026 |
|---|---|
| Completeness | 81 |
| Trust multiplier | ×1.25 |
| Composite | 34.15 · adequate |
| Freshness | 3 days since last source check |
| Difficulty | intermediate |
| Version | 0.1.0 |
| License | CC-BY-4.0 |
| Security flags | none |
Did this solve your problem?
Run an AWS data architecture review pass