Scans AWS environments and generates visual architecture diagrams
Three tools with minimal schema definition and weak descriptions. scan_vpcs has empty input schema ({}), scan_compute and scan_storage have one optional parameter each with minimal type info. All descriptions are under 100 chars and lack context on when to use each tool, what they return, or how results chain together. No output schemas documented. No error handling guidance. No parameter constraints or enums. Tool names follow verb_noun pattern (positive), but descriptions fail to explain what data is returned or how to use results downstream.
Scan compute resources (EC2, ECS, EKS, Lambda)
Scan storage resources (S3, RDS, EFS, EBS)
Scan VPCs and related networking resources
scan_vpcs has empty input schema ({}). No parameters documented, no constraints, no validation guidance.
No output schemas documented for any tool. LLMs cannot plan downstream calls or extract required fields (vpc_id, subnet_id, etc.) without knowing response structure.
Descriptions lack context on what data is returned and when to call each tool. 'Scan VPCs and related networking resources' does not explain whether it returns subnets, security groups, or route tables, or how results chain to scan_compute.
vpc_id parameter in scan_compute and scan_storage lacks description. LLMs cannot infer whether to pass a VPC name, ID, or ARN, or whether the parameter is required or optional.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 39 | <=2025-11-25 | v2 |
No error handling guidance. Code catches BotoCoreError and ClientError but returns empty lists or dicts. LLMs receive no signal about what failed, why, or what to do next.