Hayashi – Architecture Boundary Diagrams as Code (github.com)

Hayashi is a new Rust-based REST service that converts Terraform state files into FedRAMP-aligned architecture boundary diagrams as code. It parses state files, builds a directed resource graph, classifies subnets and trust zones (public, app, DB, management, IAM/KMS/audit planes), and emits Mermaid flowchart syntax styled with AWS icons, color-coded security zones, encryption labels, and NIST 800-53 annotations (SC-7, AU-2, CM-2, etc.). The service uses Cloudflare Workers AI (default model @cf/openai/gpt-oss-120b) to polish diagrams into human-friendly layouts, and returns mermaid_content plus metadata (node/edge counts, timestamp). It supports many AWS services (VPC, ELB, EC2/ECS/EKS, Lambda, S3, RDS, KMS, CloudTrail, GuardDuty, etc.) and exposes POST /v1/diagram requiring Terraform state path and diagram name; requires Rust 1.70+ and a Cloudflare account with Workers AI access. For the AI/ML and infra community this is significant because it automates production-grade, compliance-aware architecture documentation and boundary visualization—turning opaque state files into auditable diagrams that can be integrated into CI/CD, security reviews, and FedRAMP/NIST reporting. By codifying boundaries and controls as machine-readable Mermaid, Hayashi accelerates audit readiness and repeatable infrastructure governance for regulated ML deployments. Note operational considerations: Terraform state often contains sensitive info and requires secure handling and Cloudflare API credentials when used in pipelines.