cloudwright

Our take

Python tool (pip cloudwright-ai, MIT) that turns a one-line English description into a typed architecture spec, region-aware per-component cost estimates with confidence flags, compliance reports mapped to HIPAA/SOC 2/PCI-DSS/FedRAMP/NIST control IDs (with OSCAL export), and Terraform/OpenTofu/Pulumi/CloudFormation across AWS, GCP, Azure, and Databricks; the MCP server is a separate cloudwright-ai-mcp package exposing 18 tools over stdio or SSE. Design and chat commands call an LLM with your own ANTHROPIC_API_KEY (or OpenAI key), while costing, validation, review, and export run offline. Impressively broad and moving fast — v1.0 to v1.6 between March and June 2026 — but it is a solo project with ~118 PyPI downloads/week, so treat generated IaC and compliance findings as drafts to review, not audit evidence. For architects and platform engineers prototyping cloud designs.

reviewed by hand · 2026-07-05

Something wrong or dead here? Report it

Verification record

last verified

11 days ago

github stars

32

last commit

19 days ago

archived

no

license

MIT

in registry since

2026-06-16

github.com/xmpuspus/cloudwright