Skip to main content
Generate a diagram early to confirm your project data model is correct before deep analysis.

​
Why it matters

A trustworthy diagram is the base for reliable AI chat, reports, and team review.

​
Quick example

Prompt: "Highlight resources with public exposure and missing dependencies."
Expected output: A focused response that references concrete nodes and relationships in your diagram.

​
Step-by-step instructions

  1. Create a new project.
  2. Upload an ARM template file.
  3. Trigger diagram generation.
  4. Verify nodes for core network, compute, and data resources.
  5. Verify critical dependency edges.
  6. Save the project.

​
Code examples

# Validate from CLI after diagram generation
cloudeval ask "List top-level services and their dependencies" --project <project_id>

​
Expected output

  • Resource nodes are visible and labeled.
  • Dependency edges align with template intent.
  • AI can answer graph-aware questions for the project.

​
Common mistakes

  • Missing linked resources in the source template.
  • Treating unresolved template references as graph bugs.
  • Skipping validation and jumping directly to remediation.

​
Tips / best practices

  • Start with a known-good template to baseline quality.
  • Validate one subsystem at a time for large architectures.
  • Keep a short checklist for “diagram readiness.”
Last modified on March 5, 2026