Skip to main content
This quickstart is optimized for first value, not feature exploration.

​
Why it matters

If you complete these steps in order, you avoid most onboarding failures and get reliable architecture output quickly.

​
Quick example

cloudeval ask "Identify top 3 risks and recommended fixes" --project <project_id> --json
Expected output: A prioritized risk list with short remediation notes.

​
Step-by-step instructions

​
1. Create a project

  1. Open CloudEval.
  2. Sign in.
  3. Create a new project.

​
2. Add IaC input

  1. Upload a valid ARM template (.json).
  2. Confirm the file appears in project files.
  3. Validate that resources are detected.

​
3. Generate the architecture

  1. Run diagram generation.
  2. Verify core network, compute, and data nodes.
  3. Check key dependency edges.

​
4. Run AI checks

Use prompts like:
  • List internet-exposed resources and risks.
  • Show likely monthly cost hotspots.
  • What should we harden before production?

​
5. Trigger report workflows

Run project report actions for:
  • Cost insights.
  • Well-Architected/security insights.
  • Unit test style report regeneration.

​
6. Confirm plan and credits

Before team rollout, verify plan limits and available credits.

​
Code examples

# Project-scoped chat from terminal
cloudeval ask "What are the top cost drivers?" --project <project_id>

​
Expected output

  • Architecture view is generated and reviewable.
  • AI responses point to concrete resources.
  • Reports can be regenerated for your project.

​
Common mistakes

  • Uploading parameterized templates without required values.
  • Assuming report actions are free of credit constraints.
  • Treating limited exports as universally available.

​
Tips / best practices

  • Keep your first project narrow and representative.
  • Use the same template in web and CLI for parity checks.
  • Save successful prompts for team reuse.
Last modified on March 5, 2026