Skip to main content
Use this page as a fast incident checklist when CloudEval actions fail or return incomplete results.

​
Why it matters

Most failures are recoverable if you identify the layer first: auth, project data, or plan entitlement.

​
Quick example

cloudeval ask "Summarize current project health" --project <project_id>
Expected output: A valid response confirms auth, project scope, and service availability.

​
Step-by-step instructions

​
1. Check authentication

  • Web: confirm active session.
  • CLI: run cloudeval login again.

​
2. Check project inputs

  • Validate template JSON format.
  • Start with a reduced template if parsing fails.
  • Confirm expected files exist in project file tree.

​
3. Check billing and credits

If credits are exhausted, regeneration and compute-heavy actions can be blocked.

​
4. Re-run on a known-good project

Use a small baseline project to isolate whether issues are data-specific.

​
Code examples

cloudeval ask "List recent report issues" --project <project_id> --json

cloudeval ask "List files in this project and highlight missing dependencies" --project <project_id>

​
Expected output

  • You receive valid responses from project-scoped commands.
  • Entitlement-related failures are clearly visible.
  • Re-runs on baseline projects help isolate the root cause.

​
Common mistakes

  • Relying on outdated docs snippets.
  • Treating limited features as fully available.
  • Debugging report output before validating project ingestion.

​
Tips / best practices

  • Keep one “known-good” project for regression checks.
  • Re-run failed workflows with smaller template slices.
  • Track changes in plan entitlement when debugging blocked actions.
Last modified on March 5, 2026