Troubleshooting
| What you see | What to inspect next |
|---|---|
| First study has no recommended design | Its deliberately unresolved self-field screen; read the diagnosis and candidate checks. |
| CLI exits nonzero but writes a record | Search/acceptance verdicts; a completed diagnostic calculation can be non-passing. |
| Missing or mismatched material | Dataset ID and exact CSV hash for every base/graded binding. |
| Material query is unsupported | Local peak field, temperature, tape orientation and the dataset’s declared domain. |
| Coarse pass becomes inconclusive | The refined plan’s named limiting sample and fidelity boundary. |
| Spreadsheet import is rejected | Column mapping, units, n-value, nominal coordinates, bridge width, attribution, duplicates and size limit. |
| BOM is unavailable | Whether the record contains an eligible passing optimum. |
| Review package cannot rerun | Attach the exact original case and every matching external dataset. |
| Scenario ranking is unresolved | Incompatible requirements, missing dependencies or unresolved alternatives. |
| A documented feature is absent | The installed version; v0.2.0 archives predate newer source/hosted workflows. |
Browser recovery belongs to one browser/device. Keep exported workspaces and review packages for portable studies. Cancellation can retain previous completed evidence; wait for the job’s final state before starting another calculation.
When reporting a defect, include the version or commit, platform, exact command/UI action, relevant verdict and a minimal shareable input. Remove private supplier and customer information before attaching files to GitHub Issues.