What this category’s data looks like
Data sources for this category include ERP material master data, melting furnace thermometers, spectrometers, MES heat records, mold management systems, and customer technical drawings. Update frequency follows per heat, per batch, or daily cycles. Daily documents include foundry process cards, batching lists, heat records, mold histories, first article inspection records, material test reports, and nonconforming product review forms. Fields include alloy element values, furnace temperature readings, pouring duration, mold cycle counts, defect types, and more. Units involve degrees Celsius, seconds, mold cycles, and other standard production units.
What constraints these characteristics impose on the "tool calling and plugins" workflow
Data for this category updates per heat or batch and originates from dispersed sources. This requires tool calling to support real-time pulling of the latest data for a single heat or batch, and to avoid using expired cached data. Structured documents with multiple fields require plugins to support precise querying of specified fields, to reduce invalid data transmission. Cross-system data integration needs require configuring logic for chained calls across multiple plugins. Additionally, since defect tracing requires associating heat and mold cycle identifiers, tool calling must support passing unique business identifiers as input parameters, and must retain call logs for troubleshooting.
How to set configurations
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
| Workflow/API tool selection | auto, specify dedicated tools as needed | Adapts to multi-plugin calling scenarios, can automatically trigger corresponding tools for melting data, mold management, and other use cases, and aligns with the calling logic of FastGPT v4.17.0 |
| Workflow call timeout | 120 seconds | Aligns with production response timelines for foundry processes, prevents critical workflows from being interrupted by timeouts |
| Plugin call logs | Enabled | Retains call records for defect tracing and workflow troubleshooting |
| Workflow/API function-call configuration | Enable parameter validity verification | Ensures the format of incoming business parameters such as heat numbers and mold cycle counts complies with requirements |
| Workflow unique identifier | Bind to business heat ID | Establishes association between calls and specific production batches, to facilitate subsequent problem location |
Three common configuration mistakes
- Symptom: No conversation logs are retained after calling an external plugin. Cause: The plugin call log switch is not enabled, or the log storage configuration is not linked to the storage path for foundry business scenarios.
- Symptom: Workflows interrupt when calling custom Python code plugins. Cause: Correct code input parameter mapping is not configured in the workflow, and the format of foundry-specific parameters such as alloy batching and furnace temperature is not adapted.
- Symptom: Workflow calls return empty parameters. Cause: The incoming heat ID does not match the production data for the corresponding heat, or the
functionCallparameter verification is not enabled, resulting in parameters being filtered out.
How to confirm the configuration is complete
- Trigger a single-heat equipment inspection workflow, and check whether the workflow logs include detailed records of plugin calls.
- Pass a known existing heat ID, and verify that the fields returned by the tool call include foundry-specific data such as alloy element values and furnace temperature readings.
- Test passing a furnace number with an invalid format, and confirm that the workflow triggers the
functionCallparameter verification and returns a clear parameter error prompt. - Check whether the unique identifier of the workflow matches the test heat ID, to confirm the association between the call and the production batch.
The values provided are common starting points and should be measured against the reader's own samples.