What Data for This Category Looks Like
Data for this category comes from slab or billet batches, heating furnaces, mill automation control systems, online thickness and width measurement tools, surface inspection systems, and MES systems. Daily documents include rolling plans, heating schedules, rolling specifications, flatness and thickness records, surface quality reports, and quality certificates. Individual documents contain fields such as batch number, specification parameters, and test values. Thickness record fields use millimeters as the unit, temperature parameters use degrees Celsius, and rolling speed uses meters per minute. Data update frequency is per coil or per billet. Some online inspection data updates in real time.
What Constraints These Characteristics Impose on Tool Calling and Plugins
The real-time update and coil-level association features of this category’s data require tool calling to support high-frequency triggering and precise batch matching. Field differences across multiple types of structured documents require plugins to pre-set field mapping rules for this category, to avoid confusing parameters between rolling plans and quality certificates. The potential consequences of errors such as steel coil downgrading or equipment damage require that the result verification step of tool calling covers parameter validity, such as whether temperature ranges and thickness tolerances comply with rolling specifications. The coil-level traceability requirement requires plugins to bind batch numbers as call context identifiers, to prevent misuse of cross-batch data.
Configuration Guidelines
This document applies to FastGPT v4.17.0.
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
| Workflow/API tool selection | auto | Supports automatic multi-plugin calling, no manual target specification, fits the cross-review scenario of multiple documents for this category |
| Workflow/API function-call configuration | enabled | Enables structured parameter calling, fits field parsing and parameter verification for structured documents in this category |
| Plugin trigger condition | Detect rolling batch number keywords | This category’s data uses batch numbers as core identifiers, avoids unrelated plugin calls |
| Number of returned data entries per call | Top 8–12 | Single batch data volume for this category is moderate; too much data increases review load, too little fails to cover complete process parameters |
| Timeout period | 300–600 seconds | Online test data updates in real time, data must be retrieved within a reasonable timeframe to avoid excessive waiting that reduces review efficiency |
| Context binding rule | Bind current batch number parameter | Coil-level traceability requirements ensure plugin calls associate with data for the corresponding batch, preventing data confusion |
Three Common Configuration Errors
- Phenomenon: Plugin calls return AxiosError 404. Cause: The API address of the FastGPT plugin is not configured to the actual interface address of the corresponding internal MES or inspection system, and the default sample address is still used.
- Phenomenon: Process parameters returned by the plugin do not match the current review batch. Cause: The batch number binding context configuration is not enabled, and the plugin call does not associate with the current steel coil or batch data being reviewed.
- Phenomenon: Plugin calls have no response. Cause:
functionCallis set to disabled, and structured parameter calling is not enabled, causing the plugin to fail to parse structured document fields for this category.
How to Verify Correct Configuration
- Manually enter a test question containing a rolling batch number, check whether the plugin automatically triggers a call.
- View plugin call logs, confirm that the API address matches the actual interface of the internal system, and no default sample address remains.
- Verify the parameter fields returned by the plugin, confirm that they match the fields of the current review’s rolling plans and inspection records.
- Adjust the test batch number, confirm that the data returned by the plugin updates along with the batch number.
The values provided are common starting points and should be measured against the reader's own samples.