What data for this category looks like
Reference only; verify against the target FastGPT release and deployment documentation.
What constraints these characteristics impose on the tool calling and plugins workflow
Mixed input of multi-source data such as device sensors and operation and maintenance logs requires the tool calling module to support configuring multi-data source aggregation plugins to integrate data according to preset field mapping rules. The custom patrol cycle update schedule requires tool calling trigger parameters to support binding patrol time interval variables, to avoid reading expired data. Clear parameter fields and unit requirements require that field extraction rules must be locked in the tool calling configuration, to prevent non-standard numerical values from being returned. The single document length range requires that the tool calling context transfer limit must match the conventional document length, to avoid truncation of key alarm information.
How to configure settings
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
toolChoice | auto | Adapt to the dynamic alarm judgment requirements of health diagnostic reports, allowing the model to independently trigger tool calls |
functionCall | Preset field extraction function template | Ensure that tool calls only return core monitoring parameters within the report, avoiding redundant information |
| Multi-data source binding | Bind device sensor and operation and maintenance log plugins | Match the multi-source data generation logic of health diagnostic reports |
| Context length limit | 1500–2500 characters | Adapt to the conventional length range of a single report to prevent key information from being truncated |
| Request timeout setting | 300 seconds | Reserve processing time for multi-source data aggregation and field extraction |
Three common configuration mistakes
- The tool calling module shows "not triggered" in its status, and the log returns "no valid call instruction". The
toolChoiceparameter is not correctly configured asauto, or the conditions for triggering tool calls are not clearly specified in the prompt. - The tool calling module returns empty results and cannot obtain associated patrol history data. The knowledge base plugin is not bound to the tool calling link, or the trigger condition for knowledge base retrieval is not specified in the function template.
- The tool calling return result contains redundant fields not included in the report, and the field units do not meet preset requirements. The precise field extraction rules are not configured via
functionCall, and the unit verification logic is not locked.
How to confirm the configuration is correct
- Verify the trigger configuration of the tool calling module, confirm that its bound trigger event matches the generation node of the health diagnostic report.
- Perform a single test call, check whether the fields of the returned result match the preset extraction rules.
- View the tool calling running logs, confirm that all bound data source plugins return data normally.
- Verify whether the function template configured via
functionCallcan accurately extract unit information within the report.
The values provided here are common starting points and should be evaluated against the reader's own samples.
Question material comes from public community discussions. Configuration values are common starting points and should be measured against your own samples. Verified on 2026-09-28.