Data profile for this category
Daily documents for the general mechanical equipment assembly category include assembly process cards, BOMs, torque records, gap adjustment records, commissioning reports, factory inspection sheets, and packing lists. Data sources include PLM/ERP, part inspection data, assembly tool collection systems, test benches, MES, and customer technical agreements. Update frequency follows per single device or per order. Document fields include material codes, torque values (unit N·m), gap values (unit mm), and others. Structure and field formats vary across different documents.
Constraints on tool calling and plugins
Because data sources are scattered and updated per single device or order, tool calling must support integration with multi-data source plugins to pull real-time assembly data. The diversity of document structures and fields requires plugins to adapt parsing rules for multiple formats, to avoid failure to extract key information. Single-order assembly data volume is large, so tool calling context must be isolated per order to prevent mixing of data across different orders. Additionally, tasks such as BOM completeness checks and gap adjustments take significant time, so tool calling should prioritize high-time-consuming verification scenarios.
Configuration settings
The following configurations are based on FastGPT v4.17.0.
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
| Workflow/API tool selection | auto | Adapts to scenarios where tools are dynamically switched during assembly inspection |
| Workflow/API function-call configuration | Allow forced invocation | Ensures dedicated tools can be triggered when precise verification of torque and gap values is required |
| Maximum context length | 8000–12000 characters | Accommodates assembly process cards, BOMs, and inspection records for a single order |
| Single tool call limit | First 3 entries | Avoids redundancy from repeated calls for similar assembly data |
| Timeout threshold | 60 seconds | Matches typical response times of PLM/ERP systems |
| File parsing format | Supports PDF, CSV, XML | Adapts common export formats for assembly documents |
Three common configuration errors
- Phenomenon: When calling a local large language model, the returned tool call result format does not meet expectations, and cannot be parsed into structured data. Cause: The compatible configuration of the
toolChoiceparameter was not adjusted for the locally deployed large language model, and the default settings of the online API were used instead. - Phenomenon: After uploading an assembly document via an API calling workflow, the corresponding BOM field is empty. Cause: No file type identifier matching the assembly category was specified in the request parameters, causing the plugin to fail to automatically extract the BOM field.
- Phenomenon: When calling the Qwen3-OMNI model for equipment video analysis, the inspection records of the corresponding equipment cannot be associated. Cause: No context association rules for tool calling were configured, and the video analysis results were not bound to order-level assembly data.
How to verify correct configuration
- Initiate a test request that includes BOM verification, confirm that tool calling can correctly pull assembly data from the corresponding data source.
- Upload an assembly process card document, check whether the plugin can extract key fields according to the configured parsing format.
- Call the specified large language model interface, verify that the
toolChoiceparameter can automatically switch tools based on the request. - Send a file upload request via API, confirm that the returned file variable parameters match the identifiers in the request.
The values provided are common starting points and should be measured against the reader's own samples.