What This Category's Data Looks Like
The typical subjects of this category are manufacturers producing injection molded parts, automotive interior components, home appliance plastic parts, and medical plastic parts. Daily business documents include injection molding process cards, raw material drying records, mold histories, first-piece inspection reports, patrol inspection logs, and material change and machine cleaning records. Data sources cover resin batch property sheets, injection molding machine operating parameters, mold temperature controller data, MES systems, and dimensional and appearance inspection results. Data update frequency is per production batch, each mold change or material change operation, and at the end of each shift. Document fields include process parameters, raw material batch numbers, mold numbers, inspection values, and similar items, with units mostly being degrees Celsius, megapascals, seconds, and other standard industrial units.
Constraints on Tool Calling and Plugins
Because data update frequency is high and involves multi-source heterogeneous documents, tool calling must support real-time pulling of the latest production data and document content. Different types of business documents correspond to different query logic, so targeted calling rules must be configured for tools. The units and formats of data fields vary widely, so tool calling must adapt to the return formats of different data sources. Error consequences involve batch defects, mold damage, and safety issues, so tool calling must add verification links to ensure the accuracy of returned data. This category has no cross-link curing steps, so there is no need to call vulcanization-related tools, which further clarifies the scope of tool calling.
How to Configure
This configuration is based on FastGPT v4.17.0, with the following details:
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
| Workflow/API tool selection | Auto-select | Adapt to different types of process and standard document query requirements |
| Workflow/API function-call configuration | Enable multi-function calling | Support simultaneous calling of multiple types of tools such as raw material properties and injection molding machine parameters |
| Tool call timeout threshold | 300 seconds | Cover the time required to pull mold histories and patrol inspection data across systems |
| Tool call retry count | 2 times | Address occasional fluctuations in equipment data interfaces |
| Context retention length | 6000–8000 token | Fully carry associated information of injection molding process cards and material change records |
Three Common Configuration Errors
- Phenomenon: The tool call parser returns a
Tool call Parsererror. Cause: The reasoning model wraps thinking content in think tags and places it in the content field, which does not meet the format requirements for tool calling. - Phenomenon: Returning a long error code when calling the interface via aiproxy forwarded keys. Cause: The forwarding link does not correctly adapt to the protocol format of tool calling, resulting in abnormal parameter transmission.
- Phenomenon: Unable to trigger external tools when enabling the tool calling function. Cause: The MCP server is not deployed as an intermediate scheduling layer, and tool calling in new versions of FastGPT depends on this component.
How to Verify Successful Configuration
- Submit a test request that includes an injection molding process parameter query, and verify whether the tool call returns valid field content from the corresponding document.
- Check the tool call logs to confirm that the
toolChoiceorfunctionCallparameters are correctly identified and trigger the corresponding tools. - Simulate a material change and machine cleaning scenario to verify whether the tool can pull the latest material change record data.
- Test external interface calls to confirm that the interface returns normally without abnormal errors.
The values provided are common starting points and should be evaluated against relevant samples.