What This Category of Data Looks Like
Daily documents for security training and phishing drills include training slides, drill plans, phishing email templates, and drill result reports. Core data sources are employee contact lists, email gateways, training platforms, and click statistics. Data update frequency is usually quarterly or annually. Document fields include drill audience scope, template version, click record ID, review and rectification results, and other fields. Field dimensions vary across different documents, and there is no unified standardized field format.
Constraints on Citation and Traceability
Because data sources include multiple systems such as employee contact lists and email gateways, citation traceability requires associating unique identifiers from different data sources to avoid confusing different employees or drill records. Diverse document types and inconsistent field dimensions require configuring differentiated citation display rules for different documents. The low update frequency means traceability data must have long-term stability, preventing citation failure caused by data source changes. Drill result reports involve audit and rectification evidence, so traceability information must accurately correspond to specific drill links and document versions, and fuzzy matching logic cannot be used.
Configuration Settings
| Configuration Item | Recommended Approach | Rationale |
|---|---|---|
quoteMaxToken | 800–1200 characters | The core content length of phishing email templates and drill reports usually falls within this range, avoiding truncation of key traceability information |
responseData | Return only associated document metadata | Retain data source identifiers such as employee ID and click record ID to meet the accuracy requirements of audit traceability |
detail | Enabled | Display specific fields of cited documents, such as drill audience list and template version number, to facilitate review and verification |
| Citation source display location | Fixed area at the end of replies | Unify the traceability format, in line with the viewing habits of security audits |
| Maximum number of citations | Top 3 entries | Associated documents for a single drill usually do not exceed 3; excessive content will interfere with the readability of core answers |
Three Common Misconfigurations
- Unassociated knowledge base search input and response fields appear in replies. This occurs when the default citation display switch for tool calls in the workflow is not disabled, causing content from non-target data sources to be forcibly traced.
- Citation source fields are empty, and association with employee contact list or email gateway data is not possible. This happens when
responseDatais not configured to return only associated metadata, leading to data source identifiers not being correctly extracted. - No optional values are available for knowledge base variable references in the workflow. This occurs when data sources such as employee contact lists and email gateways are not configured as global variables, or the permission switch for variable references is not enabled.
How to Verify Successful Configuration
This verification is performed using FastGPT v4.17.0.
- Launch a simulated phishing drill test, and check whether the specified associated document metadata is displayed at the end of the reply.
- Access the workflow variable configuration page, confirm that data sources including employee contact lists and email gateways have been loaded as global variables and can be selected for reference.
- Adjust the value of
quoteMaxToken, and verify that cited document content in the reply is neither truncated nor over-displayed. - Disable the default citation display switch for tool calls, and confirm that only the core answer and configured traceability information are retained in the reply.
The values provided are common starting points and should be measured 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 2026-10-19.