What Data for This Category Looks Like
This category of data primarily originates from source code repositories, dependency lists, static application security testing (SAST) tools, and public vulnerability databases. Daily produced documents include code audit reports, vulnerability proof materials, repair recommendation documents, and retest records. The data update schedule is adjusted alongside project version iterations or project progress, with no fixed cycle. A single data entry includes fields such as dangerous function identifiers, vulnerability paths, impact scope, and repair solutions. There is no unified standardized unit, and content varies based on the code size of the audit target and the number of vulnerabilities, resulting in significant differences in single-document length.
Constraints on Citation Sources and Traceability
Scattered data sources linked to multiple tools and repositories require precise matching of output identifiers from corresponding tools during traceability, to avoid confusion between audit results for different code bases. Updates follow no fixed cycle, so dynamic association with the latest project version data is necessary to prevent traceability to outdated audit records. Single data entries have numerous fields and no unified unit, so original document field integrity must be preserved during traceability to avoid truncation of critical vulnerability repair information. Code audit-related data includes content linked to high-risk vulnerabilities, so traceability must strictly bind to the original generation node to prevent cross-contamination of audit results across different projects.
Configuration Settings
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
quoteMaxToken | 800–1200 token | Matches the average length of a single code audit document to avoid truncation of key repair solution content |
responseData | Full format including original document path and generation tool identifier | Ensures returned results carry original document association information required for traceability |
detail | Enabled | Displays complete vulnerability proof and retest records to meet traceability verification requirements for security audits |
| Single recall limit | Top 5 entries | Matches the number of associated documents for a single audit report to avoid excessive redundant information in results |
| Data source binding rule | Associate by project | Accurately matches audit data for the corresponding code repository to prevent incorrect recall of documents across projects |
Three Common Configuration Errors
- Returned results include cited content but the citation display function is not enabled. This occurs because FastGPT v4.17.0 enables implicit citation output by default in some scenarios, and the
detailparameter status is not correctly configured. - Cited documents do not match the current Q&A content. This occurs because data sources are not bound by project, leading to incorrect recall of code audit documents across projects.
- Variable reference format is invalid. This occurs because the platform-specified variable wrapping format is not used, and custom identifiers are passed directly, resulting in parsing failure and inability to correctly associate the corresponding audit documents.
How to Verify Correct Configuration
- Upload a test code audit document, initiate a Q&A about vulnerabilities within the document, and check whether the returned results include expected traceability information.
- Adjust
quoteMaxTokento 500 token, and verify that the length of returned cited content matches the set value. - Disable the
detailswitch, and confirm that returned results no longer include complete vulnerability proof and retest records. - Switch to a data source not bound to the current project, initiate a Q&A, and confirm that audit documents for the target project are not recalled.
The values provided here 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.