Reference only; verify against the target FastGPT release and deployment documentation.
What data in this category looks like
Data in this category mainly comes from local public data open platforms, integrated big data platforms, departmental business databases and electronic credential repositories, with a monthly update frequency. Daily management documents include public data resource catalogs, data open lists, data item descriptions, anonymization rules, interface documents and update frequency descriptions. Data fields cover data item names, open subjects, anonymization requirements, call interface addresses, update cycles, authorized usage scopes and other content. Data is stored in a structured format, and linked to the open permissions and usage boundaries of specific data assets.
What constraints these characteristics impose on conversation logs and auditing
Monthly updated public data catalogs require conversation logs to be linked to the catalog version of the corresponding cycle, to avoid logs matching outdated data items. In multi-source data access scenarios, logs must fully record the source node of data calls, interface address and anonymization execution marker, to ensure audit traceability of data flow links. The diversity of structured data fields requires logs to cover key information such as data item names and authorization scopes, to meet the complete traceability requirements of compliance audits. The compliance attribute of public data requires log retention to comply with general government data audit specifications, and must not omit full-link records of conversation initiation, data return and permission verification.
How to set configurations
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
chatId | Bind to monthly catalog version identifier | Matches the monthly update feature of public data catalogs, ensures logs are linked to valid data items for the corresponding cycle |
responseChatItemId | Link to data interface call link | Traces the specific source node of data returns, meets the traceability requirements of multi-source data audits |
detail | Enable full-field log recording | Covers key audit fields such as data item names and anonymization markers, avoids omitting compliance verification information |
| Log retention period | 180 days | Matches the general retention period requirement for government data audits |
| Single export limit | Top 10000 entries | Prevents system resource overload caused by excessive single export data volume |
| Log trigger rule | Triggered at conversation completion node | Ensures complete logs are only generated after a conversation is closed, reduces invalid log generation |
Three common mistakes
- Symptom: An error
$lookup with 'pipeline' may not specify 'localField'appears when viewing logs. Cause: The public data catalog version identifier and log query parameters are not correctly linked, causing the aggregation query to match outdated data table structures. - Symptom: Single log export only returns 50000 entries before interrupting. Cause: No reasonable single export limit parameter is configured, the system defaults to limiting single batch export data volume to avoid resource overload.
- Symptom: Data anonymization execution markers are not displayed in conversation logs. Cause: The
detailfull-field log recording configuration is not enabled, causing key audit fields to be omitted.
How to confirm the configuration is complete
- Initiate a party building learning-related public data query conversation, check whether the current cycle's public data catalog version identifier is bound in the log list.
- Attempt to export logs, confirm that the number of single exports matches the preset upper limit configuration, with no forced interruptions.
- View the detailed log of a single conversation, confirm that key audit fields such as
responseChatItemIdand data anonymization markers are included. - Check the system log retention configuration, confirm that the retention period meets the general requirement threshold for government audits.
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 2026-09-28.