What data for this category looks like
This category of data comes from cloud audit logs, IAM permission systems, CMDB resource inventory libraries, and security baseline scanning tools. Daily produced documents include compliance checklists, permission audit reports, resource inventories, and rectification tracking forms. Fields cover resource unique identifiers, permission configuration items, compliance check status, rectification tracking records, and no additional custom units. The data update cycle is monthly or quarterly; some real-time synchronized cloud audit logs have a higher update frequency. The data focuses on the permission and configuration compliance of cloud resources, core associated resource identifiers and compliance check results, and does not include non-compliant business content.
What constraints these characteristics impose on the knowledge base retrieval and recall link
Decentralized data sources and long update cycles require retrieval and recall to prioritize matching existing structured compliance documents, and avoid frequent pulling of real-time data that causes timeouts. Document fields focus on resource identifiers and compliance status, so retrieval must support precise filtering by resource ID and compliance type to reduce interference from non-compliant content. Low update frequency means the knowledge base does not need frequent refreshes, but must ensure that recall results cover all historical compliance records to meet full verification needs for internal and compliance audits. The associated attributes of rectification tracking forms and permission audit reports require retrieval to support cross-document associated recall, while also controlling the number of returned entries to adapt to the quick review needs of operations duty scenarios and avoid information overload.
How to set configurations
This configuration is based on FastGPT v4.17.0.
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
similarity | 0.75–0.85 | Matches structured fields of compliance checklists and permission audit reports, filters low-relevance non-compliant content |
limit | Top 8 entries | Adapts to the quick review needs of operations duty scenarios, avoids excessive entries interfering with access to core compliance information |
searchMode | keyword + vector | Balances precise matching of keywords such as resource IDs and semantic recall of compliance descriptions, covering different retrieval needs |
usingReRank | Enabled | Re-ranks recall results by compliance relevance, prioritizes displaying high-priority permission anomalies and non-compliant records |
collectionIds | Specify compliance knowledge base collection IDs | Limits the retrieval scope, excludes interference from non-compliant documents, and ensures results only include content related to this category |
datasetSearchUsingExtensionQuery | Disabled | This category of data has a low update frequency and focused content, no need to add additional query terms to expand the retrieval scope |
Three common errors
- Symptom: When calling knowledge base retrieval, "No valid matching results" is returned, and fallback knowledge base retrieval is not triggered. Cause: Multi-knowledge base fallback logic is not configured, only a single knowledge base collection ID is specified, and the process of automatically switching to the second compliance knowledge base when no results are found in the first knowledge base is not implemented.
- Symptom: When configuring global variables in a workflow, a string that is not a knowledge base collection ID is passed in, causing retrieval failure. Cause: Confusion between knowledge base IDs and other business identifiers, the requirement to pass in the collection ID of the compliance knowledge base is not clearly followed, and the format requirements for global variable assignment are not met.
- Symptom: The number of returned retrieval results far exceeds the preset value, and non-compliant documents are included. Cause: The
collectionIdsconfiguration to limit the retrieval scope is not set, or thelimitparameter value is too large, and filtering of irrelevant content based on the characteristics of this category of documents is not performed, resulting in non-compliant business data being mixed into the retrieval results.
How to confirm the configuration is complete
- Manually enter a typical resource ID, check whether the retrieval results only include documents of this category such as compliance checklists and permission audit reports.
- Adjust the
similarityparameter, verify whether the relevance of the returned results meets the quick screening needs of operations duty scenarios, and confirm that the threshold setting is reasonable. - Trigger a workflow call, confirm that when the specified knowledge base has no matching results, it automatically switches to the fallback compliance knowledge base for retrieval.
- View the retrieval log, confirm that the number of returned entries matches the configured value of the
limitparameter, with no additional redundant results.
The values listed above are common starting points, and should be evaluated against your own specific 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.