What the documents look like
Livestream sample selection and quality inspection happen before a broadcast. The work decides whether a product can be sold and whether it suits live presentation. The materials used at this stage come from supplier files, testing-agency reports, platform negative lists, and historical return rates. Day-to-day documents include sample evaluation forms, quality inspection reports, supplier qualifications, and prohibited-sale lists. These materials refresh weekly and per item; that cadence describes how often source files change and does not by itself create batch synchronization, full synchronization, or real-time triggers in this page.
The conversation-logs and audit step sits inside operations permission and handover management. When one sample moves between a selection manager, a supply-chain business development contact, and a quality inspector, the log must show which conversations belong to one session, which run details can be retrieved, and which fields are actually retained. The three roles share the same sample record but not necessarily the same session identity, so the handover record and the chat record have to be reconciled against a common subject.
Constraints at the conversation logs step
Session identity is separated by chatId. When chatId is not passed, the system does not use the context feature provided by FastGPT and builds context entirely from the passed messages. When a non-empty string is passed, the system reads that session from the database and uses the last item of the messages array as the user question, ignoring the remaining messages. If handover work mixes sessions from different entry points into one reconciliation, the log loses its common subject.
Run details require setting detail according to the official contract. In non-stream mode the intermediate values are stored in responseData; in stream mode they are distinguished through event. The interface returns a limited set of fields. Business data that is not returned cannot be assumed to have been retained in full. Retrieval and similarity affect content matching only; they do not carry permission or entity isolation, and the existence of a log does not mean a compliance audit has been completed.
Hypothetical drill sample: on 10 March 2025 a quality inspector submits a handover request whose original text reads "Sample A's quality inspection report has been uploaded; please have the selection manager confirm whether it can enter the 12 March livestream list." Under the same chatId another record reads "Supplier qualification pending supplement; hold the broadcast." The draft sentence to be judged is "Sample A has completed all qualification verification and can go live directly." The materials for comparison contain only the two sentences above and do not state the party, time, or conclusion of qualification verification.
Configuration and implementation checks
The table below lists only settings that the official documents describe or implementation checks tied to this page's feature and primary task. The values are starting points for this hypothetical trial, used to compare against another value, and are not official defaults.
| Configuration item | Hypothetical starting point for this trial | Check method and reason |
|---|---|---|
| chatId | One value per quality handover session | Match one-to-one with the business record, keep the length under 250, and compare whether different entry points were merged into the same value |
| detail | Set to true when run details are needed | Compare with false and confirm whether node responses appear in responseData or in event |
| stream | Chosen according to the caller's rendering needs | Compare with the other value and confirm whether intermediate values are obtained through event or through responseData |
| responseChatItemId | Unique under the same chatId | Compare with existing values; a duplicate makes this response message indistinguishable |
| source | Set to api when retrieving sessions created through the API | Compare with page sessions and confirm the scope of the returned list |
| loadCustomFeedbacks | Enabled when custom feedback must be read | Compare with disabled and confirm whether feedback fields appear in the records |
Three common errors
- Treating a field that was not returned as retained. The interface response does not give a supplier qualification verification conclusion. Concluding from the response body that qualification has been verified has as its input the response body, as its comparison object the actual list of returned fields, and as its observable result the absence of the conclusion field.
- Merging sessions with different chatIds into one reconciliation. The input is two records from different chatIds, the comparison object is each chatId, and the observable result is that the subjects differ and cannot form one session's log chain.
- Reading a pending-supplement status as passed. In the sample, "Supplier qualification pending supplement; hold the broadcast" is a to-do description. The draft sentence writes that verification is complete. The input is the two original sentences, the comparison object is the status wording, and the observable result is that the status field values are mutually exclusive.
How to verify the setup
- Session identity check. The input is the two records in the hypothetical sample and their respective chatIds. Compare whether the two chatIds are identical. If they differ, they cannot be combined for calculation, and the item is recorded as pending verification.
- Run-detail check. The input is one request with detail=true. Compare whether responseData or event=flowResponses appears in the returned body. A node response must appear for the condition to be met. When it does not appear, record the gap and do not infer a failure cause.
- Draft-sentence check. The input is "Sample A has completed all qualification verification and can go live directly" together with the two original sentences. Compare the subject, time, and conclusion fields. The original text does not state the party or conclusion of verification completion, so the sentence is unsupported and requires additional material before re-checking.