What the documents look like
For operating reviews in Near-expiry markdowns, this page examines FastGPT HTTP integrations. Near-expiry candidate and proposed-markdown records separate registration, approval and execution. The examples are hypothetical.
Constraints at the HTTP integrations step
A hypothetical store report lists this batch on a near-expiry candidate sheet awaiting inspection. Label review was registered, a proposed markdown note awaits approval, and older price-tag changes concern another batch. Candidate wording describes this paper list without proving physical expiry status or saleability. An HTTP read must retain batch, reporting period, actual inspection time and proposal state, without substituting report creation for the event. No food standard, real shelf-life limit or markdown figure is supplied.
Configuration and implementation checks
| Near-expiry markdown HTTP | Hypothetical state | Check |
|---|---|---|
| Candidate sheet | Current batch uninspected | No physical-state inference |
| Markdown proposal | Approval pending | Not executed |
| Label review | Event time needs checking | Separate creation/event |
| Historical tags | Other batch operation | Do not import outcome |
| Approval/execution | Independent evidence | Reading changes no price |
Documented conditions
| Documented item / implementation check | Condition |
|---|---|
Request address and method | The request address and method follow the external contract. |
Params and Headers | Inspect parameters, headers and authentication separately. |
| Body | Where a JSON body is used, validate its structure against the external contract. |
Variable references | Variables can come from global, system or upstream outputs. |
Debug identity fields | Test mode lacks chatId and responseChatItemId. |
Three common errors
Registration reported as executed markdown needs current approval and price-tag operation evidence checked. Report creation treated as inspection time needs original event timestamps checked. Another batch price-tag outcome applied here needs goods and batch scope checked. These concern registration versus action, time role and subject.
How to verify the setup
Retain request, raw return and answer, tracing relationships among this candidate record, proposed markdown and approval. Inspect filters and body coverage under the real contract; authentication grants no business permission. Added approval changes its covered state alone while physical checks and tag execution need separate evidence. Missing approval proves no never-approved status. Reading decides no safety, saleability or price and changes no stock.
| Material context | Scope |
|---|---|
| Document types | Near-Expiry List, Discount Approval, Spoilage Report and Promotional Price Tag |
| Candidate sources | POS Inventory Age, Store Inspection, ERP Inventory and Quality Inspection Records |
| Technical baseline | FastGPT v4.17.2 |