Workflow and documentation reviewed 2026-10-24 UTC. The product interface and detailed operating guides are currently in Chinese.
What you can check
- Brand and listing text: use screening results to identify names and terms that need closer review.
- Product images: investigate possible copyright and design-patent similarities. A similarity result is a review lead, not a legal determination.
- Store catalog: review listings across a connected store and prioritize flagged items.
- Case research: locate stored IP lawsuit records by relevant brand or case information, then verify the original docket.
Screening does not guarantee freedom from infringement, prevent every lawsuit or replace checking rights and authorization.
Walmart US: check a listing before publishing
Start with the actual SKU, proposed brand, title, product category and image files. A brand-field match and a word used descriptively in a title need different context; retain the field position rather than submitting an unexplained word list.
- Screen and reconcile: run a small brand/title sample first. Keep each result attached to its original SKU, input field and observation time; record failed and untested rows separately.
- Review trademark leads: compare potentially similar marks, related goods and current official records. The USPTO describes clearance as a broader search that also considers common-law use; a database search alone does not settle clearance.
- Check other dimensions: examine photo rights, visible logos and relevant patent or case leads separately. A completed text check does not complete image, patent or authorization review.
- Record a decision: retain source links, supporting rights documents, unresolved questions and the reviewer. Escalate uncertain rights or an actual dispute for qualified advice before making a publishing decision.
Primary references: USPTO clearance-search guidance and Walmart Marketplace intellectual-property policy. These sources do not endorse SafeSell AI.
Continue with the Walmart screening guide — Chinese, bulk SKU workflow — Chinese and multi-store review handoff — Chinese.
Can you use a supplier’s product photos?
Receiving a file does not establish who owns it or what use is permitted. For US copyright, original photographs can be protected when created; the photographer is generally the initial owner, with exceptions such as qualifying works made for hire. Confirm the rights and basis for your proposed use, rather than treating a supplier’s possession of the image as proof.
| Review | Evidence to retain | Limit |
|---|---|---|
| Origin and rights | Original file, source URL, supplier identity, claimed rights holder, permission or other documented basis for use. | A supplier statement may leave the underlying ownership unresolved. |
| Permitted use | Covered images, storefront/channel, territory, term, editing and advertising rights, plus the supporting agreement. | Check that the permission actually covers the proposed use. |
| Visible elements | Logo/text crops, reverse-image-search leads and relevant trademark, copyright or design-patent references. | Similarity is a review lead; it does not establish infringement or permission. |
| Review outcome | SKU, file version, observation time, reviewer, unresolved issues and the final asset selected. | Removing a logo or watermark does not establish rights to the underlying photo. |
SafeSell AI’s product-image workflow can help investigate candidate risks. It does not certify image ownership or supply a copyright licence. This checklist concerns Walmart US listings; a statutory exception or a disputed right needs its own context.
Primary reference: U.S. Copyright Office photography guidance. Practical follow-up: product-image self-check — Chinese.
ERP integration: inputs, outputs and failure handling
The public OpenAPI document describes authenticated REST checks. Business requests use Authorization: Bearer <API_KEY>; account-plan eligibility, scopes and available S compute balance still apply. Keep keys on your server. Read the current interactive API documentation — Chinese and OpenAPI JSON before implementation.
| POST endpoint | Documented input | Documented purpose / scope |
|---|---|---|
/api/v1/check/brand | brand, optional category | Brand-risk result; write:check. |
/api/v1/check/batch | words[] or items[{word,position}]; category context. | Per-word screening; write:check. Preserve the documented field positions. |
/api/v1/check/link | url | Product extraction and screening; write:check. A fetch failure is an incomplete check. |
/api/v1/check/tro | brand | Stored case-history leads; write:check. Verify the original docket. |
/api/v1/check/store-audit | items[{sku,title,brand,features,description}], own_brands. | Structured item review; write:audit. Preserve SKU/result mapping. |
The documented response envelope separates success: true with data from success: false with error.code and error.message, with metadata. Endpoint-specific payloads vary; the public specification is not a complete response schema for every operation.
- 400 / 401 / 402 / 403 / 429: handle invalid input, invalid key, insufficient balance, scope/plan restrictions and rate limits separately.
- 502 / 503: the link contract lists extraction failure; the case-history contract lists
tro_data_unavailable. Neither means “no risk found.” - Costs and retries: check the current endpoint terms and account before running. Before retrying, establish whether the previous operation completed or incurred usage; do not assume retries are free or idempotent.
Documentation inspection confirms what is published. It does not prove a successful authenticated brand, image, batch or store-audit request, account scopes, or charging behavior. The public web sample below uses a different endpoint.
Continue with the integration and error-handling guide — Chinese, developer entry — Chinese and MCP authorization guide — Chinese. REST, web uploads and MCP tools have separate contracts and limits.
Evidence you can inspect
Controlled brand/title example
A controlled anonymous request on 2026-10-24 sent two self-created brand/title inputs to POST /api/public/check-brands and received HTTP 200 with two results. Inspect the saved inputs and response JSON and field-by-field explanation — Chinese. This is a software-output example, not a customer success story or legal clearance. Image checks, patent checks and authenticated ERP calls were not part of that example.
Case collection and methods
The database currently contains 23,035 stored records; 8,475 have a recorded filing date in 2026.
These are collection-coverage counts, not complete U.S. litigation totals or counts of confirmed TRO orders. Filing-date quality remains under review. A case record is not a finding that a seller infringed anyone’s rights.
Read the current counts, definitions and citation instructions or download the aggregate JSON.
Common questions
Does a clear screening result mean a listing is safe to publish?
No. A completed check covers only its stated inputs and sources. Keep unresolved rights, missing evidence, untested dimensions and failed checks open for review before deciding whether to publish.
Is the public brand/title sample an authenticated ERP API test?
No. It records a controlled anonymous web request to /api/public/check-brands. It does not validate authenticated /api/v1/check/batch, image checks, account scopes or API billing.
English resources
Brand case-history research · TRO guide · Public case library — Chinese · Walmart appeal service — Chinese