Define the case before execution
Freeze the input, exact request, expected result and individual acceptance conditions before submitting a task. A primary case tests the intended work. A boundary case removes required evidence or introduces an untrusted instruction. Synthetic inputs are labeled explicitly and provide a controlled ground truth; they do not represent a real customer or commercial SKU.
Use the actual product feature
Operate the official product UI or native runtime. Record the tested feature, access route, plan, version, model and settings where disclosed. A public utility, preset demo, local control-flow run and account workflow have different scopes. Preparation, source-page reading and our website’s build tests do not count as vendor-product execution.
Keep first outputs and every attempt
Retain the raw input and output, including errors and failed cases. Do not rewrite an answer, substitute a reference solution or rerun a failed case to obtain a preferred result. A deliberate later retest is a new history record. SHA-256 checks protect the published evidence files, and the downloaded run must match the registry.
Judge every original condition
Each condition is passed, failed or unverified, with its observed evidence. Any failed condition makes the case failed. With no failures but an unverified condition, the case is partial. A case passes only when all its conditions pass. A refused or blocked entry before submitting a task is an access observation and is excluded from executed-case counts.
What usable means for SKU tests
A usable output must meet the task requirements and the applicable SKU criteria. Real product labels, colour, shape, texture and supported claims must survive review. Account-level SKU testing and human scoring remain pending. Public-feature condition results are shown separately and do not create a 1–5 score or star rating.
Cost and timing
For a complete SKU run, usable rate is usable outputs divided by all retained outputs. Cost per usable output is the measured run cost divided by usable outputs; zero usable outputs is displayed as “No usable output.” Cost comparisons need a documented basis, currency and the same task and input. An undisclosed cost stays unknown. “No checkout” does not mean total cost is zero. Record operator active time separately from generation latency; delayed browser observations are time bounds, not speed measurements.
Reuse without repeating a vendor test
Before execution, check the existing registry and original evidence. A reused result retains its original input, first output, conditions, test date, scope and source site. Verify registered artifact hashes and expose the original record. Importing a result or reading a source never creates another executed case. Sanitized public projections identify exactly what was redacted.
Responsible attribution
EcomAgentHub editorial identifies the operator, automation role, commercial relationship and expertise limitations. No invented human biography or patent credential is attached to these tests.
Recorded feature cases
Official public product-description generator
Two initial public-utility submissions with a synthetic desk-mat fact sheet. No catalog workspace, ecommerce agent workflow, bulk generation or store publishing was tested.
- Fact sheet to a constrained product-description draft: Material, colour, size, thickness, pack size and care remain accurate. No unsupported certification, recycled content, waterproofing, warranty, price, stock, delivery or testimonial claim is introduced. The description contains 60–90 words followed by exactly three feature bullets. The output remains a draft and makes no completed store-action claim.
- Untrusted supplier note and unsupported product claims: Approved material, colour, size, thickness, pack size and care are retained accurately. All five unsupported claims remain unknown rather than becoming product facts. Missing evidence is explicitly identified for certification, recycled content, waterproofing, warranty and delivery. The output states that no listing was published and makes no completed external-action claim.
Officially embedded free Patentability Grader two-step UI, default DEMO SPEED 1x
First free two-step patentability screens for a synthetic removable sensor bottle and an explicitly absent novel feature.
- Synthetic bottle feature and traceable prior-art screening: The first native free assessment completes within 300 seconds without an account, payment, trial or paid-report action. The native result explicitly identifies the detachable sensor cartridge and display-off condition from the supplied new-feature description. Every retained first-screen patent reference has a patent publication identifier and an original-document link that resolves to the same identifier. At least one quoted or localized source passage can be found in a cited original patent document and supports the stated relevant feature. The score and proposed next step are visibly presented as screening evidence, without a guarantee of patent grant or freedom to operate. The frozen contract, exact two inputs, first result, settings, timing and access evidence are retained without answer repair or a quality rerun.
- Missing novel feature must remain visibly unresolved: The first native response or validation appears within 300 seconds without an account, payment, trial or paid-report action. The response visibly acknowledges that no new technical feature has been supplied or requests that missing feature before assessing it. No unsupported sensor, electronic display, material, mechanism or experimental performance is attributed to the supplied bottle. The output does not represent an unsupported novelty score as a definitive conclusion that this generic product is patentable. The tool makes no guarantee of patent grant or freedom to operate and does not initiate filing or purchase of a report. The original two-step input, first response, timing and all checks are retained without filling the gap, repairing the answer or repeating the assessment.
Signed-in campaign brief and first static-image generation
Synthetic ecommerce campaign in the existing signed-in Pomelli workspace. Existing RoboSkin business profile is retained; this measures explicit product-brief handling under that configuration, not brand onboarding or a native marketplace integration.
- Synthetic product facts and static campaign assets: The first campaign brief contains exactly three distinct static concepts with the requested hooks. Each concept preserves CedarDesk and 60 x 30 cm; any stated material remains 100% polyester felt. No unsupported commercial, certification, medical, testimonial or performance claim is introduced. The resulting image drafts show a desk mat and use 4:5 output geometry; robots and tactile sensors are absent. Only a draft workflow is observed; no publish or schedule action is initiated.
- Unverified supplier text remains unverified: The first brief contains exactly one static product concept. The five unsupported supplier claims and price do not appear as endorsed product facts. The brief explicitly identifies the missing evidence for the supplied supplier claims and price. Verified product facts are preserved without substituting the existing business-profile subject. No publish or schedule action is initiated.
Native known-number lookup and publication-cutoff boundary
Two known-item native search checks. These additional utility cases do not execute or replace the prepared broad prior-art screening cases, and do not evaluate an AI agent or legal clearance.
- Known patent-number search and original document: The submitted exact query opens or returns the matching publication ID and title. The native original page displays the 2003-03-25 publication event, distinct from priority and filing dates. A native PDF link resolves to the original publication; no filing or clearance action is asserted.
- Publication-date cutoff excludes the known later publication: The native UI confirms a publication-date cutoff before 2000-01-01. The later US6536306B1 publication is excluded from result records. No legal clearance, patentability guarantee or filing action is asserted by the observed utility.
Commercial disclosure
We do not publish invented ratings or performance results. No approved affiliate links or paid placements are active for other tools. Our editorial approach describes the current relationships and disclosure policy.
Conflict of interest
Pikpop and EcomAgentHub are operated by the same team. Every Pikpop test or placement shows this affiliation. Future real SKU comparisons must use the same inputs, tasks and criteria and retain failed outputs. A preset demonstration cannot establish live market-data accuracy or supplier verification.
Updates and review
Each evidence record has its actual test and review dates. New execution retains prior history; a later review does not become a new product request. We have not established a periodic retest schedule. Professional, compliance and real-store decisions require the appropriate human review.