Checkout reliability lab
Choose a scenario, run the simulated journey, and inspect the checks and event trace.
“Buy one medium Sand item. Apply SAVE10 and use standard shipping.”
Everyday Carry · $80.00
M / Sand in stock · Standard shipping $5.00
Rule-driven simulated agent · No account or payment
Dataset overview
Aggregations from a seeded, reproducible event pipeline. These are simulated runs, not merchant performance.
Data pipeline and event contract
How generated events are validated and aggregated.
Generate
Seeded synthetic journeys and versioned events.
JSONL / bronzeValidate
Deduplicate, type-check and quarantine malformed records.
SQLite / silverAggregate
Build scenario, funnel and daily signals from valid events.
SQL / gold → JSONEvent contract
Every generated record carries a run and event identity, a UTC timestamp, synthetic catalog identifiers, the agent action, expected and observed state, stage, outcome, amounts and latency. The evaluator compares the observed state with the business contract.
Download 40 sample events ↓event_id · run_id · scenario_id · schema_versionCONTEXTtimestamp_utc · merchant_id · product_id · variant_idBEHAVIORagent_action · stage · expected_state · observed_stateRESULTassertion_id · passed · category · currency · latency_msSimulation boundaries
This is a sandbox simulation aligned to public commerce concepts, not a live Shopify store integration or a certified OpenAI/Shopify integration. The simulated agent is rule-driven; no LLM, real customer, payment, or purchase is involved.
About the lab
The dashboard is built from a seeded synthetic dataset. Interactive runs are separate, stay in this browser tab, and do not change the dataset.
Compare successive runs to see how one injected fault changes the outcome. No account, customer data, or payment is required.