Project Overview
An offline catalog preparation system demonstrated with synthetic data. No customer information or real store writes are used.
02 / Self-initiated workflow demonstration
A repeatable preparation and review workflow for the spreadsheet, SKU and image work behind a collection launch.
Actual local validation / 29 October 2026
500 / 2,000Synthetic products / variants checked.
Zero store writes. No client data.
The demonstration measures validation coverage, not sales or hours saved.
The brief
A spreadsheet may look tidy while duplicate SKUs, missing prices or a wrong image mapping remain hidden. The objective was to make those exceptions visible before any store operation.
The implementation
A local Python CLI normalizes supported CSV headers, checks product and variant relationships, and saves the original input alongside JSON, HTML and CSV findings. A separate image plan maps filenames, SKU, alt text and rights records.
Executed inputs and outputs
| Input | Check | Observed result |
|---|---|---|
| 500 products / 2,000 variants | Default dry run | 0 errors; report and input backup; no import file |
| The same valid fixture | Explicit review export | Draft / unpublished CSV created locally |
| Three deliberate data defects | Missing price, duplicate SKU, wrong image mapping | 3 errors; export blocked |
| One generated test swatch | Image plan, dimensions and source hash | WebP copy exported; original unchanged |
From manual checking to a repeatable review
The CLI has no Shopify API writer. Image URLs are checked for syntax and mapping, not fetched or certified as available. The sample uses reserved .test URLs and must not be imported as a real catalog.
Existing handles matched against a baseline block an import candidate. This demonstration does not establish production import or rollback acceptance.
Scope / Evidence / Limits
An offline catalog preparation system demonstrated with synthetic data. No customer information or real store writes are used.
A clear exception report and an editorial case narrative make the work understandable to an operator and a buyer.
Current Shopify CSV field names, normalized aliases, consistent report columns and SKU-based image naming. Unknown fields are reported rather than silently dropped.
The web case presents key results in readable sections and a horizontally scrollable report table. CLI files remain available for spreadsheet review.
500 products and 2,000 variants in the executed fixture; price, SKU, option relationships and gallery mapping checked. Explicit export forces draft and unpublished status.
Python CLI; default dry run; immutable input copy and SHA-256; optional baseline diff; image processing reuses the existing offline asset engine.
Three injected data defects blocked export. An actual image copy was produced without changing its original. URL availability, visual accuracy, rights and production import remain separate human checks.
CLI, README, deterministic sample generator, automated tests, HTML/JSON findings, review CSV, image plan and recovery guidance. No measured labor saving or conversion result is claimed.
Start with a representative sample and an agreed list of checks.
Discuss the batch ↗