dmlab / module-catalog-enrich

0.0.3 2026-10-30 21:54 UTC

Requires

• php: ~8.2.0||~8.3.0||~8.4.0||~8.5.0
• dmlab/module-typesafe-core: ^0.0.2
• magento/framework: ^103.0.7
• magento/framework-message-queue: ^100.4.7
• magento/module-backend: ^102.0.7
• magento/module-catalog: ^104.0.7
• magento/module-config: ^101.2.7
• magento/module-eav: ^102.1.7
• magento/module-indexer: ^100.4.7
• magento/module-mysql-mq: ^100.4.5
• magento/module-store: ^101.1.7
• magento/module-ui: ^101.2.7
• psr/log: ^2.0||^3.0
• symfony/console: ^6.4||^7.4

Requires (Dev)

• phpunit/phpunit: ^11.5||^12.5

OSL-3.0

• typesafe
• filters
• PIM
• ai
• magento2
• magento 2
• magento2-module
• layered navigation
• llm
• product attributes
• jev
• attribute enrichment
• catalog enrichment
• product data quality

Published on Packagist.org on 2026-10-30 21:55 UTC

README

AI product attribute enrichment for Magento 2: fill empty filter attributes (material, sleeve, seats, Yes/No…) from the product descriptions you already have.

Products with an empty Material, Sleeve or Number of seats never show up for shoppers who filter by them. Catalog Enrich fills those attributes from the names and descriptions you already have, picking only from the attribute's existing options. You choose per attribute: record only, wait for review, or apply automatically — and every applied value can be reverted.

Install

composer require dmlab/module-catalog-enrich
bin/magento module:enable DmLab_TypeSafeCore DmLab_CatalogEnrich
bin/magento setup:upgrade

Requirements

Magento 2.4.7+ · PHP 8.2–8.5

• Answers currently come only from TypeSafe (model Jev), through dmlab/module-typesafe-core (pulled in automatically).
• A TypeSafe API key from console.typesafe.ai; usage is billed by TypeSafe. The key is stored encrypted in your Magento; spend is capped by the Monthly Token Budget next to it. A run that hits it stops; product-grid runs are re-queued hourly, the nightly run resumes the next night.
• Queue consumers run by cron_consumers_runner (or started manually with bin/magento queue:consumers:start).

Quick start

1. Stores → Configuration → DMLab → TypeSafe — enable, paste the API key, click Test Connection.
2. Catalog → Catalog Enrichment → Rules → Add New Rule — pick an attribute. New rules start in Shadow (answers are recorded, nothing is suggested).
3. Run it: bin/magento dmlab:catalog-enrich:run --limit=100, or select products in the product grid → Actions → Enrich attributes, or enable the nightly cron.
4. bin/magento dmlab:catalog-enrich:calibrate --attribute=<code> measures precision on products that already have a value and suggests a threshold.
5. Switch the rule to Review and approve in Catalog Enrichment → Suggestions — or to Auto.

Rules: status and outcomes per attribute	Rule: threshold and extra rules

Extra Rules are the main accuracy lever: one plain sentence per line that tells the model how your catalog names things ("A tank top is sleeveless.").

Reference

Attributes	select (≤ 254 options), multiselect, Yes/No
Writes	default (store 0) values only: empty values, or missing options appended to a multiselect; skips products with a store-view value, configurable variant attributes and values a human rejected
Rules	mode off · shadow · review · auto; threshold 0.5–1 (default 0.9); fill empty_only or add_missing (multiselect); Treat No as Empty (Yes/No)
Admin	Catalog → Catalog Enrichment: Rules, Suggestions (approve / reject / revert), Attempts (Ask again on next run)
CLI	dmlab:catalog-enrich:run [--attribute] [--sku] [--limit] [--force] [--dry-run] · calibrate --attribute [--sample] · report
Settings	Stores → Configuration → DMLab → Catalog Enrichment: extra text attributes, products not visible individually, nightly cron and products per run
Background	queue consumers dmlab.catalog_enrich.products and dmlab.catalog_enrich.suggestions; nightly cron (off by default); hourly re-queue of stopped product-grid runs
ACL	DmLab_CatalogEnrich::rules, ::suggestions, ::suggestions_apply, ::attempts, ::run, ::config

A product × attribute is sent again only when its text, the options, the rule text or the model change — or when you ask (--force, Ask again). Data TypeSafe refuses is not retried until it changes; other errors back off from 1 hour up to 7 days.

Pairs with dmlab/module-typesense-search: filled attributes become layered-navigation filters on a Typesense engine.

Data and privacy

Sent to TypeSafe (US-hosted API): product name, short description and description (HTML stripped, up to 6,000 characters), the extra text attributes you chose, the attribute's label and option labels, and your rule text. No customer or order data. Zero data retention is available on TypeSafe's enterprise plan.

Architecture

• Questions — select and Yes/No are one TypeSafe Choice asked twice, in source and reversed option order, with a "not stated" option last in both; a value is accepted only when both answers pick the same option with the lower probability ≥ the threshold. A multiselect is one Noul per option. Yes/No is never a Noul, so silence is "not stated", not "No".
• Dedup — identical requests are answered from dmlab/module-typesafe-core's response cache. The attempt ledger stores one row per product × attribute with a fingerprint of everything sent. Threshold, mode and fill mode are not part of it: stored answers are re-decided locally.
• Queues — topics dmlab.catalog_enrich.products (product ids from the grid action) and dmlab.catalog_enrich.suggestions (approve / revert batches) on the MySQL db connection. Grid batches share the run lock with CLI and cron; a stopped run (budget, key, circuit) is deferred to the hourly re-queue cron.

License

OSL-3.0