angeo-dev

Claude For Commerce Magento — Development skill for Claude Code

Development community

Magento 2 and Adobe Commerce implementation of the StorefrontBackend contract from Anthropic's Claude Commerce Agents.

How to install Claude For Commerce Magento

This entry records only its repository, not the path inside it, so there is no exact command to give. Open angeo-dev/claude-for-commerce-magento and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Claude For Commerce Magento does

Magento 2 and Adobe Commerce implementation of the StorefrontBackend contract from Anthropic's Claude Commerce Agents.

Alternatives in Development

Internet Court Skill — The trust layer for agent-to-agent commerce — natural-language mandates, ERC-7710 delegated permissions, x402 4.9k ★
Commerce Agents — Reference blueprint for building shopping and merchant agents with Claude 2.6k ★
HERO Anti OverDefense — HERO = Hashing · Edge cases · Rubrics · Overbuild — the four shapes coding agents over-defend in 333 ★

README

Claude for Commerce: Magento

A Magento 2 and Adobe Commerce implementation of `StorefrontBackend`, the one integration surface in Anthropic's [Claude Commerce Agents][blueprint] blueprint.

The blueprint ships the agent: the prompt, the tool contracts, the provenance gates, the cart caps, the evals. What it does not ship is the code that turns your store into the records the agent reads. That is one class, and for Magento this repository is it.

from magento_storefront import MagentoConfig, MagentoStorefrontBackend

backend = MagentoStorefrontBackend(MagentoConfig.from_env())

Shopify published [`Shopify/claude-for-commerce-examples`][shopify] on the same day the blueprint landed. This is the Magento equivalent, built by [Angeo][angeo] alongside the `angeo/` open-source Magento AEO and agentic commerce modules.

Status

**v0.1.0 — complete, unit-tested, not yet run against a production catalog.**

Every method of the contract is implemented and covered by tests, but those tests run against fixtures. The mapping has not yet been exercised on a live store with a real attribute set, multiple store views, or a catalog large enough to hit the variant cap. Treat the catalog mapping as the part most likely to need adjusting on your store, and run `examples/smoke.py` against it before anything else.

Findings from real deployments are welcome as issues.

What you get

Every abstract method of `StorefrontBackend`, over Magento's REST API:

Agent capability	Magento
`search_products`	`/V1/search` with `quick_search_container`, so relevance matches the site's own search page; falls back to a name and SKU filter when the search engine is down
`get_product_details`	`/V1/products/{sku}`, plus `/V1/configurable-products/{sku}/children` for a family's variants
`get_cart`, `add_to_cart`, `update_cart_item`, `remove_from_cart`	`/V1/guest-carts` for guests, `/V1/carts/mine` for signed-in customers
`get_preferences`	`/V1/custo