The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill instructions direct users to execute code from a third-party NPM package (shopify-mcp) and a GitHub repository (Cesarjoquin/shopify-mcp) that are not associated with official Shopify organizations or established vendors. Running code from unverified repositories poses a risk of supply chain attack.
[EXTERNAL_DOWNLOADS]: The installation process involves fetching external content via npx, git clone, and npm install. These operations download and prepare code from external registries for execution in the user's local environment.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection. It ingests untrusted data from the Shopify Admin API, such as product descriptions, customer notes, and metafields, which could be controlled by external actors. This data is then processed by the AI agent alongside powerful tools that can modify store data.
Ingestion points: Shopify Admin GraphQL API (Products, Customers, Orders, Metafields).
Boundary markers: None. The instructions do not define delimiters or provide warnings to the agent to disregard instructions found within Shopify data.
Capability inventory: Extensive write access including create_product, update_product, create_customer, fulfill_order, set_metafield, and set_inventory_quantity across multiple scripts.
Sanitization: No evidence of content sanitization or validation of input received from the Shopify API is provided in the documentation.

shopify-mcp-server