The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to the processing of external product data. \n
Ingestion points: The skill retrieves product fields such as descriptionHtml, title, and vendor using the GraphQL queries defined in references/queries.md. \n
Boundary markers: There are no explicit instructions or delimiters used to isolate product descriptions from the agent's instructions. \n
Capability inventory: The agent has access to curl and the Shopify CLI; however, the skill mandates a read-only scope and strictly prohibits mutation operations. \n
Sanitization: No sanitization or filtering is performed on the product data before it is processed by the model. \n- [EXTERNAL_DOWNLOADS]: The skill suggests the installation of the Shopify-AI-Toolkit plugin. This is a reference to an official resource from a well-known service provider and does not escalate the security risk. \n- [COMMAND_EXECUTION]: The skill utilizes command-line tools (curl, shopify) to perform its audit functions. This usage is consistent with the skill's primary purpose and is governed by a 'Non-destructive doctrine' that forbids destructive actions.

shopify-catalog-audit