The Agent Skills Directory

[Indirect Prompt Injection] (LOW): The skill identifies a surface for indirect prompt injection by ingesting untrusted data from the Shopify API (products, orders, customers).
Ingestion points: Data is fetched via GraphQL queries and Webhooks as described in references/app-development.md and SKILL.md.
Boundary markers: Absent. Code snippets show direct processing of API responses without explicit delimiters for the agent.
Capability inventory: The skill uses subprocess calls via Shopify CLI and shell commands (npm, python).
Sanitization: Documentation recommends HMAC signature verification for webhooks but does not specify sanitization of data content before agent processing.
[Unverifiable Dependencies] (LOW): The skill triggers the installation of external packages.
Evidence: SKILL.md contains npm install -g @shopify/cli@latest.
Context: These are official tools from Shopify; while not on the pre-approved trusted list, they are standard for the described use case.
[Privilege Escalation] (LOW): The instruction to use global NPM installation (-g) often prompts for elevated privileges (sudo/Administrator) in many environments.
[Remote Access] (LOW): The command shopify app dev initiates a local development tunnel, which exposes a local port to the internet for webhook testing, a standard but notable network operation.

shopify