shopify-custom-data
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- Instructional Reinforcement Patterns: The skill uses highly imperative language and explicit overrides, such as 'TAKE PRECEDENCE', 'ESSENTIAL RULES', and 'NEVER, EVER'. While these are intended to guide the agent toward technically accurate Shopify implementations, they are structurally similar to prompt injection techniques used to override agent behavior.
- Standard CLI Command References: The documentation instructs the agent to recommend the use of Shopify CLI commands (e.g.,
shopify app init,shopify app dev,shopify app deploy). These are standard tools for the platform but involve the execution of commands within the user's development environment. - Indirect Prompt Injection Surface: The skill provides a surface for processing user-supplied data into Shopify Metafields and Metaobjects. This presents a potential area for review regarding how data is sanitized before being included in generated GraphQL mutations.
- Ingestion points: The skill processes developer requirements from the user prompt to generate data definitions and values (SKILL.md).
- Boundary markers: No explicit boundary markers or warnings are included in the generated examples to distinguish between instructions and data.
- Capability inventory: The skill facilitates the generation of network-capable GraphQL mutations (
metafieldsSet,metaobjectUpsert) for writing data to the Shopify Admin API. - Sanitization: There are no specific instructions provided for the validation or sanitization of values interpolated into the generated mutations.
Audit Metadata