The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion and processing of data stored in JSON/JSONB database columns (e.g., configuration in app/models/product.rb or line_items in app/models/order.rb). While this data could originate from untrusted external sources, the skill demonstrates the use of standard ActiveModel boundary protections.
Ingestion points: External data enters the application through JSON database columns and is processed by StoreModel::Model classes in files like SKILL.md and references/advanced-patterns.md.
Boundary markers: The skill does not explicitly define LLM boundary markers as it is focused on Rails backend logic rather than prompt construction.
Capability inventory: The skill uses database write capabilities (product.save!) and database query capabilities (Product.where).
Sanitization: The skill promotes the use of ActiveModel::Validations (e.g., validates :name, presence: true, numericality) to sanitize and validate incoming JSON data before it is persisted.
[SAFE]: The skill provides a legitimate technical guide for a well-known Ruby gem. It includes security best practices, such as an example of overriding as_json to exclude sensitive fields (e.g., secret_key) from being exposed during serialization.

store-model-coder