The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to load project-specific context from a local file to inform schema generation.
Ingestion points: The skill references loading .agents/product-marketing-context.md for content context.
Boundary markers: No specific delimiters are defined for the ingested context file.
Capability inventory: The skill is restricted to generating static JSON-LD strings; it does not involve subprocess execution, file writing, or network operations.
Sanitization: The skill does not explicitly mention sanitizing inputs, but its output is limited to structured data templates.
[EXTERNAL_DOWNLOADS]: The documentation refers users to a well-known validation service.
Evidence: Mentions Google's Rich Results Test (search.google.com) as a tool for validating implementation.

schema-markup