The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface as it is instructed to process untrusted data from local project files and external web content to generate its recommendations. * Ingestion points: Instructions in SKILL.md direct the agent to read context from files such as .agents/product-marketing.md and live HTML content via the seo-audit skill. * Boundary markers: The skill lacks explicit instructions or delimiters to help the agent distinguish between data and potential embedded instructions in the external sources. * Capability inventory: The agent generates JSON-LD code blocks and provides implementation guidance based on the ingested content. * Sanitization: No sanitization or validation logic is defined for the input data to mitigate the risk of influencing agent behavior through malicious instructions.

schema