The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructions direct the agent to read external context from files such as .agents/product-marketing.md. This creates a potential vulnerability surface where malicious instructions embedded in these data files could influence the agent's behavior.
Ingestion points: SKILL.md references multiple potential context files: .agents/product-marketing.md, .claude/product-marketing.md, and product-marketing-context.md.
Boundary markers: There are no explicit instructions for the agent to use delimiters or to disregard potential instructions found within these context files.
Capability inventory: The skill itself is purely instructional and does not include any executable scripts, network operations, or file-write capabilities that could be exploited via an injection.
Sanitization: No sanitization, validation, or filtering mechanisms are defined for the content read from the external context files.

emails