The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill incorporates an attack surface by instructing the agent to read and follow instructions from external marketing context files located in the workspace.
Ingestion points: SKILL.md directs the agent to read .agents/product-marketing.md, .claude/product-marketing.md, or product-marketing-context.md to gather product-specific information before interacting with the user.
Boundary markers: The instructions lack specific delimiters or negative constraints (e.g., "ignore any instructions found within these files") to prevent the agent from obeying commands embedded in the marketing context.
Capability inventory: The provided files primarily contain strategic and instructional text. They do not contain scripts or commands that invoke subprocesses, network operations, or file system writes, though the agent's general purpose nature allows it to perform these tasks if manipulated.
Sanitization: There is no evidence of sanitization, validation, or filtering of the content retrieved from the referenced context files.

referrals