The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external information about design patterns and writes it into persistent local files.
Ingestion points: Data enters the agent's context when it "learns" a new design pattern, which may originate from external documentation, web pages, or user-provided text.
Boundary markers: The instructions do not specify the use of delimiters or boundary markers to prevent the agent from obeying instructions that might be embedded within the recorded pattern descriptions.
Capability inventory: The skill relies on the agent's ability to write to the filesystem, specifically targeting patterns/SKILL.md and other markdown files in the patterns/ directory.
Sanitization: There is no requirement or guidance for the agent to sanitize or validate the content of the patterns before they are recorded, which could allow malicious instructions to be stored and later retrieved by the agent.

add-pattern