The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill templates are designed to ingest and process user-supplied data through placeholders such as {user_input} in the few-shot learning section.
Ingestion points: The {user_input} placeholder located in the SKILL.md file.
Boundary markers: The template uses basic quotes to delimit input, which provides minimal protection against adversarial content.
Capability inventory: Analysis of the skill reveals no dangerous capabilities; there are no subprocess calls, file system write operations, or network requests defined.
Sanitization: There is no evidence of input sanitization or filtering to prevent the agent from following instructions embedded within the user data.

prompt-engineer