The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to read and modify source code files provided by the user, which presents a surface for indirect prompt injection attacks.
Ingestion points: The skill ingest untrusted data by reading target files at paths provided via the $ARGUMENTS parameter in the 'Application steps' section.
Boundary markers: There are no explicit instructions to use delimiters or 'ignore embedded instructions' warnings when the agent processes the content of the target source files.
Capability inventory: The skill has the capability to perform file-write operations, specifically replacing JSX elements and adding import statements to the project's source code.
Sanitization: The instructions do not define any sanitization, escaping, or filtering mechanisms for the external code content before it is interpolated or processed by the agent.

ui-motion