The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to read and process external project files provided by the user, creating a surface for potential indirect prompt injection where data in those files could attempt to influence the agent's behavior.
Ingestion points: The agent is instructed to read existing project files, such as asset inventories and editorial plans, during its discovery and planning phase.
Boundary markers: The skill does not define specific delimiters or instructions for the agent to distinguish between its own logic and potentially malicious commands embedded in external content.
Capability inventory: The agent possesses the capability to write multiple markdown files to the project directory and perform web searches, which are actions that could be influenced by untrusted data.
Sanitization: The instructions do not describe methods for sanitizing or validating the content extracted from processed files.

alterlab-nmc-multimedia-story