The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is specifically designed to handle and filter untrusted user-generated content and model responses, which creates an inherent attack surface for indirect prompt injection.
Ingestion points: The skill processes external data via the +sanitize-prompt and +sanitize-response helper commands.
Boundary markers: None defined in this high-level interface description.
Capability inventory: Invokes the gws CLI tool to perform resource management and sanitization methods.
Sanitization: The skill itself serves as the sanitization layer for other agent interactions.

gws-modelarmor