The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes source code files, which are considered untrusted external data, creating a potential surface for indirect prompt injection. * Ingestion points: Source code files scoped for analysis, specifically data handlers and analytics initialization scripts (SKILL.md). * Boundary markers: The instructions do not define explicit delimiters to separate analyzed code from agent instructions. * Capability inventory: The skill is restricted to code analysis, grepping for patterns, and reporting findings based on a defined schema. * Sanitization: No sanitization of ingested source code is performed. * Context: This surface is inherent to the skill's function as an auditing tool and does not indicate malicious intent.

unawareness