The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill functions by reading and auditing local source code and test files, which serves as an ingestion point for potentially untrusted data if the audited project is compromised. This creates a surface where maliciously crafted files could attempt to influence the auditing agent's logic. The skill has capabilities to write to local files (during cleanup tasks) and execute shell commands (specifically for running tests using 'node --test'), as described in the 'Validate and report' section of SKILL.md. No specific delimiters or sanitization routines are defined for the audited content. This ingestion surface is typical for code-analysis tools and does not indicate an inherent security risk in the skill itself.

test-audit