The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze external codebase content, which serves as untrusted input. Maliciously crafted code or comments within the target files could attempt to influence the agent's behavior.
Ingestion points: The skill instructions involve using Read, Grep, and Glob tools to scan user-provided directories and files.
Boundary markers: No explicit delimiters or instructions are provided to the agent to treat audited content strictly as data, which is a known vulnerability surface for indirect prompt injection.
Capability inventory: The skill utilizing the Bash tool provides access to shell-level commands, which increases the potential impact if an injection were successful.
Sanitization: Content from the audited files is processed without filtering or sanitization.

security-audit