The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data by reading generated source code scripts, modeling documentation, and candidate pools to determine how to route review tasks. While the skill has no dangerous capabilities, the ingestion of external data without strict sanitization creates a potential attack surface.
Ingestion points: SKILL.md workflow involves reading qx_method_candidates.md, model-code-analyzer.md, code scripts (.py, .m), and folder README.md files.
Boundary markers: There are no explicit instructions or delimiters defined to prevent the agent from being influenced by instructions embedded within the files it reviews.
Capability inventory: The skill is restricted to routing logic and context passing; it contains no tools or instructions for executing shell commands, performing network operations, or writing to the file system.
Sanitization: No sanitization or validation logic is specified for the content of the ingested files.

code-reviewer