The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests user dialogue and tool schemas to evaluate agent responses. While these represent untrusted data sources, the skill is designed to enforce strict parameter discipline, which naturally mitigates attempts to inject extraneous commands via tool parameters. The validator.py script performs structural checks without sanitizing content, but its logic is restricted to safe pattern evaluation.\n- [SAFE]: All components, including the Python validation script, are focused on instructional integrity and response verification. There is no evidence of command execution, privilege escalation, or persistence mechanisms. The skill utilizes the internal skill_library.validators package, which is consistent with its stated purpose of tool call validation.

Default and optional argument discipline