The Agent Skills Directory

[SAFE]: No malicious patterns or security vulnerabilities were identified in the skill files. The instructional logic focuses on user interaction and shared understanding.
[COMMAND_EXECUTION]: The skill instructs the agent to dispatch sub-agents to gather facts from the environment (filesystem and tools). This is a functional feature for technical design and does not involve execution of hardcoded malicious scripts or access to sensitive paths.
[PROMPT_INJECTION]: While the skill ingests user input to update a design tree, it does so within a structured framework. No attempts to bypass model safety filters or override core instructions were found.

batch-grill-me