The Agent Skills Directory

[SAFE]: No malicious patterns, obfuscation, or unauthorized access attempts were detected in the skill instructions.
[PROMPT_INJECTION]: The skill accepts user input through the $ARGUMENTS variable, which creates a surface for indirect prompt injection. However, there are no instructions attempting to bypass safety filters or override agent behavior.
[EXTERNAL_DOWNLOADS]: The skill description mentions offloading data to an external cloud infrastructure for planning. This is presented as a primary feature of the skill, and the skill itself does not contain commands to download or execute arbitrary remote scripts.

gsd-ultraplan-phase