The Agent Skills Directory

[SAFE]: The skill instructions are purely architectural and informational, focusing on designing cache invalidation and eviction strategies. It does not contain executable code, remote resource fetching, or credential handling.
[INDIRECT_PROMPT_INJECTION]: The skill includes steps to gather context and relationships from the user or external data. While this represents a data ingestion point, the absence of automated command execution or complex prompt interpolation reduces the risk to negligible levels.
[COMMAND_EXECUTION]: Although the 'Bash' tool is listed in the allowed-tools metadata, the skill body contains no bash scripts or commands that would be executed automatically.

cache-evict