The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the looker-cli tool to interact with remote Looker instances. It executes commands for profile management, authentication verification, and running inline queries against the System Activity model to retrieve performance and governance data.\n- [INDIRECT_PROMPT_INJECTION]: The skill ingests LookML project files and System Activity data from external Looker instances. While this content is processed using shell tools like grep and jq, the ingestion of large project codebases represents a surface for indirect prompt injection. The skill mitigates this by requiring user approval before processing projects with more than 50 files.\n- [SAFE]: The skill contains explicit safety instructions titled 'Rules of Engagement' which strictly forbid the agent from using local filesystem tools (e.g., find, ls, grep on local paths) to ensure the audit remains scoped to the remote instance and does not compromise the host environment.

looker-audit