The Agent Skills Directory

[SAFE]: A comprehensive analysis was performed across all 11 threat categories, including prompt injection, data exfiltration, and remote code execution. No malicious patterns or security vulnerabilities were identified.
[INDIRECT_PROMPT_INJECTION]: The skill provides instructions for generating SQL models that process external data. It follows security best practices by explicitly demonstrating the use of Looker's built-in Liquid sanitization tags, such as {% condition %}, {% date_start %}, and {% date_end %}, which are designed to safely handle user-supplied filter inputs and mitigate SQL injection risks.
[DATA_EXPOSURE]: The skill includes example LookML definitions referencing common fields like 'email' and 'customer_id'. These are utilized as architectural placeholders to illustrate schema design and do not represent hardcoded secrets, unauthorized data access, or exfiltration mechanisms.
[DYNAMIC_EXECUTION]: The skill leverages the Looker Liquid templating engine for conditional SQL generation. This is the standard, intended use of the LookML language for building dynamic business intelligence models and does not involve the execution of arbitrary shell commands or unsafe runtime code evaluation.

lookml-ojof