The Agent Skills Directory

[COMMAND_EXECUTION]: The skill makes extensive use of the looker-cli utility to manage Looker resources. These actions include creating directories (project directory create), uploading files (project file create), validating projects (api project validate_project), and importing dashboards (dashboard import). These are legitimate administrative tasks within the context of Looker development.
[DATA_EXFILTRATION]: The skill uses looker-cli query runquery to execute SQL queries associated with dashboard tiles. While this is intended to verify that the dashboard functions correctly, it involves retrieving data from the connected database into the agent's context.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8) because it derives LookML content and CLI arguments from user-provided feedback and discovery specifications.
Ingestion points: User feedback regarding chart types, metrics, and layout, as well as initial project specifications from the conversation history, are used to generate LookML files and command-line parameters (e.g., {dashboard_name}).
Boundary markers: No explicit delimiters or instructions are provided to the agent to ignore potentially malicious commands embedded in the user's feedback.
Capability inventory: The skill has the capability to write files to the server, execute shell commands via looker-cli, and run arbitrary database queries.
Sanitization: There is no mention of sanitizing or validating user-provided strings before they are interpolated into LookML definitions or CLI file paths.

creating-looker-dashboard