lookml-reviewer
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill requires the installation and use of the
looker-cli, which is referenced from the official Looker Open Source GitHub repository. This is a vendor-owned resource used for legitimate project interaction. - [INDIRECT_PROMPT_INJECTION]: The agent retrieves LookML file content via
looker-cli project file catfor analysis. This content is ingested into the agent's context without explicit boundary markers or sanitization, presenting a potential surface where malicious instructions hidden in LookML comments could influence agent behavior. - [COMMAND_EXECUTION]: The workflow constructs shell commands for the
looker-cliby interpolating user-provided inputs such as project names, branch names, and file paths. This pattern creates a potential surface for command injection if the inputs are not strictly validated or escaped by the agent platform.
Audit Metadata