The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill downloads and installs the uv tool from astral.sh using a piped-to-shell execution pattern in scripts/preflight.sh. This is a standard installation method for this well-known developer tool and is considered a safe setup operation.
[EXTERNAL_DOWNLOADS]: The scripts/setup_repo.sh script installs several companion skills and the lkr-dev-cli tool directly from the author's (lkrdev) GitHub organization. These resources belong to the same vendor as the skill and are part of the required infrastructure.
[COMMAND_EXECUTION]: The workflow relies on various system tools, including lkr, gh, jq, and npx, to automate git operations, Looker project configuration, and GitHub Pull Request creation.
[CREDENTIALS_UNSAFE]: The skill handles sensitive Looker API credentials and GitHub tokens by reading them from local .env files and synchronizing them with GitHub Secrets via the gh secret set command. This is used to enable automated LookML validation in CI/CD environments and follows standard security practices for secret management.
[INDIRECT_PROMPT_INJECTION]: The skill consumes untrusted SQL queries and business logic provided by the user to generate LookML code. It maintains clear boundaries between user input and execution context, using standard tool interfaces for processing, which presents a low risk to the agent's autonomous behavior.

agentic-lookml-workflow