What problem does it solve? Running drizzle-kit generate, push, or check in CI pipelines and other non-TTY environments raises questions about which --output mode to use, whether the CLI will prompt interactively, and how to read the missing-decisions report when it exits with code 2. ## Core Features & Use Cases - Output Mode Selection: Explains the difference between --output text (human-readable, potentially interactive) and --output json (machine-readable envelope, always non-interactive). - Interactivity Reasoning: Defines the exact rule (interactive = text output plus TTY stdin) so scripts never hang waiting for prompts. - Missing-Decisions Report Parsing: Decodes the 'missing_hints: N unresolved decisions' text report, including rename_or_create and confirm_data_loss items with their --hints snippets. - Use Case: A CI job running drizzle-kit generate fails with exit code 2; use this Skill to read the text report, build the correct --hints JSON array, and re-run non-interactively. ## Quick Start Ask the AI to explain why drizzle-kit generate exited with code 2 in CI and how to resolve the missing_hints report with the right --output mode.