Writing and verification principles
We record what was run and what remains unchecked.
Start with examples and keep the originals
Run examples in a separate practice folder. Save results to a new output path without overwriting existing results. Preview the planned changes before changing filenames.
Real files may differ in permissions, column names, encoding, units, and data size. Check with a backed-up copy first.
What do we verify?
We run executable Python examples with minimal data created for this site and check output numbers, files, preservation of originals, invalid inputs, and behavior on repeated runs. Each article records the environment and scope.
For prompts and checklists, we review the consistency of example numbers and conditions. We do not label actual AI responses or Google account integration as verified.
Verification and translation limits
Results from Windows runs are not presented as verification on macOS or Linux. We have not tested every Excel feature, large files, or network drives.
AI-translated articles are marked as translations. Native-speaker and expert reviews have not yet been completed for each language. If commands, filenames, units, or expected numbers appear to differ, check both the Korean original and the actual examples.
Synthetic automotive data is for education. Judging the safety, performance, or design suitability of a real vehicle requires expert review and actual verification.
Sources and honest explanations
We check technical facts against official documentation and link to it in each article. We do not copy long passages or code from existing documents.
We do not invent search volumes, rankings, page views, revenue, time-saving rates, author credentials, or certifications. The reference date for checking official sources is 2026-09-19.