Evidence first. Commercial interests separate.

• State what a source supports—and what it does not.
• Keep vehicle applicability and uncertainty visible.
• Prefer primary records over copied summaries.
• Do not turn a code, complaint, or bulletin into a diagnosis.
• Put safety and material limitations near the decision.

Records move through source ingestion, normalization, application matching, claim drafting, evidence review, and freshness checks. A page can be technically generated without being editorially eligible for indexing.

High-consequence claims—including safety urgency, repair procedures, and applicability boundaries—receive stronger review requirements than basic navigation or glossary content.

Automated tools may help extract fields, detect revisions, cluster related complaints, or draft structured summaries. Automation is not treated as a source and does not receive its own byline.

Claims must remain traceable to source evidence. Material AI-assisted output is reviewed against the source and application before publication; unsupported inferences are removed or explicitly labeled.

Advertising, affiliate compensation, lead fees, free products, and data partnerships do not determine confidence labels, issue rankings, safety language, inclusion, or omission. Commercial links are labeled at the point of use.

Material factual corrections should update the record, checked date, and change note. Editors and reviewers should disclose relevant financial or professional conflicts and recuse themselves when independence could reasonably be questioned.