review

To check a pull request diff for SQL safety, you can run a pre-landing review that analyzes branch differences to detect data risks, race conditions, and LLM output trust-boundary violations. It applies a two-pass safety-first checklist to surface structural problems before merge.

An LLM trust boundary violation in code review occurs when unstructured model outputs directly trigger side effects without validation. Pre-landing diff analysis detects these issues by checking if structured intent matching and conditional side effects properly isolate model-generated content.

You detect scope drift in a pull request by comparing stated intent from PR bodies, TODOs, and commits against the actual diff changes. This sanity check identifies scope creep and missing requirements by mapping what the code actually modifies versus what was intended.

Yes, you can triage Greptile comments on a pull request automatically if they exist. The review process classifies Greptile comments into valid and actionable issues, already-fixed items, false positives, or suppressed comments to streamline pre-landing checks.

You map test coverage gaps by tracing changed code paths and user flows within the diff to identify missing test paths. This enforces regression-test requirements by linking structural modifications directly to necessary test coverage before landing.

Pre-landing code review works for enum completeness and conditional side effects by applying a two-pass safety-first checklist. The initial pass focuses on value completeness and safety, while the secondary pass evaluates conditional side effects and prompt problems.