review

Run the review workflow on your feature branch and it fetches the base branch, generates the full diff, and applies a checklist covering SQL safety, race conditions, and test gaps. Findings are reported with file:line citations and suggested fixes.

It checks SQL injection and N+1 queries, race conditions, unvalidated LLM output, enum completeness, conditional side effects, dead code, test gaps, crypto weaknesses, and CI/CD pipeline issues. Critical findings are separated from informational ones.

Yes, it detects the platform from the git remote URL and uses glab for GitLab or gh for GitHub to determine the target branch. If neither CLI is available, it falls back to git-native commands like symbolic-ref to find the default branch.

Yes, mechanical fixes like dead code removal, missing eager loading, stale comments, and magic numbers are applied automatically. Riskier changes involving security, race conditions, or user-visible behavior are batched into a question for your decision.

It fetches Greptile line-level and top-level comments via the GitHub API, classifies each as valid, already fixed, false positive, or suppressed, then posts evidence-based replies. Known false positives are suppressed using a per-project history file.

Automated review catches structural and pattern-based issues but cannot judge product intent, architecture fit, or visual design quality. Use it alongside human judgment for design decisions and anything requiring business context.