Greptile

Greptile is an AI code review tool that summarizes pull requests and comments inline on GitHub, GitLab and Bitbucket, billed per review credit, not per seat.

Greptile reviews pull requests before a colleague does, with the context of the codebase behind each comment. Every review opens with a plain-language account of what the change does, who it affects and why, then marks the lines that need attention. Most comments carry a code suggestion, and severity badges put the urgent findings first. Greptile is built by Tabnam, Inc., which trades under the Greptile name and is independent. The review is the core of the product. A PR Summary explains the change, an Important Files Changed table points reviewers at the files that matter, and inline comments follow on the diff. When the intent of a change is unclear, a comment can end on a clarification question. Teams steer the reviewer with custom rules and uploaded style guides, set per repository in greptile.json or per folder in a .greptile directory. Learning and custom context, the Knowledge Base MCP and cross-repo context widen what a review can draw on. Auto-approve can pass pull requests it rates as low-risk and bug-free. Greptile suits engineering teams that want every pull request read by a reviewer before a person picks it up. Teams with written coding standards can load them as rules, so each change is checked against the same conventions. Individual developers use it on their own pull requests, and maintainers of non-commercial open source projects use it on contributions. Companies that keep source code on their own servers can run the whole service in their own infrastructure, air-gapped networks included. A typical review starts when a developer opens a pull request on GitHub, GitLab or Bitbucket and Greptile's comments arrive within minutes. The author works through the findings, and one that needs more than a quick edit can go to Claude Code, Codex or Cursor through Fix with your Agent. Organizations on GitHub Enterprise Server, GitLab Self-Managed, Bitbucket Data Center, Gitea or Perforce run the same review inside their own network. Unresolved findings stay in the summary across later reviews until their threads are resolved, so an open issue is still in view at merge time.