Back to docs
journeyFirst-time installer5 min read

Getting started

Sign in, install the GitHub App, and watch your first pull request get reviewed in about five minutes.

Welcome

If this is your first time with Revvu, this page is the only one you need. We'll get you from 'I just heard about this' to 'I see review comments on my PR' in about five minutes. There's no API key to manage, no config file to write, and no infrastructure to host — Revvu is a GitHub App. Installation is three clicks plus however long it takes you to decide which repos to enable.

Before you start

You'll need a GitHub account and admin access to at least one repository. Installation takes about a minute. Your first review arrives within 60 seconds of opening or pushing to a pull request.

  • A GitHub account
  • Admin access on the repo you want reviewed (required to install a GitHub App)
  • A pull request to test against — open one or use any open PR

Step 1 — Sign in with GitHub

From the homepage, click Sign in with GitHub. You'll land on GitHub's standard OAuth approval screen — the same one every GitHub-integrated app uses. Approve it (at this stage you're only granting Revvu the ability to identify you — no repo access yet) and GitHub redirects you back to the Revvu dashboard. The dashboard is intentionally empty until you install the GitHub App in the next step.

Step 2 — Install the GitHub App

On the dashboard, click Manage GitHub App. GitHub asks where to install Revvu — your personal account or any organization you administer — and then which repositories to grant access to. Pick All repositories if you want Revvu on every repo (including ones you create later) or Only select repositories if you'd rather curate the list. You can change the selection from github.com/settings/installations at any time. The permissions screen lists the four scopes Revvu needs — see the callout below for what each one is for.

Step 3 — Open or push to a pull request

Find one of the repos you just installed Revvu on and either open a new pull request or push a commit to an existing open PR. Within seconds, the bot reacts to the PR with 👀 — that's your signal that the review is queued. About ten seconds later, a fast summary lands at the top of the PR description. Inline comments arrive over the following 30 to 90 seconds, depending on how many files changed and how much enrichment the bot pulls in.

What you'll see

Two things happen on every PR. A fast summary lands in the PR description in about 10 seconds, telling you what changed and where the risk is. A deeper review follows under 60 seconds later, posting severity-tagged inline comments on the lines that need attention.

  • 👀 reaction — The bot reacts with the eyes emoji to your PR the moment it's queued. That's your acknowledgement signal — work has started.
  • Check Run row — A row labelled 'Revvu' appears in the Checks tab of your PR. It shows 'in progress' during the review and flips to a status when the review is done.
  • PR description block — A short summary appears at the top of the PR description, fenced by HTML markers. It captures what the PR does, the key changes, and any risk areas.
  • Inline comments — Each finding lands as an inline comment on the relevant line, prefixed with a severity tag — critical, warning, suggestion, or nitpick.

If nothing happens

Most issues come from the install step skipping the repo you actually pushed to. Open the dashboard's Repos page and confirm the repo appears in the list. If it does and you still don't see a review, check the Checks tab on the PR — the in-progress row should appear within seconds of opening or pushing.

Triggering a review on demand

You don't have to push a new commit to get a fresh review. Comment on the PR with @revvu-ai please review again and the bot will queue another pass. The same mention syntax also lets you ask questions about the code or teach the bot a convention to remember next time.

What to do next

Once your first review is in, the next pages explain what's inside each comment, how to talk to the bot from PR threads, and how to tune behavior per repository.