FiretrailDocs
Open app

Quick start

This page takes you from nothing to your first analyzed pull request. It takes a few minutes, and you don't need to touch your CI, add a config file, or change anything in your repository.

1. Sign in

Go to the Firetrail app and enter your email. We'll send you a one-time passcode — there's no password to create or remember. Your organization is created automatically the first time you sign in, and you start on the Free plan.

2. Connect a repository

In the app, create a project and connect it to the GitHub repository you want analyzed. You'll go through GitHub's own installation flow, where GitHub shows you exactly what access you're granting.

Firetrail asks for the minimum it needs: permission to read pull requests and to post a check and a comment on them. It does not get write access to your code, and you can uninstall it from GitHub at any time. The details are in Connecting GitHub.

Right after that, Firetrail asks you to choose your Judges: the AI reviewers who will review this project's pull requests. Pick them, or skip and Firetrail locks in the default lineup. You can change it any time in project settings. On Business every reviewer reviews, so there is nothing to choose. See the review team.

3. Open a pull request

That's the whole setup. The next pull request that opens on that repository is analyzed automatically. Here's what you'll see, usually within a minute or two:

  • On GitHub — a check called "Firetrail — Future Debug Cost" appears on the PR, and Firetrail posts one summary comment with the score, the reasons behind it, and the reviewers' verdicts. The check is informational: it shows a result but never blocks the merge.
  • In the app — the PR appears in your project's Pull requests list, and clicking it opens the full analysis: the score with its itemized breakdown, what each check found, and the test-coverage estimate.

Already have open PRs you'd like analyzed? Click "↻ Sync open PRs" on the Pull requests list — it imports every open PR from the repository and queues them all for analysis.

Reading your first result

The number to look at is the Future Debug Cost — a 0-to-100 score where higher means the change has more of the patterns that tend to cause debugging pain later. But the score itself is the least interesting part. Right under it is the breakdown: a list of exactly what added points, like "+15 Multiple intents in one PR" or "+12 Complex logic change". That list is Firetrail's real output — it tells you what to look at before you merge.

Future Debug Cost explains the score in depth, including what to do at each risk level.

What's next

  • Read How analysis works to see what happens during those first couple of minutes.
  • Meet the review team — up to ten AI reviewers with different specialties who each give your PR an independent opinion.
  • Use a coding agent? Connect it to the Firetrail MCP and ask it for a Review Brief of that first PR.
  • When you're ready for more repositories or the full feature set, see Plans & billing.

Stuck? Email [email protected] — a human reads it.