Skip to main content
Docs·Concepts·Releases

Connect releases to GitHub

Sixty seconds of setup. Then every deploy gets watched.

📖 3 min read·🎬 90s watch

The premise: every deploy is a hypothesis. "This change won't make anything worse." Sheepit tests the hypothesis automatically — by watching your error and conversion metrics for thirty minutes after the deploy and telling you if they moved.

The 60-second setup

  1. Settings → Integrations → GitHub.
  2. Click Install. Pick your repo. Authorize.
  3. Done. Sheepit watches your default branch.

What gets auto-tagged

  • Every merge to main → a release with the commit SHA, PR title, and author.
  • Every git tag matching v* → a versioned release (v1.4.2, v2.0.0-rc1).
  • Every deploy hook your CI sends → a release with whatever environment label you pass.

What you get back

  • A deploy timeline (every release, every environment, side by side).
  • Per-release crash signatures ("these five crashes started after 1.4.2").
  • Per-release metric deltas ("checkout conversion dropped 4% in the 30 min after deploy").
  • Slack alerts when a release crosses your regression threshold — at the speed you can actually do something about it.