Search documentation

Search by page, heading, or keyword.

    Documentation menu
    Early Regression Guard

    Review your regression findings

    Every regression is grouped by the project where it was found — open one to see what changed, how to verify it, and everything needed to fix it. For how to reach this screen, see Read the result.

    What this does

    Groups every regression from a run by the project it was found in, then opens into a detail view with its own sections once you click one.

    Prerequisites

    A finished analysis run — see Read the result for how to reach this screen.

    Steps

    1. The findings list, grouped by project. A regression can surface in a different project than the one that actually changed — a change in one project can break another that depends on it, and Early lists that regression under the project that changed, tagged as a dependency finding. In the example below, backend shows 3 findings, one of which actually broke early-cli.

      Early's Regression Findings tab grouped by project: backend shows 3 findings including one tagged as a dependency finding that broke early-cli, and early-cli shows 2 of its own findings

      Each finding’s status badge tells you what it is and whether it still needs a decision:

      • Expected — Early’s own assessment that this change is not a regression.
      • Regression or Bug, with New — flagged as worth a decision, and nobody has acted on it yet.
      • Closed: — already resolved, with its reason (Fixed, Won’t fix, Fix later, Expected, or Other) shown alongside.

    2. Finding details. Click a regression to open it. The Details tab shows the business behavior before the change and after it, plus a step-by-step reproduction flow for verifying the regression — or verifying the fix, once one is applied.

      A regression's Details tab, showing the business behavior before and after the change, and numbered steps for how to verify it

    3. Investigate. This tab holds everything needed to fix the regression: why it happened, which change broke the previously-stable behavior, a root-cause analysis, and how to fix it — written out as a ready-made prompt for an AI coding tool.

      The Investigate tab, showing a structured prompt with the dependency direction, what broke, how to verify it, and the technical root cause

    Next

    Classifying, closing, assigning, and tracking a finding — covered in an upcoming release.