dive0.13.1

Dive opens a Docker or OCI image by tag, ID, or digest and shows how its contents are assembled, helping you find ways to reduce its size. Its interface breaks the image into layers and displays each layer’s files alongside the cumulative tree produced by that layer and all preceding layers. Selecting layers and navigating the tree with arrow keys lets you inspect the image interactively.

The tree marks files that were changed, modified, added, or removed, with views for one layer’s changes or the aggregate changes through that point. A lower-left pane reports basic layer details plus an experimental estimate of wasted space caused by duplicate files, files moved between layers, or incomplete removal. It expresses that estimate as both a percentage score and a total amount of wasted file space, making image efficiency easier to assess.

You can invoke Docker directly, or replace a build command with dive build -t some-tag . to build an image and analyze it immediately. Setting CI=true on a valid Dive command switches to a non-UI pass/fail result based on image efficiency and wasted space, which suits automated pipelines. The --source option selects Docker engine, the default, a Docker tar archive on disk, or Podman, with Podman available only on Linux; macOS support is limited to the Docker container engine.

It is aimed at Docker image maintainers, developers, and CI operators who need to understand layer composition and control wasted image space.