Another unusual alternative is to model your test running _as_ a Docker build. It's what I ended up doing in Trellis, which supports Depot as a runner (https://

Another unusual alternative is to model your test running _as_ a Docker build. It's what I ended up doing in Trellis, which supports Depot as a runner (https://github.com/charliermarsh/trellis https://github.com/charliermarsh/trellis): running the linter check is modeled as "build the image, then add a `cargo clippy` RUN command as the last layer".