LoginSign Up
← Back to all sparks
D

Dagger

INFRA · APIS
Velocity2.5

Portable CI/CD engine using a DAG execution model with caching so pipelines run identically locally or in the cloud

dagger.io ↗

Dagger now runs the engines for you and reports CI per test, not per job.

ci-cddeveloper-toolingcachingtestingmanaged-service
◆Current state
Dagger's recent changelog holds two product moves. Cloud Engines, in September, are hosted engines with scale-out, distributed cache and observability, billed only for resources used, with Dagger claiming repeat work costs 5× less. Test-native CI, in October, tracks individual tests, so a failure names the test and only that test is run again. Around them sit cache and plumbing changes, including COPY steps that depend only on the files they copy and SSH-mounted remote filesystems.
◆Where it's heading
Both moves rest on the same idea: treat the pipeline as a graph Dagger can see into, then avoid repeating work. A layer cache that survives unrelated file changes, per-test reruns and a distributed cache all reduce what a run must redo. Cloud Engines turn that saving into something sold, the one metered offering in these entries.
◆Prediction
Expect Test-native CI and Cloud Engines to converge, with per-test reruns offered through hosted engines. That link is inferred from the two launches; the entries neither state it nor give pricing detail.

◆Recent moves

  1. 23d ago

    GitHub merge queues

    Dagger checks now run inside GitHub merge queues, so teams can mark them required and the queue only merges when they pass on the combined changes. It makes Dagger usable as a merge gate, not only as a per-pull-request check.

  2. 23d ago

    Test-native CI

    Extends the earlier test-aware work to its practical end: the failing test is named and rerun on its own. This is where Dagger's view into the pipeline turns into saved CI time.

  3. 23d ago

    Dagger for GitLab

    Dagger extends its hosted checks to GitLab merge requests, running them without a GitLab CI job or runner. It widens Dagger's reach beyond GitHub and positions it as a CI layer that sits beside the host's own pipelines rather than inside them.

  4. 23d ago

    Dagger for Entire

    Dagger adds native support for Entire, a Git forge built for agent and human contributors, so the same checks run before and after push against a shared cache. It extends Dagger's pitch as the portable CI layer for agent-driven development into a new forge.

  5. 1mo ago

    Cloud Engines

    Dagger starts running the engine itself, with scale-out, distributed cache and observability on pay-for-use billing. It is the commercial form of the caching work around it.

  6. 2mo ago

    Remote filesystems over SSH

    A Volume can now mount a remote path over SSH as a live two-way view, with no copy into the cache. It widens what a pipeline can reach without moving data.