Trunk CI

Trunk CI records what your CI actually did. Every workflow run, every job, every step, the time each one spent waiting for a runner, and which runner picked it up. You get two things out of it:

A view on every pull request. Which checks ran on which commit, which one is red, and a waterfall of the whole run so you can see what was slow and what was waiting.
An org-wide view of your CI. A list of every job across every repository, ranked by run volume, duration, and pass rate - plus what is running and queued right now.

What it collects

Trunk builds a small model out of what your provider reports:

Trunk concept	GitHub Actions	Buildkite
Repository	The repository	Parsed from the pipeline’s git remote
Pipeline	A workflow file	A Buildkite pipeline
Pipeline run	A workflow run	A build
Job	A job under `jobs:`	A step
Job run	One execution of that job	A Buildkite job
Step	A step under `steps:`	N/A

Spans are written as a run progresses rather than at the end, so a run is visible while it is still executing.

Get set up

Ingestion is per repository and off by default. Turn it on for one repository, push a commit, and the data appears on the next run.

Set up CI

Connect GitHub Actions or Buildkite.

What you can do with it

CI on a pull request

Which checks ran on which commit, and what is blocking the merge.

The CI tab

Every job in your organization, with pass rate, duration, and run volume.

Commit traces

The waterfall of everything that ran on one commit.

​What it collects

​Get set up