Read a run

Every run says what it found: which tests ran, what failed with its error, which tests Graphide left out and why, each job’s log and timings, and the time it saved. Runs show once CI results are on for your repository.

To see what a run looks like first, open an example run, no account needed.

On the dashboard

graphide.com/dashboard lists your repositories. A repository’s page shows its last CI runs and a CI report; from there every run opens on its own page:

• The verdict, in one sentence.
• Failures first, each with its error, and marked as new, already failing on the base branch, or flaky. A job that failed without a failing test (it failed before any test ran, or outside its tests) is listed with its log.
• The tests left out, each with its reason (Which tests run, and why), and the tests that skipped themselves, in their own words.
• Jobs, timings and logs. Logs are read from GitHub when you open them, so they last as long as GitHub keeps them (90 days).
• The time Graphide saved on the run.

On GitHub

Once a repository’s results are on, each job’s GitHub summary gets a Results section that links to the run’s page.

In your terminal

gr, Graphide’s command line tool, reads the same results as your account. Install it and sign in once:

curl -fsSL https://graphide.com/install.sh | sh
gr login

Then, from a checkout of the repository:

Question	Command
Did CI pass? What failed, and what was left out and why?	gr ci show
Wait for it to finish	gr ci show --wait
Recent runs	gr ci history
What happened to one test, in this run and before?	gr ci why <test>
Every test of a run	gr ci tests
The end of a job’s log	gr ci log <job>
Find text in a job’s log	gr ci log <job> --find <text>
The end of every failed job’s log	gr ci log --failed

With no run named, a command reads the run of this checkout’s open pull request, else of its branch, and says which. Name one with --pr <n>, --branch <name>, or the run’s GitHub id and attempt (<run id>-<attempt>).

gr ci show leads with the verdict, then failures, then the commands to run next. --output json prints the answer as data, for scripts and agents.

Waiting in a script

gr ci show --wait exits with the verdict:

Exit	Means
0	Passed
1	Failed, or not known to have passed (cancelled, or lost track of before GitHub said it ended)
2	Still running at --timeout
3	No run found, or no job sent test results and none failed
4	Could not read the result: network, sign-in, results not on, or a server error

git push && gr ci show --wait --branch "$(git branch --show-current)"