Output that follows the run
One task streams live; a broad run shows one line per task; CI shows everything.
Running one task, you want its output as it happens. Running a hundred, you want a line each and the failures in full at the end. In CI, the log is the record, so every frame stays.
vx picks the flow from what you asked for. A truthy CI always gets the CI flow, and --output-logs overrides any of it.
// what it prints
vx run @demo/web#build focused: the task’s output, live
vx run build --all broad: a line per task, failures in full
CI=1 vx run build --all CI: every frameMore in Output that fits
Choose how much you see
Full logs, errors only, just the hashes, or nothing.
A table of every task
One line per task at the end: what happened and how long it took.
Failures kept for later
A failure’s output is saved, so you or an agent can read it after the terminal is gone.
A run report for your PR
A short markdown report of the run, ready for a pull request or a job summary.
Every run as JSON
A JSON file per run, with each task’s result and the time the cache saved.
Label your runs
Tag a run, like env=nightly, and find it again in history and dashboards.
See where vx spends its time
A table of vx’s own stages, so a slow run points at its cause.
Glyphs that show the cache
Each task’s line shows at a glance whether it ran, was fresh, or came from a cache.
Folded logs on GitHub Actions
On Actions, each task is a collapsible group with its result in the title.
Colors, on your terms
Truecolor output, turned off with NO_COLOR or forced on with FORCE_COLOR.
Exit codes in plain words
A task killed by a signal says which signal, not just a number.
Plain output in a pipe
Off a terminal, vx prints plain lines and never waits on a prompt.