Your build,
always at its best.

An example

From TOML to alert.

1. 01

   Define a scenario.

   An experiment.toml committed alongside your code declares what Wezel measures.

   description = "Tracks compile time of the wezel workspace"
   
   # Steps are keyed [step.<tool>.<step_name>].
   # The cargo tool builds your target and emits per-crate timing outcomes.
   [step.cargo.build]
   build_target = "workspace"
   
   # A single outcome can feed multiple summaries — Wezel tracks each independently.
   summary.workspace-build-time = { outcome = "duration.total" }
   summary.slowest-crate-time   = { outcome = "duration.max" }

2. 02

   Observe a regression.

   The CLI walks the steps, captures outcomes, and compares each summary against the trailing baseline.

   $ wezel experiment run workspace-build
   → step.cargo.build  ...........  59.7s
   → summary.workspace-build-time   59.7s   (baseline 47.3s · +12.4s · regression)
   → summary.slowest-crate-time      8.4s   (baseline  8.1s · +0.3s)
   done in 59.9s

3. 03

   Track it in CI.

   Wired into anything that runs per commit — CI, a build server, a cron — Wezel emits the alert the moment a summary drifts past tolerance.

   experiment

   workspace-build

   summary

   workspace-build-time

   commit

   a3f8c1d

   baseline

   47.3s · trailing 30 commits

   current

   59.7s · +12.4s · +26%