Skip to content
Molt

Baselines

What "healthy" means for a collector, and the two moments a human should override it.

Every comparison Molt makes is against a baseline. Get the baseline wrong and every verdict downstream is wrong with it, so this page is short and specific on purpose.

The default

If nothing is explicitly pinned, Repository.getBaseline falls back to the earliest snapshot on record for that collector. For most collectors, most of the time, this is exactly right: the first successful run is what healthy looked like.

Inspecting it

Shell
molt baseline show chaos
Text
Baseline show — chaos
  collector  c_mt101cvbc0o34ghzh
 
  captured  2026-08-20T09:12:03.000Z
  rows      60
  fields    8
  source    earliest snapshot (fallback)
  id        s_3fbd21a0-…

source tells you which of the two states you are in — earliest snapshot (fallback) or explicitly pinned — because those two states call for different levels of trust in the number.

Pinning one deliberately

Shell
molt baseline set chaos

Pins the collector's most recent snapshot as the baseline. Pass a specific snapshot ID as a third argument to pin an older one instead:

Shell
molt baseline set chaos s_3fbd21a0-1234-4c9a-9abc-1234567890ab

Do this after a legitimate redesign of the target site — a genuine new normal, not a fault. Once pinned, every future comparison measures against this snapshot, not against history.

Un-pinning

Shell
molt baseline reset chaos

Clears the pin and falls back to the earliest snapshot again. Use this if a heal was accidentally pinned as a baseline, or a one-off bad run got set by mistake — the fallback is always there to land on.

Constraint

Setting and resetting a baseline are both explicit, human-invoked actions. The engine never moves a baseline on its own outside these two commands — silently moving the goalposts is the last thing a reliability tool should do unasked.

PreviousHeal & reviewNextOnboarding a target