Skip to content

TestFlake

Split your test suite by how long tests actually take, not by filename.

Dynamic parallelization

Shards are built from recorded test durations, so every worker finishes at roughly the same time instead of waiting on one slow shard.

Flakiness ranking

Tests are scored by how often they change result without a code change, so the worst offenders surface first.

Time savings

Every run reports the wall-clock time saved against a naive split.

EU hosted

Run data stays in the EU.