Observed arrival · 2026-09-08
Sinter Separates Flaky CI Tests from New Bugs
A private-alpha tool for Elixir teams that uses test history to classify pull-request failures as known flakes or possible regressions.
Field notes
The system is split into an open recording layer and a hosted diagnosis layer. Temper writes ExUnit outcomes and commit identifiers to JSONL, then a CI step pushes the history; Sinter compares outcomes across retries and partitions without running tests or reading source code. The page illustrates the method with an ImportTest marked 23% flaky and a ParserTest marked 8% flaky, and lists Elixir-specific causes such as sandbox ownership and order dependency.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
One card from the complete issue