# Regression run configurations (Mobile, Desktop, Extension) These platforms ship several same-named runs distinguished only by Testmo Configuration. The API can only create step 1 of each; the step 2 UI pass is explained in the `bitwarden-testmo-tools` `creating-test-runs` skill's `references/multi-configuration-runs.md`, and each spec's `_comment` gives its exact pass. - **Mobile** — 2 runs, both named `Mobile`. Same filters, differing only by which automation type is excluded. `config_id` 1 = Android, 3 = iOS. Desktop and Extension use a **broad + narrow** split: one variant covers the full folder and the rest are driven by an "essential" tag. - **Desktop** — 3 runs, all named `Desktop`. Broad = **macOS** (21); narrow = **Windows** (22), **Linux** (20), driven by the `desktop-essential` tag (id 15999). - **Extension** — 4 runs, all named `Extension`. Broad = **`MacOS, Chrome`** (7); narrow = **`Windows, Edge`** (15), **`MacOS, Safari`** (10), **`Linux, Firefox`** (5), driven by the `extension-essential` tag (id 17190). Config and tag ids above verified 2026-08-28 via `GET /projects/1/configs` and `GET /projects/1/tags`. Every variant on both platforms, broad and narrow alike, takes a step 2. The two variant types differ only in how cases are selected: | | Broad | Narrow | | ---------- | ----------------------------------- | -------------------------------------------- | | Scope | whole top-level folder + subfolders | the `-essential` tag (server-side) | | Test Type | Smoke + Regression | **any** — the tag defines the scope | | Case state | Active | Active | | Automation | manual only (exclude 10, 23, 24) | manual only (exclude 10, 23, 24) | Narrow variants deliberately drop the test-type filter, since a tagged case counts regardless of type, but keep the state and automation-type filters so retired and already-automated cases stay out.