Frontend half of the 2026-05-15 amendment (backend shipped in 447f152).
- `MissionScenarioTable` component: per-scenario <table> with 7 cols
(Test | Procédure | Exécution | Source de log | Commentaires | Logs
SIEM | Cyber Incident) + Actions cell. Read mode truncates; double-
click toggles a row into edit mode where each cell becomes the right
control. detection_level lives inside the Commentaires cell as a
pill + select (no 8th column).
- MissionDetailPage Tests tab uses the new component, lifts
`editingTestId` so only one row across the whole mission is editable
at a time. Esc reverts (prompt if dirty), double-click on a different
row with a dirty draft also prompts.
- Full-bleed escape via `calc(50% - 50vw)` (same recipe as the M4 MITRE
picker). 7 dense columns breathe on wide screens, no horizontal scroll.
- `draftDiff(test, draft)` returns `null` when nothing changed → no PUT
on a no-op save. The diff carries only touched fields so the server's
per-field perm gate stays clean.
- Datetime semantics: both datetime-local inputs reuse the M7 verbatim
recipe (`iso.slice(0, 16)` + `${local}:00Z`), zero TZ shift.
Docs
- tasks/testing-m7.md §3.0 documents the column matrix + edit workflow.
- tasks/lessons.md captures the Pydantic ctx-serialisation pitfall, the
naïve-datetime guard, the table-edit pattern.
- CHANGELOG section moves "Frontend (in progress)" → "Frontend (shipped)"
and details the diff.
49 Playwright tests still green.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
9fc78e0832