fix(agent): give the harness its margin, and name the mandatory points correctly

Four review findings, all correct.

`testTimeout` was set to exactly `DEFAULT_TURN_TIMEOUT_MS`, which does not fix the
race it was meant to fix — equal deadlines only make it unbiased. The harness has
to win for a slow turn to be classified as TIMEOUT instead of dying as a killed
worker, so the vitest deadline now sits 30 s above it.

`overBudget` listed shape changes, non-move events and parked-run ends, but the
mandatory set also always holds the first and last sample: a two-point track under
`maxPoints: 1` overflows for a reason the message did not name, and the model was
told something false. Both the JSDoc and the message now say mandatory points, and
enumerate all four kinds.

And two comments: the test said the shape flips every 100 ms where `sweep` samples
at 50 ms and alternates on every index, and a comment in `workbench/l0/` pointed at
`README.md`, which from there resolves to a file that does not exist.

Author: EtienneLescot

src/lib/ai-edition/timeline/cursor-track.test.ts

@@ -222,9 +222,10 @@ describe("buildCursorTrack — compression", () => {
 
 	it("says so when the mandatory points push it over maxPoints", () => {
 		// The ceiling is soft: `maxPoints` budgets the rate and the gap floor, and the
-		// points nothing can put back are exempt. A recording that flips the pointer
-		// every 100ms is all exempt points, so the budget cannot hold — and the track
-		// has to say that rather than let the model read 100 rows as "within budget".
+		// mandatory points are exempt. `sweep` samples every 50 ms and the shape here
+		// alternates on every index, so every sample is mandatory and the budget cannot
+		// hold — and the track has to say so rather than let the model read 100 rows as
+		// "within budget".
 		const flipping = sweep(200, { shape: (i) => (i % 2 === 0 ? "arrow" : "text") });
 		const track = buildCursorTrack({
 			assetId: "asset_1",

src/lib/ai-edition/timeline/cursor-track.ts

@@ -67,11 +67,12 @@ export interface CursorTrack {
 	/** True when maxPoints forced a coarser rate than `hz` would give. */
 	truncated: boolean;
 	/** Present ONLY when the ceiling did not hold. `maxPoints` budgets the rate and
-	 *  the gap floor; the points nothing can put back — a shape change, a non-move
-	 *  event, the ends of a parked run — are exempt and stack on top, so a capture
-	 *  rich in them lands above the ceiling. Absent means the budget held. It is a
-	 *  separate field from `truncated` on purpose: that one says "you are seeing
-	 *  less than you asked for", this one says the opposite. */
+	 *  the gap floor; the MANDATORY points are exempt and stack on top — the first
+	 *  and last sample, a pointer-shape change, a non-move event, the ends of a run
+	 *  longer than the max gap — so a capture rich in them lands above the ceiling.
+	 *  Absent means the budget held. It is a separate field from `truncated` on
+	 *  purpose: that one says "you are seeing less than you asked for", this one
+	 *  says the opposite. */
 	overBudget?: string;
 	/** When true, every point's virtual-timeline position equals its `atSec`, and
 	 *  `virtualSec` is omitted from the points. Goes false as soon as a clip is
@@ -328,9 +329,10 @@ export function buildCursorTrack(options: CursorTrackOptions): CursorTrack {
 	// field is absent when the budget held, so the common payload is unchanged.
 	const overBudget =
 		points.length > maxPoints
-			? `${points.length} points for a ceiling of ${maxPoints}: pointer-shape changes, ` +
-				`non-move events and the ends of a parked run are never dropped, and this ` +
-				`recording has enough of them to land above the budget.`
+			? `${points.length} points for a ceiling of ${maxPoints}: the mandatory points are ` +
+				`never dropped — the first and last sample, pointer-shape changes, non-move ` +
+				`events and the ends of a parked run — and this recording has enough of them ` +
+				`to land above the budget.`
 			: undefined;
 
 	return {

vitest.workbench.config.ts

@@ -16,12 +16,13 @@ export default defineConfig({
 		globals: true,
 		environment: "node",
 		include: ["workbench/**/*.wb.ts"],
-		// ponytail: imported, not recopied. A `.wb.ts` that drives a live turn is
-		// cut by whichever cutoff fires first, and this one used to sit at 120 s
-		// while the harness moved to 300 s — so vitest would have killed the turn
-		// before the harness could say it had timed out, which is the failure the
-		// harness comment exists to prevent. One constant, two enforcers.
-		testTimeout: DEFAULT_TURN_TIMEOUT_MS,
+		// ponytail: derived from the harness cutoff, and deliberately ABOVE it. A
+		// `.wb.ts` driving a live turn is cut by whichever deadline fires first;
+		// this one used to sit at 120 s while the harness moved to 300 s, so vitest
+		// killed the turn before the harness could classify it. Equal values would
+		// only make that race unbiased — the margin is what guarantees the harness
+		// wins and the run gets a TIMEOUT verdict instead of a dead worker.
+		testTimeout: DEFAULT_TURN_TIMEOUT_MS + 30_000,
 		reporters: ["default"],
 		// ponytail: the fixed cost of the suite is the dynamic
 		// `await import("deepagents")` in chat-service.ts:346 — hundreds of

workbench/l0/real-fixture.wb.ts

@@ -75,8 +75,8 @@ describe("la fixture réelle — le document", () => {
 	it("n'a PAS de piste caméra dans le document, quoi qu'il y ait sur le disque", () => {
 		// Le dossier d'enregistrement contient bien un fichier webcam ; le document
 		// écrit par l'app, lui, porte `cameraTrack: null`. La fixture ne corrige
-		// pas ça — voir README.md § « La prise réelle ». Cette assertion existe pour
-		// que personne ne « complète » la fixture sans s'en apercevoir : le modèle
+		// pas ça — voir workbench/README.md § « La prise réelle ». Cette assertion
+		// existe pour que personne ne « complète » la fixture sans le voir : le modèle
 		// verra `hasCameraTrack: false`, et c'est l'état réel du projet.
 		expect(realScreencastDocument().assets[0].cameraTrack).toBeNull();
 	});