diff --git a/cloud/apps/acceptance/desktop/run-desktop.mjs b/cloud/apps/acceptance/desktop/run-desktop.mjs
index 609918964e5..2eb45d555c0 100644
--- a/cloud/apps/acceptance/desktop/run-desktop.mjs
+++ b/cloud/apps/acceptance/desktop/run-desktop.mjs
@@ -176,8 +176,17 @@ async function launchApp(userData, label) {
})
child.stdout.removeAllListeners('data')
child.stderr.removeAllListeners('data')
- child.stdout.pipe(out)
- child.stderr.pipe(out)
+ // From here on, every app log line carries a wall-clock stamp so it can be lined up with steps.
+ const stamp = (stream) => {
+ let rest = ''
+ stream.on('data', (d) => {
+ const lines = (rest + d.toString()).split('\n')
+ rest = lines.pop() ?? ''
+ for (const l of lines) out.write(`${new Date().toISOString()} ${l}\n`)
+ })
+ }
+ stamp(child.stdout)
+ stamp(child.stderr)
log('instance', instance.cdpUrl, instance.rendererUrl, 'runner pid', child.pid)
const deadline = Date.now() + 300_000
let seen = null
@@ -603,6 +612,7 @@ async function uiDelete(step, name, label, timeoutMs = 90_000, watchAfterMs = 0)
const dialog = (await d.innerText()).replace(/\s+/g, ' ')
await shot(step, `${label}-confirm`)
const t0 = Date.now()
+ const clickedAt = new Date(t0).toISOString()
await d.getByRole('button', { name: /^Delete( Workspace)?$/ }).click()
let outcome = null
while (Date.now() - t0 < timeoutMs) {
@@ -621,6 +631,8 @@ async function uiDelete(step, name, label, timeoutMs = 90_000, watchAfterMs = 0)
}
outcome ??= { kind: 'timeout' }
outcome.ms = Date.now() - t0
+ outcome.clickedAt = clickedAt
+ outcome.doneAt = new Date().toISOString()
if (outcome.kind === 'removed' && watchAfterMs) {
// Teardown can fail after the row is gone; look for the error wherever the app puts it.
const t1 = Date.now()
diff --git a/cloud/apps/acceptance/findings.json b/cloud/apps/acceptance/findings.json
index c97a87f4eb1..a52d00d150b 100644
--- a/cloud/apps/acceptance/findings.json
+++ b/cloud/apps/acceptance/findings.json
@@ -3,7 +3,7 @@
"Wake delivery (2026-11-10, target p95 < 6 s): on golden snap_aQmW (includes #141, #151, #162), one persistent connection per wake, n=22: total p50 5.13 s, p95 6.49 s, p99 6.72 s. Just over target at p95. Phases (p50 / p95): Vercel resume to orcad spawn 2.98 / 4.27 s; orcad ready 1.36 / 1.68 s; first terminal list incl. handshake 0.43 / 0.45 s; terminal create 0.14 / 0.17 s; echo 0.14 / 0.16 s. Before: 6.8 / 11.0 s (#141 only, reconnect per call) and 7.7 / 9.5 s (pre-#141). #151/#162 cut create+echo from 1.8 s to 0.29 s. What is left is Vercel resume (about 58% of p50) and orcad ready.",
"Rows closed 2026-11-10: real 15-min idle sleep: an idle VM with no client stopped by itself after 19.2 min. Agent sign-in persistence: Claude answers, onboarding is pre-seeded, and the auth homes are kept fresh, after a wake and after a rollover. Auto-resume after Restart on m4air: one continue 29.4 s after the click, turn finished, no double continue. Wake on message: a message to a sleeping workspace was delivered in 6.1 s. Scheduled routine: fired by itself and succeeded on a fresh VM, which was then destroyed. File search: ripgrep over the pairing found both needles in 5,002 files, p50 88 ms. Usage metering: control-plane outbox, then billing hours, then the Stripe meter. Overage cap blocks new starts at $5. Dunning end to end: grace expiry (past_due_since backdated 97 h) leads the control-plane cron to suspend the running VM in 123 s; paying lets the account start again and the VM resumes. Desktop keystroke n=50: p50/p95/p99 84/94/144 ms; the 407 ms outlier did not recur. Web client is post-MVP.",
"Control-plane billing gate (closed via core #166): a dev:billing-test-* account goes through the real billing Worker. No plan → 403 no_subscription; active plan → 200; 4th concurrent VM on 1 seat → 403; included hours used up → 403; failed renewal → 403 payment_past_due; grace expired → the running VM was suspended by the cron in 71 s and resume → 403; paid → resume 200. Staging team checks pass again (teams are deleted by the shared cleanup script).",
- "CLI create (closed via W5 #167): orca worktree create --on cloud --repo … with an API key, headless: running in 3.7 s; orca cloud workspace destroy works; signed out gives cloud_not_signed_in. Move, Sleep, then Delete: no wake, destroyed, Vercel 404, no misleading toasts. The row took 31 s to disappear, against about 2 s for other asleep deletes (sent to core-desktop as a perf note).",
+ "CLI create (closed via W5 #167): orca worktree create --on cloud --repo … with an API key, headless: running in 3.7 s; orca cloud workspace destroy works; signed out gives cloud_not_signed_in. Move, Sleep, then Delete: no wake, destroyed, Vercel 404, no misleading toasts. Row removal went from 31 s to 1.5 s / 1.7 s after core #171, which skips the hooks check for asleep sandboxes.",
"orca.yaml setup in the VM (closed via core #168): an API-created workspace on test-repo branch w9-setup-script ran scripts.setup inside the VM exactly once, including after sleep/wake. The environment-build skip path is unit-tested by core; the desktop path runs after the trust prompt.",
"W1-1 (fixed, #118/#123): builds used to advance only while a client streamed /events; GET polling sat in running for 25 min. With the Durable Object runner deployed, builds finish when the client only polls GET: build 1 of a fresh environment in 17 s, build 2 in 18 s. An earlier 250 s first build had started before #123 went live. The harness now polls by default; ORCA_ACCEPTANCE_BUILD_EVENTS=1 follows the events stream.",
"D-1 (bug, core desktop, high): — fixed in #127 (bab8369a229); main-process paths verified headlessly by core against the dev control plane; desktop UI re-run pending (no-windows rule): starting an Orca Cloud create before signing in, then signing in from the card, created a sandbox owned by the fallback dev user (desktop-dogfood), not the signed-in user. Keep-alive, Copy SSH Command and Delete then got not_found. Likely cause: getHostedSandboxConfig() caches the token for the app process and does not reset it on sign-in (src/main/hosted-sandbox/hosted-sandbox-config.ts).",