docs: record Boat staging HMR and Claude restore boundary

This commit is contained in:
Dotta committed 2026-10-11 02:20:05 -05:00
1 parent 26542bb9d0
commit b3855f10b4
8 files changed
+694 -8

No files matched your search

+32 -8
View File
@@ -2,21 +2,21 @@
**Experimental and unmerged. Local journeys and partial staging qualification passed on the sources below; final-source staging acceptance remains incomplete.** A successful build, probe, or earlier-source journey does not establish the remaining product conditions.
## Current source boundary (2026-10-11 06:13 UTC)
## Current source boundary (2026-10-11 07:04 UTC)
| Surface | Source and evidence |
| --- | --- |
| Combined implementation | `818f94594db17bb3249cadc33fd7e26535102421`, including the current-environment computer selection, safe rejection diagnostics, pending-cleanup disconnect correction, and bounded Claude executable verification after restore. At 06:11, 54 latest-head checks passed and exact-head Greptile was 5/5 with no unresolved threads. One earlier duplicate workflow was cancelled; its replacement passed. |
| Core branch | `6c2a0a2bd86ef411bbc0e273b179d5bc4f1a3733`. Canonical Boat migration 0333 follows unchanged upstream 0332. At 06:11, 54 checks passed, four were skipped, and exact-head Greptile was 5/5 with no unresolved threads. |
| Verified staging source | Deployment21 verified `388a16905b5d1924ce88a1fe550d0f65ebbaf00e` at 05:54:57.175 UTC through the official workflow. It preserves applied staging migration history. An ordinary saved Codex turn succeeded, but its next continuation and automatic retry failed with `native_session_cleanup_quarantined` before provider work. Diagnosis remains open; no quarantine was cleared. |
| Staging candidate | `991baf77cf891f3ac21aa26af7c060b4751a79df` additionally contains the disconnect correction and Claude verification fix. Deployment22 has not started. |
| Local retry | Local44 uses compatible controller `441e88c09a4b16fd4f1403e6c4e813d00653edf5` and the qualified immutable provider pack `sha256:d7275d9204e60a0a5dcb8dc536da95e945e6b90b7bdfda964d82084895ec5952`. BOA20 passed cold and warm turns with the fixed pack, then failed its ordinary post-idle continuation at installation verification. Same-run recovery also failed; diagnosis remains open. BOA19's earlier failed saved-session continuation and the database history remain preserved. |
| Combined implementation | `26542bb9d04a09a9249041ea1debe0b1001b7638`, including shared native/JavaScript restore verification and fixed-enum retirement diagnostics. Exact-head CI passed all 53 checks, and review scored 5/5 with no unresolved findings at 07:00 UTC. |
| Core branch | `b1030f6233fbea7c7a4669d567fc101ec0a85996`. Canonical Boat migration 0333 follows unchanged upstream 0332. Shared integrity verification passed 85 cases with six existing platform skips and both affected TypeScript checks. Retirement diagnostics passed 67 selected executor cases; the duplicate-event correction passed 11 cases and server TypeScript. Both changes were independently reviewed. Exact-head CI passed 54 checks with four skips; review scored 5/5 with no unresolved findings. |
| Verified staging source | Deployment22 verified `92be5d8596fd457d624280e49dd18b122ca82ee7` at 06:55:24.595 UTC through the official workflow. It includes retirement diagnostics and preserves applied staging migration history. Ordinary saved Codex continuation and state-preserving warm HMR passed. Desktop rendering passed, but exact URL navigation through computer tools failed with malformed address text; diagnosis continues. |
| Staging candidate | `a5ed24f7eeb50abd35375c6b1864bb424fe699b3` additionally includes shared JavaScript/native verification and diagnostic deduplication. Its preview-only image and migrator build passed; deployment23 was dispatched through the official helper at approximately 07:18 UTC and is not yet verified serving. |
| Local retry | Local45 uses compatible controller `3f0efd3f7274e61590cf4bf3e237b75b4f7ff167` and qualified immutable pack `sha256:4e9ede0507204690c01d943f61e5189efdfe38855f4f8952d88045ce7a71f13d`. BOA21 passed cold and warm turns and natural archive. Its post-idle session reopened successfully, but turn startup timed out before provider work; automatic cleanup recovery ended authority-indeterminate. Diagnosis is ongoing. Local44 passed cold/warm turns but failed ordinary post-idle continuation at the separate JavaScript verifier. Its failed task, pinned package and database history remain preserved. |
Staging19 establishes the native runtime, desktop, and warm development journeys on its recorded source. Staging20 subsequently passed legacy Claude and a guided legacy Codex follow-up. Latest-head CI is green, but live final-source acceptance is incomplete and the full local suite remains **34,914 passed, 24 failed, 338 skipped**. Saved-session cleanup, corrected Claude post-idle continuation, and live sibling cancellation isolation remain open. See the dated observations below for source-specific evidence and failures.
Staging19 establishes native runtime, desktop, and warm development journeys on its recorded source. Staging20 passed legacy Claude and a guided legacy Codex follow-up. Final-source live acceptance remains incomplete, and the full local suite remains **34,914 passed, 24 failed, 338 skipped**. The initiating historical cleanup-quarantine cause, corrected Claude post-idle continuation, and live sibling cancellation isolation remain open. Dated observations below preserve each source's successes and failures.
The [consolidated retry receipt](assets/2026-10-10-boat/evidence/retry-qualification-summary.json) preserves earlier sources, timestamps, failures, and receipt hashes. [Staging19 receipts](assets/2026-10-10-boat/evidence/staging19-live-evidence.json) contain the newer successes. A deployment proves serving source, not product acceptance. No provider stop was forced or cancelled, no quarantine was cleared, and no database history was reset or rewritten. Dated observations below retain their original cutoffs.
Latest automated checks: [exact-head CI and review receipt](assets/2026-10-10-boat/evidence/ci-review-0611.json).
Latest automated checks: [exact-head CI and review receipt](assets/2026-10-10-boat/evidence/ci-review-0700.json).
## Product outcomes
@@ -265,3 +265,27 @@ The warm owner retired normally and Boat archived with stop completion at 06:26:
Core `3812ccfebb` / combined `a7425b8d0c` adds fixed-enum retirement diagnostics to the existing persistent run log. It captures the actual retiring owner/generation and records whether close was never invoked, pending, completed, or failed. No retirement, settlement, grace-period, or retry decision changes. Independent review caught stale generation attribution before commit; its production-path regression now passes. All 67 selected executor cases, the eight-case affected rerun, and server TypeScript passed. Deployment22 of compatible staging source `92be5d8596fd457d624280e49dd18b122ca82ee7` is building through the official workflow, request `72d124e9-4eae-4883-930a-0449c41d55fa`. It is not yet serving or a recovery pass.
The retained pack's exact read-only verifier then isolated the rejection to `inspectCommand`, the separate JavaScript entry-point checker, with `executable_changed`. It had not yet read the native executable. A repeat verified the same JavaScript digest and then passed the native executable's first-read ctime transition through the new bounded verifier. The first-before JavaScript stat was not captured; its post-read ctime matches the first verifier's timestamp. No provider command ran in either probe. The normal viewer closed at 06:35:53. A shared bounded verification implementation is under review; this diagnosis is not a successful saved-task recovery. [Failure and diagnosis receipt](assets/2026-10-10-boat/evidence/local44-post-idle-verifier-failure.json).
## Shared verifier and deployment22 serving (2026-10-11 06:56 UTC)
Core `16f1c67531` replaces duplicate command/native integrity checks with one bounded same-descriptor primitive. Both paths retain pinned content hashes and strict launch identity. Four JavaScript regression cases failed on the prior implementation; all 85 integrity cases then passed, with six existing platform skips and both TypeScript checks passing. Independent review approved the change. Follow-up `b1030f6233` prevents duplicate immediate-close diagnostics while retaining late and distinct cleanup failures; its 11 focused cases and server TypeScript passed.
Deployment22 verified `92be5d8596fd457d624280e49dd18b122ca82ee7` at 06:55:24.595. The existing staging BOA2 task received a normal continuation after rollout, requesting actual desktop tools, the unchanged proof hash, and the existing Vite listener. Its outcome is pending. Deployment23's candidate `a5ed24f7ee` is only being prebuilt through the official preview channel, request `1e5e3b19-8fb5-4797-9a65-18589f75afe8`; no second rollout has started. Local45 selected the new qualified pack for fresh BOA21 and is preparing its first run. These are source/selection observations, not completed lifecycle acceptance.
## Staging22 saved continuation and warm HMR (2026-10-11 07:04 UTC)
After the normal controlled deployment, ordinary BOA2 continuation `2d067f32` retained the original normalized session, logical runner and provider thread. It read the unchanged proof and started the existing Vite project. The browser’s normal Computer Connect rendered the shared desktop; the normal generated preview link rendered the app. Computer navigation itself failed: the agent and live viewer both observed Chromium’s malformed omnibox text. This remains a failed navigation attempt, not a desktop-input pass. No saved authority or quarantine was reset.
The heading-only follow-up `9cc22b3c` passed real warm HMR. The open preview changed from heading `0450` to `0702` while retaining unsaved input `staging22 unsaved draft 0702` and page-session `1791702073810`. Independent samples retained owner `f414c237`, boot `91790981`, runner PID30593/start33571, Codex30796/start33705, CUA driver30866/start33714 and Vite31815/start34936. The final checkpoint was settled, with an empty outbox and no pending terminal delivery; the original proof hash was unchanged. The viewer closed normally afterward. [Before](assets/2026-10-10-boat/staging22-hmr-before.png) · [After](assets/2026-10-10-boat/staging22-hmr-after.png).
[Scoped source, lifecycle, input and HMR receipt](assets/2026-10-10-boat/evidence/staging22-live-evidence.json). Boat’s installed schema explicitly permits the lowercase hotkeys used by both attempts; the exact driver or focus cause remains unestablished.
These observations apply to deployed source `92be5d85`, not the still-building deployment23 candidate. The successful controlled-deployment continuation does not identify the initiating cause of deployment21’s historical quarantine.
## Local45 shared-verifier lifecycle (2026-10-11 07:07 UTC)
Fresh BOA21 cold run `f32eda91` and ordinary warm run `dba6f9fd` completed successfully with both original proof hashes intact. They retained normalized session `f408ef63`, logical runner `35d2bd79`, Claude thread `41a0d5cd`, physical owner `fbc45caa`, and the same boot/launch identity. [Visible warm proof](assets/2026-10-10-boat/local45-claude-warm-proof.png).
After the configured warm deadline at 07:03:57.669, the owner retired normally and Boat archive stop `stop_629e46626e5468c5` completed at 07:04:20.880. A read-only provider observation confirmed archived state before the ordinary same-task post-idle message at 07:06:44. Run `ec0b5e2b` reused the retained package and original saved identities on a new owner/boot. Session attachment and both snapshots completed, passing the earlier installation-verification rejection. The subsequent `turn.start` acknowledgement timed out after 30.003 seconds; the four committed events were session-only, with no provider turn-start or output. Automatic quarantined-cleanup recovery then ended `authority_indeterminate` at 07:12:22. **Post-idle acceptance failed.** No guest probe prewarmed the restored package, and no session reset or package replacement was used. A later normal Computer Connect enabled scoped timestamp-only diagnosis after the failed attempt had settled and Boat had archived.
The post-failure read-only diagnosis found ACPX turn preparation persisted at 07:09:59.511, but the last-request marker and Claude transcript still referred to the warm turn. This bounds the stall before the ready-for-prompt checkpoint; it does not distinguish process spawn, initialization, or session loading. Normal Computer Connect authorized that read after archive; Close Computer completed at approximately 07:17:42. No provider command was executed in diagnosis. [Lifecycle and scoped diagnosis receipt](assets/2026-10-10-boat/evidence/local45-claude-lifecycle.json) · [Visible failed continuation](assets/2026-10-10-boat/local45-claude-post-idle-failure.png).
@@ -0,0 +1,100 @@
{
"checkedAt": "2026-10-11T07:00:18.365830+00:00",
"prs": [
{
"pr": 15813,
"head": "b1030f6233fbea7c7a4669d567fc101ec0a85996",
"reviewHead": "b1030f6233fbea7c7a4669d567fc101ec0a85996",
"mergeable": "MERGEABLE",
"counts": {
"SKIPPED": 4,
"SUCCESS": 54
},
"failures": [],
"focusChecks": [
{
"__typename": "CheckRun",
"completedAt": "2026-10-11T06:49:45Z",
"conclusion": "SUCCESS",
"detailsUrl": "https://github.com/paperclipai/paperclip/actions/runs/38119091410/job/114409945153",
"name": "ci / e2e shard (8/8)",
"startedAt": "2026-10-11T06:46:02Z",
"status": "COMPLETED",
"workflowName": "PR"
},
{
"__typename": "CheckRun",
"completedAt": "2026-10-11T06:55:42Z",
"conclusion": "SUCCESS",
"detailsUrl": "https://app.greptile.com/paperclip-org-3/github/paperclipai/paperclip/-/pull/15813",
"name": "Greptile Review",
"startedAt": "2026-10-11T06:45:25Z",
"status": "COMPLETED",
"workflowName": ""
}
],
"pending": [],
"unresolved": [],
"moreThreads": false,
"greptileSummary": [
{
"url": "https://github.com/paperclipai/paperclip/pull/15813#issuecomment-6099291617",
"updatedAt": "2026-10-11T06:55:49Z",
"reviewedCommits": [
"b1030f6233fbea7c7a4669d567fc101ec0a85996"
],
"score": [
"5"
]
}
]
},
{
"pr": 15804,
"head": "26542bb9d04a09a9249041ea1debe0b1001b7638",
"reviewHead": "26542bb9d04a09a9249041ea1debe0b1001b7638",
"mergeable": "MERGEABLE",
"counts": {
"SUCCESS": 53
},
"failures": [],
"focusChecks": [
{
"__typename": "CheckRun",
"completedAt": "2026-10-11T06:50:34Z",
"conclusion": "SUCCESS",
"detailsUrl": "https://github.com/paperclipai/paperclip/actions/runs/38119092419/job/114409948224",
"name": "ci / e2e shard (8/8)",
"startedAt": "2026-10-11T06:46:13Z",
"status": "COMPLETED",
"workflowName": "PR"
},
{
"__typename": "CheckRun",
"completedAt": "2026-10-11T06:52:00Z",
"conclusion": "SUCCESS",
"detailsUrl": "https://app.greptile.com/paperclip-org-3/github/paperclipai/paperclip/-/pull/15804",
"name": "Greptile Review",
"startedAt": "2026-10-11T06:45:27Z",
"status": "COMPLETED",
"workflowName": ""
}
],
"pending": [],
"unresolved": [],
"moreThreads": false,
"greptileSummary": [
{
"url": "https://github.com/paperclipai/paperclip/pull/15804#issuecomment-6098591173",
"updatedAt": "2026-10-11T06:52:06Z",
"reviewedCommits": [
"26542bb9d04a09a9249041ea1debe0b1001b7638"
],
"score": [
"5"
]
}
]
}
]
}
@@ -0,0 +1,198 @@
{
"recordedAt": "2026-10-11T07:00:01.646823+00:00",
"taskId": "ef14981f-cdab-42ef-a963-309c052e5808",
"identifier": "BOA-21",
"controllerSource": "3f0efd3f7274e61590cf4bf3e237b75b4f7ff167",
"fixSource": "16f1c67531ea1e46fe93ece4326c982bb53e5f04",
"providerPackDigest": "sha256:4e9ede0507204690c01d943f61e5189efdfe38855f4f8952d88045ce7a71f13d",
"coldAndWarmRuns": [
{
"id": "f32eda91-446b-4eae-bacc-ff95211302f4",
"status": "succeeded",
"created_at": "2026-10-11T06:45:42.106Z",
"started_at": "2026-10-11T06:45:42.143Z",
"finished_at": "2026-10-11T06:56:59.596Z",
"error_code": null,
"error": null,
"native_session_id": "f408ef63-1ad4-4229-9bf0-2eb8ecdbe5ad",
"runner_instance_id": "35d2bd79-6e2c-4e09-a375-8bfc132ab3a2",
"session_id_before": null,
"session_id_after": "41a0d5cd-fa74-46bd-b098-0167e80c6758",
"execution_stage": "preparing"
},
{
"id": "dba6f9fd-3537-490b-ac47-b455b8a0d388",
"status": "succeeded",
"created_at": "2026-10-11T06:57:42.335Z",
"started_at": "2026-10-11T06:57:42.372Z",
"finished_at": "2026-10-11T06:58:57.712Z",
"error_code": null,
"error": null,
"native_session_id": "f408ef63-1ad4-4229-9bf0-2eb8ecdbe5ad",
"runner_instance_id": "35d2bd79-6e2c-4e09-a375-8bfc132ab3a2",
"session_id_before": "41a0d5cd-fa74-46bd-b098-0167e80c6758",
"session_id_after": "41a0d5cd-fa74-46bd-b098-0167e80c6758",
"execution_stage": "preparing"
}
],
"warmOwner": {
"id": "fbc45caa-0c3b-40f5-8e1c-5736dbfa31b2",
"kind": "runner",
"phase": "warm",
"runId": "dba6f9fd-3537-490b-ac47-b455b8a0d388",
"deadline": "2026-10-11T07:03:57.669Z",
"generation": 2,
"process": {
"nonce": "f3633c99-0dd1-49d3-aa74-e3f1ecdd7180",
"bootId": "1ecbeef9-edbe-4347-ab7a-c24e0a528441",
"unitName": "paperclip-fbc45caa-0c3b-40f5-8e1c-5736dbfa31b2.service",
"launchGeneration": 1
}
},
"liveRunnerExecutableProof": {
"sandboxId": "bx_ndnggjad",
"exitCode": 0,
"timedOut": false,
"data": {
"checkedAt": "2026-10-11T06:57:03.136619+00:00",
"ownerId": "fbc45caa-0c3b-40f5-8e1c-5736dbfa31b2",
"bootId": "1ecbeef9-edbe-4347-ab7a-c24e0a528441",
"launchGeneration": 1,
"pid": 57415,
"startTicks": "86874",
"runnerSha256": "8f90f35f94f2266b0a441a0013b73c3a7ba1f736de6a21b541c96da0f8de5f05",
"runnerBytes": 37554768,
"expectedHashMatches": true,
"admissionStages": [],
"journalExitCode": 0
},
"error": null
},
"postArchiveContinuation": {
"status": "failed",
"runId": "ec0b5e2b-aea3-42d3-b2ac-f0a0cf714393",
"nativeSessionId": "f408ef63-1ad4-4229-9bf0-2eb8ecdbe5ad",
"runnerInstanceId": "35d2bd79-6e2c-4e09-a375-8bfc132ab3a2",
"providerThreadBefore": "41a0d5cd-fa74-46bd-b098-0167e80c6758",
"providerThreadAfter": "41a0d5cd-fa74-46bd-b098-0167e80c6758",
"physicalOwner": {
"id": "0dfa2b99-4fa1-476e-908f-9fdf69347a80",
"kind": "runner",
"phase": "active",
"runId": "ec0b5e2b-aea3-42d3-b2ac-f0a0cf714393",
"deadline": null,
"generation": 1,
"process": {
"nonce": "d4efda7e-739b-4284-ab6c-e87b8784f7d9",
"bootId": "2968e236-5a78-4193-9e6c-a113bb913074",
"unitName": "paperclip-0dfa2b99-4fa1-476e-908f-9fdf69347a80.service",
"launchGeneration": 1
}
},
"selectedProviderPackDigest": "4e9ede0507204690c01d943f61e5189efdfe38855f4f8952d88045ce7a71f13d",
"sessionReopen": "resumed",
"primaryFailure": {
"observedFinishedAt": "2026-10-11T07:10:53.266Z",
"code": "provider_transport_failed",
"command": "turn.start",
"cause": "ACPX sidecar request timed out at turn.start",
"lastAdmissionStage": "ready",
"lastAdmissionElapsedMs": 2520,
"sessionOnlyEventTypes": [
"session.resumed",
"session.capabilities.updated",
"session.goal.snapshot",
"run.attached"
],
"providerOutputObserved": false,
"turnStartIssuedAt": "2026-10-11T07:09:59.312Z",
"transportCloseStartedAt": "2026-10-11T07:10:29.315Z",
"elapsedUntilTransportCloseMs": 30003
},
"automaticRecovery": {
"finishedAt": "2026-10-11T07:12:22.183Z",
"code": "adapter_failed",
"message": "runner_state_identity_mismatch: authority_indeterminate",
"identityRotated": false,
"succeeded": false,
"kind": "scheduled_quarantined_cleanup_recovery"
},
"guestObserverReadsBeforeAdmission": false,
"fixtureResetOrManualRetry": false,
"acceptance": "not_passed"
},
"observerPolicy": "After the cold runner hash at 06:57:03, no guest observer reads occurred through the failed post-idle admission and cleanup. Later bounded read-only diagnosis used a normal viewer after terminal failure (07:16:10\u201307:17:17); viewer closed normally 07:17:42. No provider commands, resets, or cache mutations.",
"naturalArchive": {
"owner": {
"checkedAt": "2026-10-11T07:06:28.446Z",
"owners": [
{
"id": "fbc45caa-0c3b-40f5-8e1c-5736dbfa31b2",
"kind": "runner",
"phase": "retired",
"runId": "dba6f9fd-3537-490b-ac47-b455b8a0d388",
"deadline": null,
"generation": 2,
"process": {
"nonce": "f3633c99-0dd1-49d3-aa74-e3f1ecdd7180",
"bootId": "1ecbeef9-edbe-4347-ab7a-c24e0a528441",
"unitName": "paperclip-fbc45caa-0c3b-40f5-8e1c-5736dbfa31b2.service",
"launchGeneration": 1
}
}
]
},
"provider": {
"checkedAt": "2026-10-11T07:06:28.906Z",
"id": "bx_ndnggjad",
"state": "archived",
"health": "ok",
"hydrated": true,
"stop": {
"id": "stop_629e46626e5468c5",
"status": "completed",
"requestedAt": null,
"lastAttemptAt": "2026-10-11T07:04:20.880Z",
"endedAt": "2026-10-11T07:04:20.880Z",
"error": null
}
},
"guestReadsSinceWarmTurn": false
},
"updatedAt": "2026-10-11T07:12:53.893381+00:00",
"proofFiles": [
{
"name": "boat-final-claude.txt",
"sha256": "14fa23911a1a95ede22a93f43fc027934b20a1ba497f8e6120abc8769498e124",
"verifiedIn": [
"cold",
"warm"
]
},
{
"name": "boat-claude-framing-proof.txt",
"sha256": "afaf525f4bf0c38614c5df0cf97140e7d1bf9cad96faa293e2317dc82116af4a",
"verifiedIn": [
"cold",
"warm"
]
}
],
"postFailureReadOnlyDiagnosis": {
"viewerOwnerId": "020eea2f-f880-44c7-9c6f-4d6d78135611",
"observedAt": "2026-10-11T07:17:17.209Z",
"viewerClosedAt": "2026-10-11T07:17:42Z",
"acpxLastPromptAt": "2026-10-11T07:09:59.511Z",
"acpxLastRequestMatchesFailedTurn": false,
"acpxAgentStartedAt": "2026-10-11T06:58:37.408Z",
"claudeLastAssistantTimestamp": "2026-10-11T06:58:56.236Z",
"claudePostIdleMessagesObserved": false,
"runnerLifecycle": "suspended",
"providerLifecycle": "closed",
"activeTurnId": null,
"providerExitUnconfirmed": false,
"conclusion": "Turn preparation persisted; live ready checkpoint was not observed. The exact spawn/initialize/session-load/ownership blocker remains undetermined.",
"providerExecution": false,
"guestFileChanges": false
}
}
@@ -0,0 +1,364 @@
{
"schemaVersion": 1,
"recordedAt": "2026-10-11T07:08:10.911066+00:00",
"source": {
"deployment": 22,
"servingCommit": "92be5d8596fd457d624280e49dd18b122ca82ee7",
"verifiedAt": "2026-10-11T06:55:24.595Z",
"status": "succeeded",
"verification": "verified"
},
"scope": {
"sandboxId": "bx_b8sawfyp",
"companyId": "76e38a59-75e7-5ff8-a4bf-ebb78a7b8ebe",
"environmentId": "8c78970d-76b4-42e5-b45e-eca4deb85a67",
"agentId": "d7a6fbcf-b0aa-4449-bdc0-875cbdf2a246",
"taskId": "7d3472d4-461f-44ce-90ed-139e878df200"
},
"results": {
"ordinarySavedContinuation": {
"status": "passed",
"runId": "2d067f32-3614-41eb-8f6d-670502aff86c",
"observedAt": "2026-10-11T07:01:34.625704+00:00",
"normalizedSessionId": "06496696-b312-4268-ac40-66770aaa6f1a",
"logicalRunnerId": "0005a613-101d-475f-9409-8b8e8b87e9ce",
"providerThreadId": "01a12881-5d3b-72a0-b4c3-d993f5b089e1",
"matchesPreviouslyRecordedSavedIdentity": true,
"proofSha256": "f9253e57dadaefe9276e99a9cd0cb9fad68d2704dd820b61e026edf6e6527c46",
"proofBytes": 24,
"proofMatchesOriginal": true,
"runnerLifecycle": "ready",
"providerLifecycle": "session_open",
"outboxCount": 0,
"pendingTerminalDeliveryIsNull": true,
"completedTurnAuthoritative": true,
"controllerDeploymentRecovery": "Ordinary same-task continuation after controlled deployment22 preserved the original saved identities and proof; no manual task/session reset was used.",
"limit": "This is controlled-deployment recovery, not an active controller-crash test. The initiating deployment21 cleanup-quarantine cause remains unknown."
},
"warmHmr": {
"status": "passed",
"runId": "9cc22b3c-ba7d-4f01-bb5a-7367349f68aa",
"beforeObservedAt": "2026-10-11T07:01:34.625704+00:00",
"afterObservedAt": "2026-10-11T07:03:36.891899+00:00",
"sameProcessIdentities": true,
"sameBoot": true,
"bootId": "91790981-bff6-43a8-8f82-ef7d4c397df8",
"ownerId": "f414c237-68a1-4451-9545-dd8342865c04",
"processes": [
{
"role": "runner",
"pid": 30593,
"startTicks": 33571
},
{
"role": "codex",
"pid": 30796,
"startTicks": 33705
},
{
"role": "computer_use_driver",
"pid": 30866,
"startTicks": 33714
},
{
"role": "vite",
"pid": 31815,
"startTicks": 34936
}
],
"sameNormalizedSession": true,
"sameLogicalRunner": true,
"sameProviderThread": true,
"proofUnchanged": true,
"vitePort": 5173,
"vitePid": 31815,
"postTurnRunnerLifecycle": "ready",
"postTurnProviderLifecycle": "session_open",
"postTurnOutboxCount": 0,
"postTurnPendingTerminalDeliveryIsNull": true,
"postTurnCompletedTurnAuthoritative": true,
"browserObservation": {
"source": "Parent normal browser observation and before/after screenshots",
"observedAtApprox": "2026-10-11T07:03:15Z",
"headingBeforeSuffix": "0450",
"headingAfterSuffix": "0702",
"unsavedDraftBeforeAndAfter": "staging22 unsaved draft 0702",
"pageSessionBeforeAndAfter": "1791702073810",
"reloadPerformed": false
}
},
"computerNavigation": {
"status": "failed_to_reach_target",
"runId": "2d067f32-3614-41eb-8f6d-670502aff86c",
"target": "https://example.com/",
"parentVisualObservation": "Connected desktop showed Chrome New Tab with malformed omnibox suggestion; the agent reported navigation failure.",
"actualToolCallsObserved": true,
"exactTranscriptThreadId": "01a12881-5d3b-72a0-b4c3-d993f5b089e1",
"transcriptTimeRange": [
"2026-10-11T06:55:00Z",
"2026-10-11T07:01:00Z"
],
"inputSummary": [
{
"timestamp": "2026-10-11T06:58:48.249Z",
"tool": "launch_app",
"arguments": {
"name": "Google Chrome"
}
},
{
"timestamp": "2026-10-11T06:58:53.945Z",
"tool": "bring_to_front",
"arguments": {
"pid": 31558,
"window_id": 29360131
}
},
{
"timestamp": "2026-10-11T06:58:53.947Z",
"tool": "click",
"arguments": {
"pid": 31558,
"window_id": 29360131,
"x": 210,
"y": 50,
"delivery_mode": "foreground"
}
},
{
"timestamp": "2026-10-11T06:58:54.053Z",
"tool": "hotkey",
"arguments": {
"pid": 31558,
"window_id": 29360131,
"keys": [
"ctrl",
"l"
],
"delivery_mode": "foreground",
"x": 210,
"y": 50
}
},
{
"timestamp": "2026-10-11T06:58:54.056Z",
"tool": "hotkey",
"arguments": {
"pid": 31558,
"window_id": 29360131,
"keys": [
"ctrl",
"a"
],
"delivery_mode": "foreground",
"x": 210,
"y": 50
}
},
{
"timestamp": "2026-10-11T06:58:54.066Z",
"tool": "press_key",
"arguments": {
"pid": 31558,
"window_id": 29360131,
"key": "BACKSPACE",
"delivery_mode": "foreground"
}
},
{
"timestamp": "2026-10-11T06:58:58.831Z",
"tool": "type_text",
"arguments": {
"pid": 31558,
"window_id": 29360131,
"x": 210,
"y": 50,
"text": "https://example.com/",
"delivery_mode": "foreground"
}
},
{
"timestamp": "2026-10-11T06:58:58.833Z",
"tool": "press_key",
"arguments": {
"pid": 31558,
"window_id": 29360131,
"key": "ENTER",
"delivery_mode": "foreground"
}
},
{
"timestamp": "2026-10-11T06:59:09.016Z",
"tool": "get_window_state",
"arguments": {
"pid": 31558,
"window_id": 29360131,
"include_screenshot": true,
"max_elements": 200,
"max_depth": 8
}
},
{
"timestamp": "2026-10-11T06:59:38.075Z",
"tool": "bring_to_front",
"arguments": {
"pid": 31558,
"window_id": 29360131
}
},
{
"timestamp": "2026-10-11T06:59:40.727Z",
"tool": "click",
"arguments": {
"pid": 31558,
"window_id": 29360131,
"x": 194,
"y": 50,
"delivery_mode": "foreground"
}
},
{
"timestamp": "2026-10-11T06:59:43.716Z",
"tool": "hotkey",
"arguments": {
"pid": 31558,
"window_id": 29360131,
"keys": [
"ctrl",
"l"
],
"delivery_mode": "foreground"
}
},
{
"timestamp": "2026-10-11T06:59:46.100Z",
"tool": "hotkey",
"arguments": {
"pid": 31558,
"window_id": 29360131,
"keys": [
"ctrl",
"a"
],
"delivery_mode": "foreground"
}
},
{
"timestamp": "2026-10-11T06:59:47.901Z",
"tool": "press_key",
"arguments": {
"pid": 31558,
"window_id": 29360131,
"key": "BACKSPACE",
"delivery_mode": "foreground"
}
},
{
"timestamp": "2026-10-11T06:59:49.856Z",
"tool": "type_text",
"arguments": {
"pid": 31558,
"window_id": 29360131,
"text": "https://example.com/",
"delivery_mode": "foreground"
}
},
{
"timestamp": "2026-10-11T06:59:52.223Z",
"tool": "press_key",
"arguments": {
"pid": 31558,
"window_id": 29360131,
"key": "ENTER",
"delivery_mode": "foreground"
}
},
{
"timestamp": "2026-10-11T07:00:00.468Z",
"tool": "get_window_state",
"arguments": {
"pid": 31558,
"window_id": 29360131,
"include_screenshot": true
}
}
],
"hotkeySchemaObservation": {
"at": "2026-10-11T07:07:05.335650+00:00",
"initializeSucceeded": true,
"toolsListSucceeded": true,
"inputActionsRequested": false,
"documentedExample": [
"ctrl",
"c"
],
"actualHotkeys": [
[
"ctrl",
"l"
],
[
"ctrl",
"a"
]
],
"lowercaseModifierAdvertisedValid": true,
"foregroundModeAllowed": true,
"optionalCoordinatesAdvertisedForFocus": true
},
"limits": [
"Correct target text and advertised-valid lowercase hotkeys were sent. The exact driver or focus failure is not established.",
"The second navigation attempt sent calls sequentially several seconds apart, so the first tightly grouped calls alone do not explain all observed failure.",
"Actual computer-tool invocation is proved; successful navigation is not claimed for this source."
]
}
},
"viewerLifecycle": {
"normalCloseAt": "2026-10-11T07:03:59Z",
"source": "parent normal UI closure",
"postCloseArchiveStatus": "not_observed_in_this_receipt"
},
"evidenceCollection": {
"processAndCheckpointMethod": "Guarded read-only provider status and scoped guest metadata under the existing product-authorized runner.",
"inputDiagnosisMethod": "Exact saved provider thread; only computer-tool inputs inside the current run time window, followed by installed-driver MCP tools/list. No computer input actions issued by the observer.",
"observerLifecycleMutation": false,
"observerSourceEdits": false,
"guestReadsStoppedAfterSchema": true
},
"limitations": [
"Staging21 quarantine initiating cause remains unknown.",
"This receipt does not repeat all four runtime qualifications on deployment22.",
"Sibling-stop isolation and post-viewer normal archive are not proved here.",
"Configured warm expiry timestamp was not exposed by the guest checkpoint."
],
"evidence": [
{
"name": "staging-deploy-22-result.json",
"sha256": "ead8a73e2962ae0df086dad7abd9d43a4b257c1d4597b0b49c8babb1564847df"
},
{
"name": "retry-staging-22-nativecodex-baseline.json",
"sha256": "eba1d355e26b6f93f98a15c36759ce2f9075073e8b816bdc0be74897637ebc96"
},
{
"name": "retry-staging-22-nativecodex-hmr-settled.json",
"sha256": "37131bc94b52a9fc7aeeee52e8fe1a6816cd12c68fb947dafdc91f34a82588a4"
},
{
"name": "retry-staging-22-computer-inputs-filesystem.json",
"sha256": "4013f380975fe925b64d18e9502058e274479195b3c1f1cfa5ed44e0714f47be"
},
{
"name": "retry-staging-22-hotkey-schema.json",
"sha256": "73af47e83a5fab0df16f67bc707f5eb99561f27293a84d6658a110a18e5c3e89"
},
{
"name": "staging22-hmr-before.png",
"sha256": "4e2178e2ddd2ff5a53d071a94ed22440e10e9e89a2a761dcadde8e2c8bd86be3"
},
{
"name": "staging22-hmr-after.png",
"sha256": "eaabee0be9a4325e24ec9cf1134d252e5246292e8b185c35bd39531e8fd3546f"
}
]
}
Binary file not shown.

After

Width:  |  Height:  |  Size: 93 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 87 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 19 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 19 KiB