github code-yeongyu/oh-my-openagent v5.1.81

one hour ago

Changed

  • The programming skill's language gate now routes rendered web work (.html, .css, .jsx, .tsx, .vue, .svelte, .astro, or any task that ships a UI) to the frontend skill before the first line of UI code. It tells the agent to take the look from the frontend skill's references when there is no design direction (no named brand, reference, DESIGN.md, or existing styled UI to match), and to finish only after visual-qa passes on the rendered surface. (#14604, fixes #14602)

Fixed

  • Memory sync: a sign-out now retires the trust pin of the account it signs out of, so omo account show --account <id> no longer answers joined once that sign-in's keys are gone. Plain omo account forget (the desktop's sign-out of a CLI-item session) used to retire nothing. It now reads the stored sign-in first to find its account. A stored sign-in that cannot be read is still deleted and its pin is left; one warning line is printed only after the delete succeeds. A forget that fails on a locked store reports only that error (#12308, #12384). (#13642) (#14533)
  • Memory sync: some sign-ins earned their pin before omo sync recorded their account. On a refused refresh, plain forget, forget --device and adopt --replace, omo now asks the service for that account, as omo logout already did, and retires the pin when the service names it. It asks only while a live pin exists and the access token has not expired; on a refused refresh, the token may still be valid in the 30 s before expiry, when the refresh runs. An expired token is never sent. A pin whose account stays unknown is kept, and omo prints one line naming omo account retire-sync --account <id> to clear it, because omo account show keeps answering joined for it (#12383, #13662). (#13642) (#14533)

Added

  • omo sync status (and --json) shows this device's memory sync state: the signed-in account, whether this device has joined and whether its sync switch is on, the account's devices with their sync membership and last-seen time, pending join requests, and whether an engine is running on this machine. It writes no sync state, prints no key material, and each section reports its own failure instead of failing the command. (#14437, fixes #14435) (#14533)

  • 097ce1e Merge pull request #14604 from code-yeongyu/skill/programming-frontend-pointer

  • e9aa3b0 docs(programming): keep an existing styled UI as design direction; fix the row reference for non-script files

  • 9e448c6 Merge pull request #14576 from code-yeongyu/ci/platform-red-first-red

  • a4edd85 docs(changes): fragment for #14604

  • eaef6a3 docs(programming): route rendered web work to the frontend and visual-qa skills (#14602)

  • a321e72 Merge pull request #14580 from code-yeongyu/train/14555-windows-platform-red

  • 7fa1efb Merge pull request #14575 from code-yeongyu/ci/speed-critical-path

  • 9b5c5ba ci(platform-red-tracker): a passing macOS status job stays success when macos-combined ran

  • 013af34 ci(knip): declare jq, which script/platform-jobs.test.ts runs

  • ea1a2fe ci(platform-red-tracker): a macOS status job proves only what macos-combined ran

  • 9312ca7 Merge remote-tracking branch 'origin/fix/14555-sync-status-unreadable-pin' into HEAD

  • 5d841a2 Merge remote-tracking branch 'origin/fix/inactive-backlog-windows-ebusy' into HEAD

  • 784106c Merge remote-tracking branch 'origin/fix/14556-win-service-auth' into HEAD

  • c2d9e6c ci(platform-red-tracker): read the platform job classification from one jq file

  • b0a5e83 ci(platform-red-tracker): name the merge whose dev run first failed each macOS/Windows job

  • d50841a test(memory-sync): run the search-permission case of sync status only where mode bits exist

  • 01cb170 style(service-auth): format the platform-gated tests

  • 0eaecf3 Merge pull request #14533 from code-yeongyu/train/memsync-cli-4

  • 88b8a40 fix(thread): finalize seeding statements so gateway tests release the db on Windows (#14555)

  • 13d18f6 chore(train): regenerate bundles after merging dev 09342bc (v5.1.80 source state)

  • 9e5b23c test(service-auth): run the real Linux state file only on a POSIX host (#14556)

  • 99bed2e Merge dev 09342bc (v5.1.80 source state) into train/memsync-cli-4

  • 09342bc Merge pull request #14524 from code-yeongyu/release/v5.1.80-source-state

  • 10d8c5f chore(train): regenerate bundles after merging dev 8032185 (train 11)

  • eb53e9c Merge dev 8032185 (train 11) into train/memsync-cli-4

  • 8032185 Merge pull request #14537 from code-yeongyu/release-train/omo-20261121-thread-train-11

  • 8a8311a docs(changes): fold the train members' fragments into one train fragment (14533.md)

  • 60411ef docs(changes): one package-log block per log in the train 11 fragment (#14537)

  • 7e07cbf docs(changes): fold the member fragments into the train 11 fragment (#14537)

  • ac9ae57 Merge PR#14521 @b73da831ba into thread train 11

  • c63476b Merge PR#14482 @dd22c8ef7b into thread train 11

  • d3520ca Merge PR#14487 @00b14e9845 into thread train 11

  • d1f5139 Merge PR#14480 @fba8b162b5 into thread train 11

  • c06ef71 Merge PR#14502 @8790887279 into thread train 11

  • 961948d Merge PR#14514 @0b74338b74 into thread train 11

  • d604802 Merge PR#14416 @74b903a2d2 into thread train 11

  • 29bc0b1 Merge PR#14513 @c00e6420f5 into thread train 11

  • 74dff18 Merge PR#13640 @ee29e3c501 into thread train 11

  • b1391d5 Merge PR#14518 @0e3d300fdd into thread train 11

  • f6e2522 Merge PR#14536 @45942f1199 into thread train 11

  • 04a66b8 Merge PR#14532 @96d4e9b7f9 into thread train 11

  • bdb3b0e Merge pull request #14517 from code-yeongyu/release-train/omo-20261121-thread-train-10

  • 45942f1 test(thread): settle the post-dispose call with settled(), not expect().rejects (gateway matcher ban)

  • 96d4e9b chore(knip): drop the removed liveRecordRpcChildPids from the pre-existing unused-export list (#14529)

  • 268368e chore(bundles): regenerate omo.js, omo-wake.js and thread-sdk for the store dispose fix (#14535)

  • 485b5ea add changelog fragment for #14536

  • 682d026 fix(thread): gateway store dispose lets in-flight calls settle before it closes the worker (#14535)

  • 96d4092 docs(changes): #14437 changelog names service-auth/devices.ts listLiveDevices (review LOW L1)

  • a764529 chore(train): regenerate bundles for memsync-cli train v4 (#14437, #13642, #14434)

  • 7e3e132 train: merge #14434 (81455ea)

  • 0a10f1a train: merge #13642 (8325120)

  • 310e3d2 train: merge #14437 (dc83319)

  • dd22c8e style(thread): format the rematch regression test

  • b848160 fix(thread): rethrow only model_state_unavailable as the engine's refusal on a failed rematch (review of #14482)

  • 00b14e9 fix(thread): keep the #14270 misplaced-file hint when a device answered a silent lookup (review L1 of #14487)

  • 732f3d3 Merge remote-tracking branch 'origin/dev' into fix/14477-silent-probe-reason

  • 65399ad fix(thread): keep the engine's set_model refusal on a failed rematch and document both new outcomes (review L1, L2 of #14482)

  • fba8b16 build(service-auth): regenerate bundles for the desktop legacy label rename (#14475)

  • 0267aa5 Merge remote-tracking branch 'origin/dev' into fix/14445-set-model-provider-suffix

  • 2aa99a8 Merge remote-tracking branch 'origin/dev' into fix/14475-desktop-legacy-device-label

  • dc83319 Merge origin/dev 4a8c752 into feat/memsync-sync-status-verb

  • 0e3d300 chore(omo-native): regenerate service-auth at the train 9 top (collapse_key from #14431 reaches the service-auth bundle)

  • b73da83 changelog: fragment for #14521

  • c9fe020 test(thread): type the unnamed-row assertions and format

  • c00e642 build(omo-sdk): regenerate the omo SDK bundle

  • a01b6b5 Merge remote-tracking branch 'origin/dev' into fix/14389-reland-ultrawork-arming

  • a8b7054 fix(omo-sdk): the sign-in refusal's next action names no command

  • c7195ff fix(thread): name an unnamed row from the folder its path names

  • 445942c fix(thread): resolve a listing row's device by exact id only (#14305)

  • 74b903a Merge remote-tracking branch 'origin/dev' into refactor/fleet-hint-omo-relay

  • 68f9d13 test(remote): pin candidateLines' return value; reword a title off the deleted resolver

  • 92d14db docs(changes): fold the #14469 and #14431 fragments into the train 9 fragment (#14518)

  • 4fd3aa2 docs(changes): fold the #14440 fragment into the train 10 fragment (#14517)

  • 593c73e chore(bundles): regenerate generated bundles once at the train 9 top

  • 920dc69 Merge remote-tracking branch 'origin/dev' into wt/13640-linux-state-file

  • db27bac Merge PR#14440 @2e376e8eef into thread train 10

  • f63b181 Merge remote-tracking branch 'origin/dev' into fix/11632-dag-cases-in-memory-journal

  • 0498050 test(thread): an unnamed row names its path's folder (RED)

  • 3fc29b2 test(thread): a listing row's device resolves by exact id only (RED)

  • b1f1ace Merge dev (after #14501) into train 9

  • 4a8c752 Merge pull request #14501 from code-yeongyu/chore/regen-generated-14468

  • d2f1b01 docs(changes): fragment for #14514

  • a392978 docs(changes): fragment for #14513

  • 8f064be build(omo-sdk): regenerate bundles on dev; drop four unused imports in account-cli

  • 6c2d88d Merge dev @486d98b992 (train 8 #14415) into the review follow-up

  • be08f4a chore(bundles): regenerate on top of dev 486d98b; changelog fragment (#14468)

  • 21695b0 feat(omo-sdk): the sign-in verb answers a side-effect-free --check probe

  • f53f382 feat(omo-sdk): the sign-in verb takes its options as one JSON line on stdin

  • 02ec428 build(omo-sdk): regenerate service-auth and the omo SDK bundle; docs

  • 5f2eb92 style(omo-sdk): format

  • b8f8143 feat(omo-sdk): sign in by device code and rename a device

  • 252aed4 test(omo-sdk): device-code sign-in verb, SDK signIn and devices.rename (RED)

  • 6960e48 Merge remote-tracking branch 'origin/dev' into chore/regen-generated-14468

  • 4ba1d78 Merge dev (after #14415) into train 9

  • 486d98b Merge pull request #14415 from code-yeongyu/release-train/omo-20261119-thread-train-8

  • bfd9e5c Merge pull request #14467 from code-yeongyu/fix/14465-brand-names-plugin

  • 6e91ae7 Merge remote-tracking branch 'origin/dev' into wt/13640-linux-state-file

  • cefb8a8 Merge #14431 (d36dacd) into train

  • d4f483c Merge #14469 (726a3f8) into train

  • 02bc927 chore(bundles): regenerate on top of dev 3fe56c0 (#14468)

  • 2be368c Merge remote-tracking branch 'origin/dev' into chore/regen-generated-14468

  • baa0266 Merge remote-tracking branch 'origin/dev' into fix/11632-dag-cases-in-memory-journal

  • 3fe56c0 Merge pull request #14495 from code-yeongyu/fix/11632-scheduler-test-in-memory-journal

  • 80dd931 add changelog fragment for #14502

  • fa645fb chore(bundles): regenerate the stale generated bundles and bundle inputs on dev (#14468)

  • f949b96 chore(changes): fold #11624's fragment into #14495 (one fragment per PR)

  • 52e9574 test(dag): journal the e2e ordering and missing-skill cases in memory (#11632)

  • 2e37731 test(dag): build in-memory store paths with the file store's own builder; refresh the timeout rationale

  • e91b151 style(dag): bun fmt the in-memory store, store contract and store validator exports

  • 018ec10 add changelog fragment for #14495

  • 916c8b4 chore(knip): drop the stale dagKeyHash pre-existing row; the in-memory store imports it

  • 9e1a91a test(dag): journal the scheduler cases in memory (#11632)

  • 4a50ec3 Merge remote-tracking branch 'origin/fix/11621-infra-retry-in-memory-journal' into fix/11632-scheduler-test-in-memory-journal

  • dc3a1b9 Merge pull request #14486 from code-yeongyu/fix/14470-serial-reads-causal

  • 648b521 build: regenerate the generated artifacts after merging dev (bundle inputs, service-auth, extensions, codex install)

  • 67bbff2 Merge origin/dev into fix/14465-brand-names-plugin; regenerate plugin/scripts/install.mjs (Set B)

  • 806fb95 Merge remote-tracking branch 'origin/dev' into wt/13640-linux-state-file

  • dd7d69f Merge dev @0b56df2abb into thread train 8

  • a31c37d test(thread): the unnamed-row fixture names its folder in path (#14415)

  • d36dacd Merge remote-tracking branch 'origin/fix/14428-inactive-target-backlog' into fix/14428-thread-backlog-drain

  • 726a3f8 test(omo-native): pin the second store-ops isClaimantDead call site (#14466 review)

  • 0b56df2 Merge pull request #14448 from code-yeongyu/fix/14430-error-streak-alert

  • 3207ee9 Merge remote-tracking branch 'origin/fix/14428-inactive-target-backlog' into fix/14428-thread-backlog-drain

  • 259725f Merge dev @af161f9a4d (#14458) into thread train 8

  • 145fe6c test(thread): seed the inactive-target backlog in one transaction instead of per-test timeouts (#14466)

  • af161f9 Merge pull request #14458 from code-yeongyu/fix/14429-reopen-unknown-session

  • a2842eb docs(changes): add fragment for #14487

  • 649186c fix(thread): preserve silent lookup reasons

  • 80d5731 fix(thread): restore model state when the thinking suffix is refused

  • 1f7936b test(thread): serial-reads test asserts the order causally, not by one sample (fixes #14470)

  • 870abf9 test(thread): reproduce partial switch on thinking refusal

  • 695635b test(thread): reproduce dropped silent lookup reasons

  • 1acefa9 docs(changes): fragment for #14482

  • 6b3405f Merge origin/dev into fix/14465-brand-names-plugin

  • 3d4af11 fix(thread): apply model thinking suffixes and identify absent providers

  • aca7ff7 docs(changes): #14480

  • 0d2e6f1 fix(service-auth): rename a desktop-generated legacy device label on refresh (#14475)

  • abf0a19 test(service-auth): a desktop-generated legacy label is renamed on refresh (#14475, RED)

  • 5ac71bd Merge #14469 (4a60b50) into #14431: expired and dead-claim rows leave the per-lane backlog (#14466)

  • 3663db8 test(thread): reproduce target model suffix and provider misses

  • 2c9ce28 fix(thread): listing names an unreadable store in backlog_error; tool-level collapsed test (#14431 review 3)

  • ba35e2e Merge remote-tracking branch 'origin/dev' into fix/14428-thread-backlog-drain

  • f22daf7 fix(thread): listing survives an unreadable store, an empty lane fits any accepted delivery, absorbed repeats say collapsed (#14431 review 2)

  • 81455ea test(memory-sync): drop the repository reference from the pair-shape comment

  • 00eaeb0 build: regenerate omo-device-account and omo-feedback after the store helpers went private

  • c3dcfcb fix(thread): collapse key keeps the queued row's marker on rollback and every send's receipt (#14431 review)

  • 1d91444 chore(knip): the state-file fixture is a by-path entry, mkfifo a known test tool, and the store's file helpers stay private

  • f5bce41 Merge dev @6555140b8b (#14447) into thread train 8

  • e3ccdcc build: regenerate omo-device-account and omo-feedback after merging dev

  • 3a38140 Merge remote-tracking branch 'origin/dev' into refactor/fleet-hint-omo-relay

  • 211fffa chore(bundle): regenerate omo-task.js for #14437's busy-host probe delta on dev 6555140

  • f8d8e3d Merge remote-tracking branch 'origin/dev' into fix/14429-reopen-unknown-session

  • 7bb318d Merge remote-tracking branch 'origin/dev' into feat/memsync-sync-status-verb

  • 6257fa6 Merge remote-tracking branch 'origin/dev' into d14/win32-pipe-pair-tests

  • 5649123 Merge remote-tracking branch 'origin/dev' into fix/signout-pin-retire-12308

  • de4ba9b Merge remote-tracking branch 'origin/dev' into wt/13640-linux-state-file

  • c6a3eb1 Merge remote-tracking branch 'origin/dev' into fix/14430-error-streak-alert

  • 6555140 Merge pull request #14447 from code-yeongyu/fix/14427-fmt

  • 4a60b50 fix(thread): leave expired intake rows to the target's reconcile (#14466 review)

  • 3b82895 test(thread): fill the target cap through store intake, one call per row (#14428)

  • f3dbf5e test(thread): fill the target cap with one drain pass instead of a timeout bound (#14428)

  • 3df4b16 test(thread): bound the full-backlog cases explicitly (#14428)

  • b662fd2 Merge remote-tracking branch 'origin/dev' into fix/14428-thread-backlog-drain

  • 22aec0f test(thread): busy-backlog QA scenario, v16 migration expectations and knip entries (#14428)

  • a9d9e32 docs: record the bundle and category verification refresh (#14447)

  • f0fdf2e Merge remote-tracking branch 'origin/dev' into refactor/fleet-hint-omo-relay

  • 3e8101a docs(changes): fragment for #14469

  • acafacd fix(thread): let an inactive target's backlog expire instead of staying queue_full (#14466)

  • 26d8041 chore(knip): drop the stale completionOutcome pre-existing row (#14430)

  • b4aebab docs(changes): fragment for #14467

  • 659fa1d fix(omo-native): name the injected plugin in the brand profile so an attaching start does not refuse it (#14465)

  • 849bdb1 test(thread): cover the empty-path fallback too (#14458 delta review LOW)

  • b274aa2 docs(changes): fragment for #14458 (#14429 review M1)

  • a56afdb Merge dev @03423f1943 into thread train 8

  • d37ab58 Revert "Merge PR#14392 @f4243bd92e into thread train 8": #14392 leaves train 8 (#14415)

  • f67df1a Merge remote-tracking branch 'origin/dev' into fix/signout-pin-retire-12308

  • 4c123c6 fix(thread): drop an ended session's streak state; document that operator_alerts is read once (#14430 review r2 LOWs)

  • 6096ca1 chore(bundles): omo.js budget = 1565721 B measured on the dev merge (e1004d4) + 4 KiB

  • 03423f1 Merge pull request #14462 from code-yeongyu/feat/14442-deepseek-gpt-ultrawork

  • 6bda73c docs(changes): fragment for #14462

  • f87b57c feat(ultrawork): DeepSeek V4.1 Flash receives the GPT ultrawork prompt (#14442)

  • b3aeaf0 fix(thread): thread_handoff's receipt reports the working folder too (#14429 review M1)

  • ca51612 Merge remote-tracking branch 'origin/dev' into fix/14430-error-streak-alert

  • e1004d4 Merge pull request #14460 from code-yeongyu/feat/14441-ultrawork-gpt-directive

  • 8306277 docs(changes): fragment for #14460

  • 28e258d chore(bundles): regenerate for the GPT ultrawork directive; omo.js budget = this regen (1559896 B) + 4 KiB

  • db42f0a feat(ultrawork): default and GPT directives as two whole prompts

  • 5cb0adb test(eq): re-record EQ.omo E21-E24 goldens for exact GPT-6.2 category gates

  • 0098c77 fix(thread): read operator_alerts from the user layer only; make the alert component test order-independent (#14430)

  • 76a8ae5 Merge remote-tracking branch 'origin/dev' into fix/signout-pin-retire-12308

  • efef34e chore(omo-native): regenerate service-auth bundle for sync-creation guidance

  • 96bd3f4 fix(thread): a thread on another device reports its working folder, not the folder's name (#14429)

  • 382c700 build: bundle-inputs lists the two new service-auth sources

  • 1cdd811 test(audit): treat codex-model-catalog.ts as a model-identity file

  • 74b493d fix(service-auth): the account lookup of a discard gets the fix-up budget, not the 30 s request timeout (#13642 review L1)

  • d151e64 fix(memory-sync): sync status tells a refused socket from an unreadable one and names the probe's idle-clock effect

  • 38c59e4 build: regenerate the service-auth bundles from this source

  • 230b526 fix(memory-sync): sync status reads engine presence without attaching, shares the device read, and reports each section's own failure

  • 940885c fix(service-auth): report a kept pin only when one is live, and warn that it may belong to another signed-in account (#13642 review L1, L2)

  • 6822512 test(memory-sync): cite the desktop pair contract with its repository (review L2)

  • 416983d test(service-auth): the state-file launcher runs are beforeAll setup, not a raised test timeout

  • 39c4c01 build: keep dev's generated bundles; the release train regenerates them

  • 906da17 Merge branch 'dev' into fix/7185b-linux-no-keyring-state-file

  • 2bb3bfd changes: fragment for #14448

  • 013cd15 chore(bundles): omo.js budget = 1556648 B measured with #14430 + 4 KiB

  • 293610e feat(thread): alert the operator when a session ends 10 runs in a row with an error (#14430)

  • 74f9081 test(memory-sync): the operator socket decides no_engine

  • 7ca726e style: format tests merged unformatted by #14427

  • 2e376e8 docs(changes): fragment for #14440 (#14305 U7b)

  • 6de1bf5 Merge dev @8b75b271f2 (#14427) into thread train 8

  • 689f7c9 docs(changes): name the sync status fragment after #14437

  • c02d720 test(thread): RED error-streak detector driven by an injected clock (#14430)

  • 40e5623 docs(changes): sync status fragment

  • 4d8ed83 feat(memory-sync): omo sync status shows this device, the account's devices, pending joins and the engine

  • a652bb1 fix(service-auth): a kept sync pin of an unknown account is reported with the retire-sync command (#13662)

  • 8b75b27 Merge pull request #14427 from code-yeongyu/fix/14155-exact-gpt62-defaults

  • b9c468b fix(routing): use exact GPT-6.2 defaults for deep categories and Codex

  • 8f8fa82 test(memory-sync): run the three channel tests on win32 over the fd 5/6 pipe pair (#12698)

  • 135b0ac Merge remote-tracking branch 'origin/dev' into fix/signout-pin-retire-12308

  • 6761972 docs(changes): fragment for #14431

  • 7baf0c1 fix(thread): steer reserve, collapse key and backlog facts for a target that stops taking its queue (#14428)

  • e1177b6 chore(omo-native): regenerate service-auth at the thread train 8 top after dev @8ed8b4ccdc (#14415)

  • bc6f016 Merge dev @8ed8b4ccdc into thread train 8

  • 6e175be fix(thread): turn_conflict's details name the row by its key; candidateLines has one doc (#14305 U7b)

  • 1ab29be test(thread): RED - turn_conflict's details name the keyless row; the keyless tests assert typed fields only (#14305 U7b)

  • 0031186 test(config): exercise current Astra IDs in config and rules fixtures

  • f1f7350 Merge remote-tracking branch 'origin/dev' into feat/14305-u7b-relay-transport

  • 8ed8b4c docs: remove sponsor section from all README translations

  • 89ff247 Merge pull request #14425 from code-yeongyu/fix/memory-sync-cli-refusal-20261120

  • bf41a6b docs(changes): note CLI sync setup guidance

  • 9041d75 fix(omo-native): explain first-device sync refusal by reason

  • 74bb9e0 Merge remote-tracking branch 'origin/dev' into refactor/fleet-hint-omo-relay

  • b036a83 Merge pull request #14412 from code-yeongyu/fix/14384-windows-test-paths

  • 3b89490 docs(changes): fold the #14406 and #14374 fragments into the train 8 fragment (#14415)

  • 558c172 Merge dev @2818426be3 into thread train 8

  • cc1815c test(eq): the eval tool description carries the shortened fleet hint

  • 733b502 chore(thread): #14406 review nits: the roster read once per restore, sessionOn's doc in place, the silent-service restore leg tested (#14305 U7b)

  • f41c20d refactor(thread): the last deviceOnline callers check the device through resolveTypedDevice; deviceOnline and rowThreadId go (#14305 U7b)

  • 4cfc217 Merge PR#14374 @8f2c5a60f5 into thread train 8

  • 59e8630 fix(thread): answers about a session with no durable id carry its row key (#14305 U7b, review M2 of #14386)

  • 732cc49 Merge PR#14406 @3e2f5b4dbb into thread train 8

  • 3cfc5f8 test(thread): RED - answers about a session with no durable id name it by its row key (#14305 U7b, review M2 of #14386)

  • 2818426 Merge pull request #14408 from code-yeongyu/test/hero-stable-all-locales

  • 5d064bd Merge pull request #14401 from code-yeongyu/feat/download-to-stable

  • fe8061b test(remote): give the EACCES endpoint case its secret file on Windows, where the endpoint resolves through it (#14384)

  • 57931d7 chore(changes): number the fleet hint fragment

  • 2393d14 refactor(skill-pointers): the fleet hint points at the omo-relay skill

  • 3e2f5b4 fix(thread): a restored / monitor watches that session on this device (#14305 U7, review L1 of #14406)

  • 66ba0b9 test(thread): RED - a restored / monitor watches that session (#14305 U7, review L1 of #14406)

  • 5af3c05 fix(thread): a restored device-wide monitor on an offline or silent device keeps its watch (#14305 U7, review M1 of #14406)

  • 62d0856 test(skill-pointers): device and fleet wording points at the omo-relay skill

  • 682f90e test(thread): RED - a restored device-wide monitor on an offline device keeps following its listing (#14305 U7, review M1 of #14406)

  • 6636278 Merge remote-tracking branch 'origin/dev' into feat/14305-u7-remove-old-resolvers

  • a634982 docs(changes): fold the #14379, #14392, #14397, #14402 and #14409 fragments into the train 8 fragment (#14415)

  • 865dd43 Merge PR#14392 @c492d8666c into thread train 8

  • 5372c34 Merge PR#14397 @758c9c0357 into thread train 8

  • 523725f Merge PR#14392 @f4243bd92e into thread train 8

  • b77fd5a Merge PR#14402 @9e0558d909 into thread train 8

  • fa2e3b9 Merge remote-tracking branch 'origin/dev' into feat/14364-omo-sdk-entry

  • d5e79a7 Merge PR#14409 @a7e8ca9f99 into thread train 8

  • 42e6e4e Merge PR#14379 @d6d95ec830 into thread train 8

  • 9dd731a Merge origin/dev (thread train 7) into fix/14389-ultrawork-arming-off-switch

  • c46257d Merge pull request #14404 from code-yeongyu/release-train/omo-20261119-thread-train-7

  • df9c896 Merge PR#14387 @ba86a52f7b into thread train 7

  • 9f4413a fix(thread): the restore fallback watches on the resolved device id and sends only an id unlisted (#14305 U7)

  • 06d32d8 test(thread): RED - the restore fallback resolves the device part and never sends a name unlisted (#14305 U7)

  • bec5297 Merge remote-tracking branch 'origin/dev' into feat/14364-omo-sdk-entry

  • c46d833 test(eq-omo): re-record darwin-arm64 E21-E24 goldens for train 7, sandbox outside the home tree (#14404)

  • 0be4edf test(eq-omo): re-record linux-x64 E21-E24 goldens for train 7 in a clean environment (#14404)

  • 9cb03cf docs(changes): fold the #14378, #14381 and #14393 fragments into the train 7 fragment (#14404)

  • ac44a5a Merge PR#14393 @d68599991a into thread train 7

  • 6d1b646 Merge PR#14386 @9be350a3fd into thread train 7

  • 758c9c0 fix(thread): an unreached remote create removes the worktree it made or names it, every create is checked for reach, and a route_unlisted is final only after b is asked again (#14397 review M1 L1 L2 L3)

  • 51b9dac test(thread): an unreached remote create removes its worktree, an alias-only create is checked, a missing routing id is said, and one late route_unlisted ends nothing (red, #14397 review M1 L1 L2 L3)

  • 5ae86dc Merge remote-tracking branch 'origin/dev' into fix/14394-remote-create-unreachable

  • 412741a Merge PR#14381 @8d95580240 into thread train 7

  • a53b45c Merge origin/dev (a9f9939) into fix/14389-ultrawork-arming-off-switch

  • 9e0558d docs(changes): fragment for #14402

  • aff3383 fix(thread): a close's settlement drops rows it lost the race for; a nothing_queued wake reports drained marks (#14383)

  • d6d95ec Merge remote-tracking branch 'origin/dev' into fix/14351-settled-hooks-bounded

  • 4f8b0d5 Merge PR#14378 @c6acf8bca8 into thread train 7

  • a7e8ca9 chore(changes): fragment for #14409

  • 67a4f3a feat(thread): thread_list rows on this device name the engine version of their host (#14278)

  • bc7dd9e test(web): the flag-off hero test covers all 9 locales and each locale's own label

  • a9f9939 Merge pull request #14407 from code-yeongyu/test/eq-darwin-goldens-host-independent

  • 54c9853 test(eq-omo): restore darwin-arm64 E21-E24 goldens to the last host-independent recording

  • 680e976 chore(thread): lint fix and stale comments after the full gate pass (#14305 U7)

  • eea1c84 Merge pull request #14356 from code-yeongyu/release-train/omo-20261119-thread-train-6

  • 2cc4d36 Merge pull request #14395 from code-yeongyu/feat/14388-ulw-execute-light-profile

  • 5353ee1 Merge pull request #14400 from code-yeongyu/fix/12022-test-macos-realpath

  • 5faf6af fix(web): the home page's desktop button opens the stable download page, worded as the desktop app download

  • 2969af4 refactor(thread): relayAddressEntry reads row.durable_id; rowThreadId is left to relay-transport.ts (#14305 U7)

  • 5ed6ec0 refactor(thread): relay-session-host's open checks its device through resolveTypedDevice (#14305 U7)

  • d30779b fix(remote): sessions.resolve ambiguity candidates carry unnamed (#14305 U7)

  • f222d3e test(remote): RED - sessions.resolve candidates say when a name is a fallback (#14305 U7)

  • 8d25d05 refactor(thread): every caller imports the account reads from thread/address; remote/account.ts and the holderTarget re-export go (#14305 U7)

  • b86820c refactor(thread): delete relayAddressBook and its last-listed memory; thread_list feeds the module's one (#14305 U7)

  • ef0d4af docs(changes): fragment for #14401

  • 632ce3b fix(thread): a restored monitor on a device that is offline or silent watches the stored key (#14305 U7)

  • 2e3730d test(thread): RED - a restored monitor on an offline device keeps watching (#14305 U7)

  • ebed66d fix(thread): rowName names this device's listed threads and every candidate; a keyless row is passed back by rowKey (#14305 U7, review M3 of #14386)

  • c9a6803 test(thread): RED - unnamed threads named by rowName on this device's leg and in candidates (#14305 U7, review M3 of #14386)

  • 0be2df8 fix(web): send /download to the stable desktop download page; ?channel=beta keeps the beta page

  • c38f309 refactor(thread): U7's files read row.durable_id instead of rowThreadId (#14305 U7)

  • 4c008f6 refactor(thread): delete resolveAcross and its helpers; remote-ops device helpers go private (#14305 U7)

  • 896d0d1 refactor(thread): monitor targets resolve through the address module (#14305 U7)

  • ce6c632 test(thread): RED - monitor targets resolve their thread part and device name through the module (#14305 U7)

  • bbb12c3 docs(thread): name resolveAddress where AGENTS.md still named resolveAcross / resolveStoredSession (#14305 U7)

  • a920fbf refactor(thread): thread_create's offline check goes through the module's device resolution (#14305 U7)

  • ae48587 test(thread): RED - thread_create reopening a thread on an offline device names the device (#14305 U7)

  • 4eb0f5d refactor(thread): drop resolveEntries; its two callers build the book they hand the module (#14305 U7)

  • 5c910c9 refactor(thread): move holderTarget into thread/address (#14305 U7)

  • ab92611 test(eq-omo): re-record linux-x64 E21-E24 goldens in a clean environment (#14356)

  • b8b4e34 test(update): compare the versioned-install fixture by its real path so macOS /var -> /private/var matches (#12022)

  • 3cf29fa Merge remote-tracking branch 'origin/dev' into feat/14364-omo-sdk-entry

  • 9be350a fix(thread,remote): review M1, L1, L4, L5 of #14386 (#14305 U6)

  • 8f2ebe0 test(thread): RED for review M1 and L1 of #14386 (#14305 U6)

  • e59b624 fix(thread): judge a remote create's reach by the service's route_unlisted only, from the name or a first prompt's state read, so no create costs a round trip it did not (#14394 gate)

  • f7942f4 fix(ci): regenerate the Codex installer and bundle inputs for #14392, drop the unused KeywordTypeSchema export

  • 6e811f1 docs(changes): fragment for #14397

  • d685999 docs(changes): fragment for #14393

  • 61068b5 fix(thread): a remote create ends a session the device's route does not reach and refuses host_unavailable, never ok with an orphan thread (#14394)

  • ca6a781 test(thread): a remote create whose new session the device's route does not reach is refused, with nothing left (red, #14394)

  • 1a2d7ef test(thread): the reopen frame on b names the folder find_session reports (#14296)

  • f1991bd docs(changes): fragment for #14395

  • 2a0a93e chore(bundles): set omo.js budget to the #14392 measurement + 4 KiB

  • 3c1d0f4 Merge remote-tracking branch 'origin/dev' into fix/14389-ultrawork-arming-off-switch

  • e271f83 feat(ulw-execute): --light keeps the plan, uses one worktree and verifies once at the end (#14388)

  • abce676 fix(thread): a reopen names the session's recorded folder, so the engine never probes the file's path on a loaded host (#14296)

  • c33f7e5 test(thread): a reopen names the folder its session file records (red, #14296)

  • e9f6a22 docs(changes): fragment for #14392

  • cdd244e Merge dev @bf159fb831 (thread train 5 #14332) into thread train 6

  • bf159fb Merge pull request #14332 from code-yeongyu/release-train/omo-20261119-thread-train-5

  • ba86a52 test(thread): a send to a thread live under another engine's key on its device reaches that holder at once (refs #14298)

  • 8d95580 fix(omo-relay): the cell imports the fleet SDK; name thread tools in their eval form

  • cd24706 test(eq-omo): re-record linux-x64 E23 on an idle host: the deterministic 4-step memory status (#14287)

  • 13b4dd3 Merge train5 @c16067f14c into thread train 6

  • ca5902c test(eq-omo): E21-E24 goldens carry the corrected task model descriptions; lease QA names a custom agent

  • 9669b45 Merge PR#14363 @b41242db44 into thread train 6

  • b41242d fix(thread): a create answers a prompt stored for a woken session as queued waiting on the wake, and the wake floor is at least one wake's measured cost (#14363 delta review L1 L2)

  • abd077f test(thread): a prompt stored for a woken session is accepted, and the wake floor never drops below one wake's cost (red, #14363 delta review L1 L2)

  • b9e068e chore(changes): note the omo-relay skill

  • 61ad2d6 docs(changes): fragment for #14379

  • e251838 Merge pull request #14373 from code-yeongyu/docs/14372-features-defaults

  • 2090ccd fix(memory): a settle skips the state write only when state.json already holds it (#14351)

  • 742d62b refactor(thread): delete accountListing; relayAddressBook reads the round itself (#14305 U6)

  • b338a7c chore(thread,remote): format the list unit's files; accountListing is module-private now (#14305 U6)

  • 4960ced fix(thread): thread_list on the one row identity, merge, naming and device state (#14305 U6)

  • dfaf319 fix(remote): sessions.list on the one row identity, merge, naming and device state (#14305 U6)

  • e45c73b test(thread,remote): RED for the list unit's row identity, names and --device (#14305 U6)

  • 8fb9087 test(native): the failing-busctl case writes its own stub, as /bin/false is /usr/bin/false on macOS (#14348)

  • bb6bf97 test(omo-relay): helpers answer one outcome per target and write nothing

  • d788e86 chore(changes): fragment for #14378

  • 0f3f7a3 chore(knip): list the graceful-close QA scenario run-all.mjs runs by path (#14324)

  • 581d408 Merge PR#14363 @0fcdd7cd47 into thread train 6

  • 0fcdd7c fix(thread): a create answers a first prompt the thread admitted to its queue as queued, and the wake memory floor is a quarter of a limit below 4 GiB (#14363 review M1 L1)

  • f1ec5e7 test(thread): a first prompt admitted to a queue is accepted, and a small memory limit still wakes (red, #14363 review M1 L1)

  • e304202 Merge pull request #14360 from code-yeongyu/docs/14357-signin-categories

  • 75bd1b3 docs(changes): fold the #14339, #13994 and #13895 fragments into the train 6 fragment (#14356)

  • 2b2114a docs(changes): keep dev's #14337 and #14346 fragments; the train 6 fold moved them into 14356.md

  • be47d56 Merge PR#13940 @8c2bf2e1cc into thread train 6

  • 72468a7 Merge PR#13895 @14323a6225 into thread train 6

  • 0f32723 style(thread): format the create-live-prompt files (fmt gate)

  • 640d1bd style(thread): format the thread train 6 merge resolutions (#14355, #14339)

  • bef42ff docs(cli): omo login --api needs --device or --no-browser while accounts are closed (#14357)

  • fb26510 chore(bundles): set omo.js budget to the thread train 6 top measurement + 4 KiB

  • 929de41 fix(thread): a create that opens a thread live already sends its prompt to it, answering the delivery or prompt_not_started (#14335)

  • 78c5d75 Merge PR#14355 @f433feb879 into thread train 6

  • d5d5bcf fix(memory): settle work coalesces by supersession and journals reconcile incrementally (#14351)

  • 03ab3ad Merge PR#14339 @5825d630a4 into thread train 6

  • d88fdf8 Merge PR#13994 @c5866d6330 into thread train 6

  • 14c8310 Merge PR#14353 @c8dab2918d into thread train 6

  • 4728f77 Merge PR#14299 @377fe50438 into thread train 6

  • 3291c8c Merge PR#14324 @64ac9319f5 into thread train 6

  • 05d6555 Merge PR#14348 @5e3c44437e into thread train 6

  • c618fc0 Merge train5 @80b769d056 into thread train 6

  • f7ba8ec Revert the #14321 merge out of thread train 6

  • 5add9f4 Merge remote-tracking branch 'origin/dev' into docs/14357-signin-categories

  • 80b769d Merge dev 10714ddf (#14361 dev-red fix) into thread train 5

  • 10714dd Merge pull request #14361 from code-yeongyu/fix/dev-red-14345-14337

  • 4db631f fix(ci): format doctor.js and regenerate the stale omo-task bundle on dev (#14345, #14337)

  • 54223d5 docs: document terminal sign-in, split category chains by harness, fix shared-skill count (fixes #14357)

  • 29624a6 docs(changes): fold member fragments into the train 6 fragment (#14356)

  • 10fea9c docs(changes): name the thread train 6 fragment for its PR (#14356)

  • c8dab29 Merge branch 'feat/14313-graceful-close' (with release-train/omo-20261119-thread-train-6) into fix/14352-wait-capped-limit

  • 377fe50 Merge remote-tracking branch 'origin/release-train/omo-20261119-thread-train-6' into tsf-close/14299

  • 64ac931 Merge remote-tracking branch 'origin/release-train/omo-20261119-thread-train-6' into tsf-close/14324

  • f433feb test(thread): the parked-thread wake test passes its send mode as delivery, the field thread_send takes (#14240 caller of an unknown field)

  • 3a2b2cc Merge remote-tracking branch 'origin/dev' into tsf-close/14299

  • 5825d63 fix(thread): the relay leg reports the lane the receiver answered, and a follow_up no longer needs the state probe (#14333)

  • 5c80bd4 test(thread): the relay result names the lane the receiver's answer proves; a follow_up survives a lost probe (red)

  • 6dc0625 Merge PR #14299 @030e18d0a8 into thread train 6

  • 4202694 Merge PR #13077 @5d04e8a711 into thread train 6

  • 5e3c444 test(omo-native): make the hung-busctl stub record its real pid; document the bus pre-check's limits (#14341)

  • 50df41e Merge pull request #14310 from code-yeongyu/test/14270-desktop-host-leg

  • 005797e Merge remote-tracking branch 'origin/dev' into fix/held-session-wake

  • 9b6ee2e fix(thread): thread tools refuse a field they do not take, naming the field meant; a create states prompt null when given none

  • 1aef562 test(thread): unknown or misnamed tool fields are refused, and a create states prompt null when none was given (red)

  • 7c040a3 fix(thread): a create whose first prompt does not start a turn fails prompt_not_started, never ok and silent

  • d35f7bc test(thread): a create whose first prompt never starts fails prompt_not_started instead of answering ok (red)

  • 07bec83 fix(thread): a stored unattended-only hold blocks no later wake (carrier and waiting_on read one predicate)

  • b7a993b test(thread): a session the carrier woke and handed back wakes again for a later delivery, held unattended or not (red)

  • f70d961 test(thread): the pair loop guard tests pin its numbers, not its wording

  • d54aa61 test(thread): a delivery to a session a host restart parked resumes under a dev engine stamp

  • 01fa0b0 fix(thread): no fixed woken-session cap; the carrier defers a wake only on measured host memory (host_memory_pressure)

  • a7ad9d1 test(thread): parked sessions wake without a count cap, deferred only on measured host memory (red)

  • 3707d9d fix(thread): the pair bucket answers a truthful pair_loop_guard, not overloaded, on both legs

  • 6912c97 fix(thread): the queued answer's waiting_on keeps its kind's type (tsgo)

  • 308a8cf test(thread): a 9th burst send to one session is refused pair_loop_guard with the measured burst (red)

  • 48dd684 fix(thread): no per-turn cap on how many sessions one turn reaches

  • f015ad1 fix(thread): an engine 'held' admission answers queued waiting on the hold, on the local and relay legs

  • 8dfb297 fix(thread): the wake's engine gate reads a dev engine's CalVer prefix

  • 2e5a060 docs(changes): fold the #14321 fragment into TRAIN6.md

  • 5d04e8a Merge remote-tracking branch 'origin/dev' into fix/13074-kibitzer-hook-never-blocks

  • 92b134d Merge PR #14321 @ac06c390d1 into thread train 6

  • da2cdc9 Merge thread train 5 @182e24cd7d into thread train 6

  • 1053030 refactor(thread): resolution is module-private once resolveStoredSession is gone (#14322 with #14328)

  • 9d0a0bd chore(knip): list the default-delivery QA scenario run-all.mjs runs by path (#14334)

  • d84a3ef Merge pull request #14337 from code-yeongyu/fix/doctor-engine-commands-14336

  • 9363c60 fix(omo-native): gate the host scope on the user bus busctl uses; drop unused export; cover the timeout path (#14341)

  • 071e05f test(thread): one reopen table proves the holder kind reaches the route (#14270 review)

  • 182e24c test(audit): categorize the /bin/omo literals in self-update-versioned-install.test.ts (member #14256)

  • c315cf2 fix(thread): a graceful close names the leg's cap only when the cap bounded its drain (#14352)

  • ac068db docs(changes): add the user-facing lines for #14282 and #14276 that the thread train 5 fold left out

  • db4f43c fix(omo-native): move the host supervisor into its own scope via StartTransientUnit, never failing the launch (#14341)

  • adacbff Merge remote-tracking branch 'origin/dev' into test/14270-desktop-host-leg

  • 1416c63 docs(changes): fold the thread train 6 member fragments into TRAIN6.md

  • 8b4d72a Merge PR #13975 @b40c5121be into thread train 6

  • b7ee168 Merge PR #14334 @fc22f87e44 into thread train 6

  • b6dbd66 Merge PR #14309 @fe2c5179cf into thread train 6

  • ef0a719 Merge PR #14304 @93d376f3c4 into thread train 6

  • fabf1ea Merge PR #14283 @395b2894ce into thread train 6

  • 9e0c8ef Merge PR #14331 @ee9907a57d into thread train 6

  • 5d00a27 test(native): drop the reachability pin of the sendAddressBook call in resolveStoredSession, deleted by #14322 (thread train 6 merge of #14328)

  • 9a6fd5b Merge pull request #14346 from code-yeongyu/fix/14344-installer-real-root

  • b8656bb Merge PR #14328 @208dbac564 into thread train 6

  • 43f6e19 Merge PR #14327 @e632537b62 into thread train 6

  • c576af8 Merge PR #14322 @8f37ff3fed into thread train 6

  • 1baf721 test(get-worker): format root-guard test

  • 8c2bf2e Merge rpc-relay-parity-measured (#13895 at 14323a6) into the branch

  • 14323a6 fix(knip): the live parity driver is an exports-unchecked entry (run by path from qa:rpc-relay-parity)

  • 0dcc384 Merge PR #14318 @0cabf948e2 into thread train 6

  • 0786a1e fix(omo-native): start Linux engine hosts in their own systemd user scope (#14341)

  • a937c0d docs(changes): fragment for #14346

  • edd8d07 fix(get-worker): install for a real root shell; refuse only sudo/doas escalation (#14344)

  • 8861e99 Merge remote-tracking branch 'origin/dev' into fix/doctor-engine-commands-14336

  • 96095ef fix(omo-native): doctor keeps only omo's own commands; engine commands come from the shared list (#14337 review LOW-1, LOW-2)

  • e1bd464 Merge dev d03b389 (train 4 #14284 + #14255) into thread train 5

  • d03b389 Merge pull request #14284 from code-yeongyu/release-train/omo-20261119-thread-train-4

  • e4b101f Merge rpc-relay-parity-measured (#13895 at 4a2d72a, dev merged) into the branch

  • 4a2d72a Merge origin/dev (c9c4cfb) into the branch; parity table regenerated with bun script/rpc-relay-parity.ts

  • a2ef67b test(omo-native): drop the copied kill-guard dead code and keep a Windows run's runtime in the fixture home (#14255 review LOW-3, LOW-4)

  • 19a52ac qa(thread): default-delivery waits for the steered turn to settle before step 3 (#14333)

  • c5866d6 Merge origin/dev (c9c4cfb) into the branch; generated bundles carry dev's copy

  • b40c512 Merge origin/dev (c9c4cfb) into the branch; generated bundles carry dev's copy

  • f932a8c Merge pull request #14255 from code-yeongyu/fix/session-of-pid-dispatch-14253

  • af42d01 docs(changes): fragment for #14339

  • d82e7a2 fix(thread): an idle start reports effective=start, not follow_up (#14333)

  • f6a7c63 docs(changes): fragment for #14337

  • a612454 fix(omo-native): doctor reads one-shot engine commands from the shared list; session-of-pid test header (#14336)

  • ce613b1 test(thread): one turn reaches 40 sessions, all delivered (red: the per-turn cap of 16)

  • ee9907a fix(native): a typed listing key is set at the thread, link and details.thread fields, not by rewriting serialized output (#14305 delta review 2 M2 follow-up)

  • 550ccd5 test(native): a typed listing key names the session at the thread and link fields, never inside a reply's text (RED, #14305 delta review 2 M2 follow-up)

  • 6272d2b test(thread): a send to a target its engine holds answers the hold, local and relay legs (red)

  • fc22f87 fix(thread): review r1: keep omo-native's contracts import type-only, assert the literal default, cover create's first prompt (#14333)

  • 9d4469b test(thread): the wake's engine gate reads a dev engine's CalVer prefix (red)

  • 2e8c52e docs(changes): omo remote attach checks the relay before attaching (#14305 delta review 2, L1)

  • ba0f9e8 refactor(native): one name for the relay check: remoteAvailable, re-exported for attach (#14305 delta review 2, L2)

  • 65468c6 fix(native): a listing key the person typed names the session in every --json answer (#14305 delta review 2, M2)

  • 0ad0f36 fix(remote): the this-device holder check answers from the reads the resolution already made (#14305 delta review 2, M1)

  • 4391a9c test(native): second delta review findings (RED, #14305 delta review 2 of #14331)

  • 83a998a qa(thread): live default-delivery scenario: no mode starts an idle TUI and steers a busy one; follow_up waits (#14333)

  • 9a1a7ff Revert "chore(bundle): regenerate this PR's own bundle delta (#14333)"

  • 31f091d chore(train): Set A regen at the thread train 5 top after dev c9c4cfb (#14303)

  • 7ac6a4a docs(changes): fragment for #14334

  • 1a60e7f chore(bundle): regenerate this PR's own bundle delta (#14333)

  • 484cafd fix(thread): one shared default delivery, auto, on every send surface (#14333)

  • 87a05d6 Merge dev c9c4cfb into thread train 5 (carries #14303)

  • c5d909f docs(changes): fold the thread train 5 member fragments into 14332.md

  • 1e41f87 chore(train): Set A regen at the thread train 5 top

  • 6c46be6 fix(native): every omo remote verb checks the relay first again, names a typed key in the answer, and maps thread_list advice in --json (#14305 delta review H1, LOW2, LOW3)

  • 6d23a6f fix(remote): resolveTypedAddress refuses a durable id this device lists too, and answers a thread held here from its first resolution (#14305 delta review H2, LOW1)

  • eb08a12 test(native): delta review findings for the omo remote address CLI (RED, #14305 delta review of #14331)

  • 569e203 Merge PR #14287 @ed9afe1f43 into thread train 5

  • 32e00c3 Merge PR #14326 @cc4edefbe2 into thread train 5

  • a581a9c Merge PR #14256 @b0475d7e49 into thread train 5

  • cd3c15b Merge PR #14307 @bd54812e5c into thread train 5

  • 77f91d0 Merge PR #14228 @eaf6231620 into thread train 5

  • dc9ddae Merge PR #14276 @fc57619930 into thread train 5

  • f50f527 Merge PR #14282 @530d5007bd into thread train 5

  • 9f49b2f Merge PR #14292 @6a73675a8d into thread train 5

  • 5e9b21b Merge PR #14285 @1274b936b7 into thread train 5

  • 7439c89 test(native): the five session-verb sequences read one filtered listing and no account-wide one (#14305)

  • d1c382c docs(remote): the Addresses section states when a device is asked itself, what here answers, and how candidates print (#14305 review M3, L1, M4)

  • d857cdd fix(native): /session and device/ are usage again, and attach checks its target without throw-then-catch (#14305 review L2, L4, M2, M4, L1)

  • 1caeb6b fix(native): the session verbs read the relay state once, keep the typed listing key as thread, and print candidates (#14305 review L3, L2, M2, M4, L1)

  • 0aeb7c4 fix(remote): resolveTypedAddress refuses a durable id two devices list and answers a thread held here as the module's not_found (#14305 review M2, L2)

  • bbbbc0b fix(native): revoke lists an ambiguity's candidates one per line and carries their ids in --json (#14305 review M4, L8)

  • af5fb6d fix(native): sessions --device prints an ambiguous device's typed failure in --json, with its candidates one per line (#14305 review M1, M4, L7)

  • 4a7f32a test(native): review findings for the omo remote address CLI (RED, #14305 review of #14331)

  • d02db7a docs(changes): fragment for #14331

  • 05c8bfe test(native): the session verbs' call sequences record the listing filter and pin the address module's reads (#14305)

  • 1600733 docs(remote): one Addresses section for the omo remote grammar (#14305)

  • 1cb6bae feat(native): omo account devices revoke names a device by id, name or prefix and refuses an ambiguity (#14305)

  • c64f2d1 feat(native): omo remote verbs resolve one typed address through the address module; remote-session-target goes (#14305)

  • 5587230 feat(remote): seam 4, resolveTypedAddress, the one resolver for a typed thread address (#14305)

  • 0794807 test(native): the sessions --device name cases move to the list unit (#14305)

  • 4e25357 test(native): the remote CLI and device revoke take every address form (RED, #14305)

  • 208dbac test(native): the reachability pin lists sendBookFor's call of sendAddressBook, the one site L4 moved there (#14305 review L4 of #14328)

  • ac06c39 fix(build): fold the engine version define into compileDefinesForOmoBinary (#14321 review)

  • 4e8af35 docs(changes): #14328 fragment names the candidate device id and the offline-device refusal

  • e632537 test(remote): attach runs its open under what is left of the caller's one deadline; the fragment names the shared deadline (#14327 delta review L-d)

  • f877abc test(thread): route parity asserts the listing_incomplete details, not message wording (#14305 review L7 of #14328)

  • 91c91c9 test(thread): an unattested reply written / names its device, so unique_holder does not apply (#14305 review L6 of #14328)

  • 50c5995 fix(thread): thread_create's existing-thread lookup asks this device's book per address part, so here/ finds a session file (#14305 review L3 of #14328)

  • 8ba7dee fix(remote): attach judges a name miss as the session members do, listing_incomplete while the device missed the round (#14327 delta review L-f)

  • 64741cd test(thread): thread_create opens here/ of a thread known only by its session file, with no account (RED, #14305 review L3 of #14328)

  • 8fba59d test(remote): attach by a name on an online device that missed the round is listing_incomplete (RED, #14327 delta review L-f)

  • 0557590 refactor(thread): one sendBookFor beside sendAddressBook hands the resolver this device's book per address part; tools, create and locate drop their copies (#14305 review L4 of #14328)

  • 7351e98 fix(remote): the roster-miss path reuses the listing it read, so it reads the listing once (#14327 delta review L-a)

  • 45c25a1 test(remote): a listing-named device's address reads the listing once (RED, #14327 delta review L-a)

  • ae55079 fix(thread): a graceful close of another device's thread drains at most 35 s; wait_ms needs graceful (#14324 review)

  • 8f37ff3 docs(thread): wrap the address paragraph of omo-thread.md to the file's width (#14322 delta review N2)

  • 43620d8 test(thread): one SDK object wakes a session that started after its earlier send (#14322 delta review N1)

  • caee0be fix(remote): a listing-named device reads offline in every answer, as sessions.list shows it (#14327 delta review L-b)

  • 06c3d7c test(remote): a listing-named device's surfaced row is offline as sessions.list shows it, while the core resolves it online (RED, #14327 delta review L-b)

  • 13c6919 fix(thread): address module answers device_offline for what an offline device's rows in hand do not resolve, and never probes it; core exports deviceOffline (#14305 review M2 of #14328)

  • b3079f3 refactor(remote): the resolved device's row comes from the roster or the listing-named device, no unreachable fallback (#14327 delta review L-c)

  • 3409b9e Revert "test(thread): a no-remote SDK send to an id the store does not own lists the host once (RED, #14322 review L5)"

  • ea62ff8 Revert "perf(thread): a no-remote SDK send lists the host once: the gateway reads the view the send resolved over (#14322 review L5)"

  • 98b6a02 chore(remote): drop the doc comment Stage left behind in sessions.ts (#14327 delta review L-e)

  • 003cbb7 test(thread): / resolves only from rows in hand, else device_offline with nothing sent (RED, #14305 review M2 of #14328)

  • 105b8b9 fix(thread): a device-qualified thread id hands the listing and roster already read into the pinned ask; it re-reads only for the #14280 second pass (#14305 review M1 of #14328)

  • 7e03e4f test(thread): / reads that device's listing once (plus the #14280 re-read on a miss) and the roster once (RED, #14305 review M1 of #14328)

  • 2365feb fix(thread): address module checks a device-qualified hit against this device's book only, and reads the account-wide round only when the device misses a non-id part (#14305 review H1 of #14328)

  • e675de1 test(thread): a device-qualified address that its device answers reads no account-wide listing; a miss reads it once only for a non-id part (RED, #14305 review H1 of #14328)

  • e729765 test(thread): graceful-close QA reports a close that answers before it fences (#14313)

  • f4c416d docs(changes): #14327 fragment names attach under the id-shape rule

  • 7ed76e0 fix(remote): sessions.resolve, the session members and attach spend one deadline across the roster, listing and member (#14327 review L1)

  • 10bfaa7 fix(native): restore the comma the usage edit dropped (#14322 review L1)

  • 1927b51 docs(changes): #14322 fragment names the scope_denied change and the local-first binding operations

  • 7a302a8 docs(thread): omo thread's reference matches the code: send to another device, host_unavailable for the binding operations, exit 1 for a gateway unsupported, no self (#14322 review M2, L1)

  • ae70c69 fix(native): omo thread's usage no longer offers self, which never resolves for a cli: caller (#14322 review L1)

  • 7b14152 test(thread): the session_on_other_device test asserts details, not message wording (#14322 review L6)

  • 372ab20 style(thread): drop the blank line the resolveStoredSession removal left in internals.ts (#14322 review L3)

  • 71fd48e chore(thread): drop the SDK send's redundant id_prefix: true, the resolver's default (#14322 review L4)

  • 0a1fb10 perf(thread): a no-remote SDK send lists the host once: the gateway reads the view the send resolved over (#14322 review L5)

  • dd8f950 test(thread): live graceful-close QA on a busy daemon session and a busy terminal (#14313)

  • 9319105 test(thread): a no-remote SDK send to an id the store does not own lists the host once (RED, #14322 review L5)

  • f3714f1 test(thread): an SDK send to a name only another workspace holds is scope_denied unless all_scope (#14322 review M1)

  • 1fd2621 revert(thread): the SDK send's book reads every workspace again, as thread_send and locate do; the resolver judges scope (#14322 review M1)

  • bff7f45 fix(thread): store-only SDK operations resolve on this device first and consult the account only on a local not_found (#14322 review H1)

  • b5ed975 test(thread): a name a local session and another device's thread share binds and reports locally (RED, #14322 review H1)

  • 702ec19 docs(changes): fragment for #14328

  • 3a80e4d test(remote): a device-qualified resolve runs the listing under what is left of the caller's one deadline (RED, #14327 review L1)

  • 01a0681 docs(remote): the id-shape predicate's comment names the core idShaped rule, not on-device.ts (#14327 review L3)

  • ffa2f7c fix(remote): attach sends a session part to its device only when it is id-shaped or a listed key (#14327 review L7)

  • 7ccf87d test(remote): attach refuses a session part that is no id shape and no listed key, with the device never attached (RED, #14327 review L7)

  • 58d07f2 fix(remote): an alias's listing_incomplete names the address asked for, not the bare device id (#14327 review L5)

  • 03a992b refactor(remote): the resolved device always carries its row; the member branch that refused a missing row goes (#14327 review L4)

  • 961ce32 fix(remote): a listing-named device the roster misses resolves by its exact id again (#14327 review M2)

  • fab4246 test(remote): a device the roster misses but the listing names resolves by its exact id (RED, #14327 review M2)

  • 08ff97d fix(remote): a device miss keeps details.id, the core's device and candidates added beside it (#14327 review M3)

  • d342d1a test(remote): a device miss keeps details.id beside device and candidates (RED, #14327 review M3)

  • 59e5a1b fix(remote): a device-qualified address that two rows answer is ambiguous_target again (#14327 review M1)

  • e905dff test(remote): two rows answering a device-qualified address are ambiguous_target (RED, #14327 review M1)

  • cc4edef fix(thread): a held steer whose re-send is refused turn_conflict settles its receipt (#14320 review)

  • 148818e fix(thread): address module resolves a device part from the roster and an id-shaped thread part from that one device, on one status round per resolution (#12709, #14305)

  • 45c304d fix(thread): address module resolves a listing row that publishes no thread id by its listing key, on its device and bare (#14305)

  • e47698a test(thread): a device-qualified id reads only that device; a row with no thread id resolves by its listing key (RED, #12709, #14305)

  • 8b415ce refactor(thread): delete replyOnDevice; its last caller now resolves through resolveAddress's pin, whose table (resolve-refusals.test.ts and reply-device.test.ts) holds every case it had; no test called it directly (#14305)

  • 36e7cdc style(thread): bun fmt (#14305)

  • 987052c feat(thread): locate, send and create hand the resolver this device's book per address part instead of stripping here/ (#14305 review M2)

  • ec0eea8 fix(thread): address module asks this device's book for the thread part once the device part names this device (ResolveContext.localFor, #14305 review M2)

  • cb85246 test(thread): / asks this device's book for the thread part (RED, #14305 review M2)

  • 1a14df0 test(thread): the two-device fixture gives this device a hex id of its own, so here and the id are tested apart (#14305 review M2)

  • c6ef0a7 fix(thread): address module asks the one device a / names when its listing does not show the id, as a pinned reply does (#14305 review H1)

  • d354dbe test(thread): / on another device asks that device; table through resolveAddress and thread_read (RED, #14305 review H1)

  • b8ec49c feat(thread): address module passes an unlisted id-shaped session part through to its device; core exports idShaped (#13748, #14305)

  • 97aa60e test(thread): a miss while b missed the round is host_unavailable naming b, not not_found (#14305, reverses #14172's choice)

  • 89bd5bc test(thread): count every roster read of a send, through the remote client too; a reply now reads the account once (#14305)

  • 8d2416b feat(thread): an unattested reply checks its sender's one holder on the resolution's own account read (#14048, #14305)

  • dddf25b feat(thread): address module refuses a thread two devices list when the caller asks for a unique holder (#14048, #14305)

  • 1459a40 feat(thread): address module names why a pinned reply could not tell (#14280 vocabulary) and who answered a probe (#14305)

  • 512f12b base only: PR #14285 (#14280 reply reasons) merged locally; dropped by the rebase onto dev

  • 4996f91 feat(thread): thread_send and thread_handoff resolve their target through resolveAddress (#14305)

  • 4d533da feat(thread): thread_create resolves existing threads and fork sources through resolveAddress (#14305)

  • da51d9c test(thread): thread_send, thread_handoff and thread_create take / (RED, #14305)

  • c87fcbc base only: release train 4 (#14284) merged locally so tools.ts carries #14263; dropped by the rebase onto dev

  • 7b5ea02 fix(thread): an SDK send reads the session files of the caller's scope, as the agent's thread_send does (#14322)

  • e1726af docs(changes): fragment for #14327

  • 125cf34 fix(remote): an alias holder's online device that missed the round answers listing_incomplete (#14305)

  • c1c4ef4 test(thread): a held steer whose re-send is refused turn_conflict settles its receipt (RED, #14320 review)

  • 0b63d25 test(remote): an alias holder on an online device that missed the round is listing_incomplete, not device_offline (RED, #14305)

  • 7f5431d fix(remote): the device refusal fallback names no typed text, so sessions.resolve typechecks (#14305)

  • 68eff9f fix(remote): attach resolves its device part through the core's device resolver (#14305)

  • c9aa59f test(remote): attach resolves its device part by name in any case, not_found{device} otherwise (RED, #14305)

  • 72997a9 refactor(remote): aliases parse a handle once through the shared parser; sessionOf goes (#14305)

  • 329937e fix(remote): unlisted pass-through only for id-shaped session parts; listing_incomplete from the core's deviceListing (#14305)

  • e740053 test(remote): only an id-shaped session part passes unlisted to its device; an incomplete device's miss is listing_incomplete (RED, #14305)

  • c5e9006 refactor(remote): session members resolve handles through the address module, typed refusal for names (#14305)

  • af40602 refactor(remote): sessions.resolve parses and resolves devices through the address module (#14305)

  • 112dd1a test(remote): an unresolvable session part is a typed refusal, never device pass-through (RED, #14305)

  • f2450b6 test(remote): device prefix, NFKC name, here part and listing_incomplete misses (RED, #14305)

  • 0bb0d1a fix(thread): an SDK send resolves through the address module with or without a remote surface (#14322)

  • 3aaef55 test(thread): an SDK send with no remote surface resolves here/ and / through the address module (RED, #14322)

  • 81581c1 docs(changes): fragment for #14326

  • cd352e4 fix(thread): a cross-device send refused turn_conflict re-sends as a follow-up; a busy target is never 'unsupported' (#14320)

  • 18c86ed test(thread): a refused turn_conflict on a cross-device send (RED, #14320)

  • 0cabf94 test(thread): the last-listed memory names only the remembered device when two devices miss the round (#14318 delta review)

  • d4a5fe7 docs(changes): fragment for #14322

  • 5682973 perf(thread): the last-listed memory is updated in place, not copied per resolution (#14318 review L3)

  • e50f9a3 fix(thread-address-core): here with no device id is a refusal, not a device named here (#14318 review L2)

  • 2de6296 chore(thread-address-core): Node and Bun globals fail the typecheck of its sources (#14318 review L1)

  • 4b55029 test(thread): the two-device fixture names when the offline device was last seen (#14318 review M4)

  • ba77678 fix(remote): a device filter is checked by exact id again, so "here" is not_found (#14318 review M1)

  • 0922b4d fix(remote): omo remote resolves a typed device only when it is online, by every form (#14318 review H2)

  • 336f5fa feat(thread): thread_close { graceful } drains the turn and answers every unread message (#14313)

  • 39ac842 refactor(thread): delete resolveStoredSession, its last caller moved to the address module (#14305)

  • 4b23044 feat(thread): the thread SDK resolves every address through the shared module (#14305)

  • cc41f92 test(thread): the SDK resolves every address form through the shared module (RED, #14305)

  • 40bc908 docs(changes): fragment for #14321

  • 1514f3a docs(changes): fragment for #14318

  • 77b461a fix(thread): only the caller's online flag decides whether an offline device resolves, by every form (#14305)

  • ff2d456 style(thread): format the desktop-host leg test

  • fe2c517 Merge remote-tracking branch 'origin/fix/14278-roster-wait-state-truncation' into fix/14308-local-listing-state

  • 93d376f fix(thread): a listing of archived threads also reports the files it left unread; the incomplete-device refusal no longer says the device did not answer (review #14304 M1, L1, L3; L2 filed as #14315)

  • fc57619 fix(thread): find_session skips the primary by key, passes over a foreign dir's none, and bounds the fallback (#14271 review)

  • 395b289 fix(thread): an errored message with no content never renders the whole message with its uncapped error; carry dev's bundles (review #14283)

  • 68b2bf9 perf(thread): the omo.js side of locate and the lazy resolver stay within budget (#14305)

  • a27789e perf(thread): accountDevices stays as omo.js had it; the last-seen roster is the sidecar's (#14305)

  • eaf6231 fix(thread): the create refusal names the binary's schema first, the store's second (#14228 round 4)

  • a5341fd fix(thread): a device row carries last_seen_ms only when the service names one (#14305)

  • a9611f9 chore(thread): keep only used names on the address barrels (#14305)

  • 9023c90 style(thread): format the address module (#14305)

  • b09dfed feat(thread): callers resolve devices and addresses through the address module (#14305)

  • a1c7b25 refactor(remote): seams.ts parses with the core address grammar (#14305)

  • 4d327a8 feat(thread): address module API checkpoint (#14305)

  • dbf33cb feat(thread): the thread client sidecar re-exports resolveAddress (#14305)

  • 51b8873 test(thread): a desktop-app holder reopens a closed CLI TUI session like a CLI daemon (#14270)

  • fb70a0a chore(changes): fragment for #14309

  • c6cc517 fix(thread): thread_list says when an engine of this device did not answer, and keeps its threads' note (#14308)

  • 152e8dd fix(thread): one changelog fragment named for the PR; the create refusal says the thread exists (#14228 round 3)

  • c9c4cfb Merge pull request #14303 from code-yeongyu/fix/14300-windows-taskkill-bound

  • e942aa1 test(thread): one address module table and / through the control tools (RED, #14305)

  • dd77939 chore(changes): fragment for #14307

  • 039e8ef feat(thread): thread_list rows carry a running thread's provider retry wait as waiting_on (#14278)

  • d7d356f Merge remote-tracking branch 'origin/dev' into fix/thread-liveness-typed-state

  • fffcc30 chore(changes): fragment for #14304

  • 86e0d09 fix(thread): remove the last remnant of the dropped SDK schema branch (#14228 review round 2)

  • 1eb9f9b fix(thread): thread_list reports partial devices and the inactive threads it left out (#14295)

  • acc144b fix(thread): review round 2 - drop the dead SDK branch, clean the merge debris, align the changelog (#14228)

  • 030e18d fix(thread): name thread_create in its eval form in the target_closed next action (#14273)

  • a397780 fix(thread): a close on request settles the queued intake it superseded (#14273)

  • 4e214d2 Merge remote-tracking branch 'origin/dev' into fix/14271-find-session-route

  • 45d5ff1 fix(thread): find_session fallback passes over non-answers, asks the primary once, and names why it failed (#14271 review)

  • d2a4011 chore(train): Set A regen at the thread train 4 top after dev 822bd92 (v5.1.79)

  • 1a194b3 Merge dev 822bd92 (release v5.1.79) into thread train 4

  • 530d500 fix(thread): a reopen on another device checks a given name for conflicts (#14274)

  • b2d347c docs(changes): fragment for #14292

  • 16442fa style: bun fmt (#14265 lines)

  • 1427a40 chore(train): Set A regen after #14262's fix at the thread train 4 top

  • 0a1f478 docs(changes): #14279's host_unavailable answer in the thread train 4 fragment (review L1)

  • 80cb5c6 Merge #14262 @03f24e7b95 (route_unlisted keeps its reason when find_session is refused; train 4 review H1) into thread train 4

  • 661f614 fix(thread): keep the hold wait typed as a DeliveryWait kind

  • 03f24e7 fix(thread): a parked-row re-ask that find_session cannot answer keeps the route_unlisted refusal (#14240, #14237)

  • 87ca154 Merge dev (thread train 4 base abfb5be) into fix/14240-parked-live-row-wake

  • 687f713 fix(thread): format the merged sources (fmt check)

  • 1274b93 changes(14285): describe the reply refusal reasons

  • ac55184 chore(bundles): set omo.js budget to the thread train 4 top measurement + 4 KiB

  • c3ef1bb fix(thread): a reply to a sender on another device names why it cannot resolve; not_found only when the device answered (#14280)

  • a38e758 docs(changes): fold the thread train 4 member fragments into 14284.md

  • a8810d9 chore(train): Set A regen at the thread train 4 top (dev abfb5be)

  • d4e215d Merge origin/dev after train 3 (omo#14248) landed; keep dev's generated files (train-4 member rule)

  • 370df06 Merge #14279 (host_unavailable names the devices that answered (#14270)) into thread train 4

  • 441ed2d Merge #14265 (ship the thread client typebox in the compiled runtime, Fixes #14204) into thread train 4

  • 30fef89 Merge #14264 (older entry cursor answers cursor_stale, Fixes #14180) into thread train 4

  • 721643b Merge #14263 (bound thread_send reply-route device read by the call budget, Fixes #14226) into thread train 4

  • acea172 Merge #14262 (send to a parked live-listed thread re-asks the holder and wakes it, Fixes #14240) into thread train 4

  • 0c2b029 Merge #14261 (refuse thread_create with neither prompt nor thread, Fixes #14050) into thread train 4

  • aa89df6 Merge #14254 (attach-pool held-read expiry takes the port clock (#14252)) into thread train 4

  • 9feaea2 Merge #14281 (test(remote): neutral fixture device label) into thread train 4

  • e05caa1 fix(thread): thread_read shows a provider error's text instead of [] (#14278)

  • 57e70f7 Merge remote-tracking branch 'origin/dev' into fix/remote-tui-neutral-fixture-label

  • 1f4aa07 fix(thread): a reopen keeps the thread's name and applies a name given at reopen (#14274)

  • df5b237 chore(changes): fragment for #14281

  • 70d3e04 test(remote): neutral fixture device label

  • 1750fac fix(thread): host_unavailable for an unheld session id names the devices that answered (#14270)

  • 3ce9c69 fix(thread): find a closed thread's file on any engine its device lists when the primary predates find_session (#14271)

  • 7e8bb66 fix(thread): a route_unlisted retry announces its delivery once and keeps its stamp hop (#14240 review L)

  • 11d3aa5 docs(changes): fragment for #14265

  • d788725 fix(binary): inline typebox into the thread client sidecar so the CLI can load it (#14204)

  • 782cddd docs(changes): the change fragment for #14264 (#14180)

  • bc840d3 docs(changes): the change fragment for #14263 (#14226)

  • ee3e6e6 docs(changes): the change fragment for #14262 (#14240)

  • 6b1e23f docs(changes): the change fragment for #14261 (#14050)

  • 3abe6ee fix(thread): an older entry cursor answers cursor_stale; the CLI hint names older messages (#14180)

  • 2a6081f fix(thread): bound thread_send's reply-route device read by the call budget (#14226)

  • 7ee8234 fix(thread): a send to a live-listed thread its host parked re-asks the holder and wakes it (#14240)

  • c9d2531 fix(thread): refuse a blank thread_create prompt, which opened an unnamed empty thread (#14050)

  • b0475d7 fix(update): adopt finished version dirs, stop on a live installer, prune leftovers, regenerate the wrapper (#12022)

  • 61fe785 test(omo-native): no per-test timeouts in the built-binary test (lint ratchet)

  • 6a2ac89 docs(changes): add the #14256 fragment

  • 9baf96c fix(thread): adopt dev's gateway_schema_too_new refusal; extend it to thread_create's first prompt and the CLI SDK (#13927)

  • 87f6974 fix(update): update the omo-versions launcher layout on PATH (#12022)

  • d88ea39 test(omo-native): spawn the built binary exactly as g13's session attestation does (#14253)

  • e7d344c docs(changes): fragment for #14255

  • 84effb7 fix(omo-native): pass session-of-pid to the engine unchanged; built-binary test and an engine-dispatch guard (#14253)

  • 19c4757 test(thread): the attach pool's held-read expiry takes the port's clock; the 10-minute bound is tested (#14252)

  • 461c5eb Fold the review LOWs (#14235) into the typed schema refusal branch

  • ca3f177 Merge origin/dev after the 5.1.78 cut: keep the typed schema refusal through the thread_create refactor

  • 98e4397 fix(thread): review LOWs of the typed schema refusal - docs, comments, and a direct SDK row (#14228)

  • 6f3cc66 fix(thread): the CLI surface maps a too-new store schema typed too (#13927)

  • bf7d86b fix(thread): a store schema newer than this binary refuses typed, not internal_error (#13927)

  • 8a33a8c Merge origin/dev (630a673) into the branch; regenerate bundles once

  • 50bd9de chore(build): regenerate bundles after merging dev

  • c68a498 Merge dev into rt/13972-inbound-nextaction

  • dd39f65 chore(build): regenerate bundles after merging dev

  • cfc010c chore(build): regenerate bundles after merging dev

  • 3923bec Merge remote-tracking branch 'origin/dev' into oms/13994

  • 4687311 Merge remote-tracking branch 'origin/dev' into oms/13975

  • 705e906 chore(bundles): regenerate omo.js on #13972 aea22da with this PR's inbound next actions

  • 96908fd Merge branch 'rt/13942-nextaction' into rt/13972-inbound-nextaction

  • ff9df39 docs(changes): name the fragment for #13994, the number GitHub assigned

  • b4f8416 fix(thread): connector next actions for inbound's binding_inactive and unsupported

  • 19e4ad0 Merge origin/dev into rt/local-engine-msg

  • aa853d5 fix(remote): name this device's engine for every local attach failure, not only a missing one

  • e651364 fix(remote): name this device's engine, not the device service, when no local engine answers a follow or attach

  • 9ec48be fix(rpc): an extension-cancelled navigate/edit is not applied either (#13940 review L1)

  • 001ed06 fix(rpc): only an outcome "refused" is unapplied; the engine's success outcomes stay served (#13940 review M1)

  • 7b57b6d fix(rpc): a reply with an unapplied outcome is not served locally (unload_session is differs, #13940 review)

  • a5ad485 fix(rpc): the parity table shows a reply's unapplied outcome (unload_session refused, #13895 review)

  • 9cb68fc docs(changes): #13895 fragment states only numbers the parity table shows (113 of 128, 15 excluded, 36 unsafe-to-probe rows)

  • 3a0d95b docs(changes): #13895 fragment gives the measured split of the harmless-form probes (23 refused, 13 no-op)

  • f287223 docs(changes): #13895 fragment states exactly what was measured (113 of 128; 15 connection_local excluded; 36 in a harmless form)

  • 0e226a1 Merge branch 'dev' into rpc-relay-parity-measured

  • e178663 docs(changes): fragment for #13895

  • ac13d3c Merge branch 'dev' into rpc-relay-parity-measured

  • 42aaa91 test(rpc): measured local-versus-relay parity for every engine RPC method (#13478)

  • b4fe844 fix(service-auth): keep discardedOf module-private (knip: unused export)

  • 8be0c2d docs(changes): #13640 fragment follows the guarded migration

  • 43ac5e5 fix(service-auth): never let the Linux state file replace another keyring sign-in (#13640 review)

  • c2c3c96 fix(service-auth): fold the #13642 review LOWs

  • af4c05c chore(changes): fragment for #13642

  • c95af80 fix(service-auth): every sign-out retires the sync pin of the account it signs out of (#12308, #12384, #12383)

  • ceb42ba docs(changes): fragment for #13640

  • 5e6cf60 fix(service-auth): keep the Linux sign-in in a private state file when no Secret Service exists

  • 94d2ead docs(changelog): record the Kibitzer tool_result hang fix

  • 4b17f5e Merge remote-tracking branch 'origin/dev' into fix/13074-kibitzer-hook-never-blocks

  • 7c075f2 Merge remote-tracking branch 'origin/dev' into fix/13074-kibitzer-hook-never-blocks

  • a0b8b79 fix(memory): the Kibitzer tool_result hook starts recall delivery without awaiting it (#13074)

  • 5f45c8f test(memory): a Kibitzer recall delivery that never settles must not hold the PostToolUse hook (#13074)

  • 4d5d721 add changelog fragment for #11624

  • 558ba29 merge origin/dev into fix/11621-infra-retry-in-memory-journal

  • 923105d test(dag): hold the in-memory store to the file store contract

  • a335412 test(dag): journal the infra-retry cases in memory

bun add -g omo-ai

Don't miss a new oh-my-openagent release

NewReleases is sending notifications on new releases.