npm omo-ai 5.1.74
v5.1.74

30 minutes ago

A secret input or grant approval that settled while the relay link was down is now dismissed on the terminal that raised it after omo remote attach reconnects, instead of staying open with an answer the session would refuse. (#13773)

The computer-use conformance preflight test cleans up its compiled stand-in engine reliably on Windows. The test's temporary directory is now removed with the shared retrying helper, because Windows briefly keeps a just-run executable locked and Bun ignores rmSync's retry options. QA tooling only; no runtime behaviour changes. (#13299, part of #13221)

Fixed

  • The EQ.omo driver resolves a symlinked temp dir before it makes its sandboxes, so a recording on macOS no longer carries /private<sandbox> paths or an untrusted project. A new test fails when a platform's EQ.omo goldens were recorded with a senpi version more than a week away from the pinned one. (#13865)
  • CI: the desktop-bench comparison summary prints its regression thresholds (x1.3 and +0.5 ms) instead of undefined. It now reads them from compareBench's result rather than destructuring names eval-core.mjs does not export. (#13838, follow-up to #12876 review L1, #12567)
  • omo remote bootstrap installs device service 0.1.39 instead of 0.1.27, which was older than the wake and delivery minimum and answered open/create to the device with unsupported. (#13803)
  • A wake toward a device service older than the wake minimum now names the service's version, the minimum and the fix, instead of a bare unsupported. (#13803)
  • omo remote devices lists the device it runs on first (the account roster held only the other devices), and --json marks each row this_device. omo remote sessions --device with a device that is not on the account now fails with exit 2 naming it, instead of printing an empty list. (#13458, part of #13402)
  • thread_interrupt also stops the background jobs the interrupted turn started (a bash command it ran in the background or that auto-detached), so the thread no longer wakes a minute later when that job ends and replies as if it had never been interrupted. The result names what was stopped as background: { status: "stopped", stopped, failed }; an engine that predates the stop answers background: { status: "not_supported" }. A thread on this device and a thread on another device run the same interrupt code and answer the same shape. (#13444)
  • Computer use v3 (fake backend): the source guard that bans environment reads now also reports a grouped use std::{...} import of the environment module when a // or /* */ comment comes before the item inside the group. A comment can no longer hide an item, or split one at a comma or brace inside it. (#13363)
  • Desktop engine: the client retries a connection reset before the first reply byte only on Unix sockets, never on a Windows named pipe, where the daemon may already have run the request. (#12744)

Changed

  • The thread_read wait limit (600,000 ms) is now internal to the thread schema instead of an unused export; the limit itself is unchanged. No behavior change. (#13864)
  • Internal: the Windows backend documents when its posted key route still sends a key under a held Alt: for modifier-only chords, when releasing held keys, and while a failed Alt release is still recorded as held. A Windows test is also renamed to match the order of its checks. Nothing changes for users. (#13824, fixes #13821)
  • Internal: the module doc of the X11 input's client-process reader now names the /proc/<pid>/exe read added for the background-input check. Nothing changes for users. (#13820, fixes #13818)
  • /remote enable, /remote disable and /remote control in the TUI now answer in a sentence ("Remote sessions are on for this device.", "Remote control is off: ...") instead of the CLI's raw JSON. When /remote enable cannot finish, it shows the CLI's reason (for example "Could not download the OmO device service.") instead of an exit code, and a warning the CLI prints while succeeding (for example that the device is not linked) now shows after the result. (#13796)
  • The remote module's public types no longer depend on internal modules, so the importable remote SDK ships clean type declarations. No command or tool behaves differently. (#13782)
  • The remote module gains its client factories, one per kind of caller (a person, including omo remote; a session; a connector; the thread SDK's own principal; the device service's bootstrap entry), and its public entry. Its alias operations work when the caller hands in the account's alias registry, naming each session by the same durable id the thread tools use, and a device that does not answer is reported as offline rather than as having no such session. Nothing calls them yet, so no command or tool behaves differently. (#13750)
  • The remote module now declares every operation a surface needs (part 4a-2): opening, messaging, reading, following, waiting on, interrupting and closing a session on any device, its model and thinking level, attaching with pending approvals, aliases and the wake switch. The operations that have not landed yet answer "unsupported" as data, and nothing calls them yet, so no command or tool behaves differently. (#13747)
  • Remote access now has its own module (part 4a-1 of the one remote module design): typed results, one code list, a deadline table and a clock seam, with the service, devices and session list members. Nothing uses it yet, so no command or tool behaves differently. (#13498)
  • The device service client takes one injectable clock for its handshake, request, endpoint-command and attach waits, so each of those deadlines can be tested on fake time. Real timers stay the default and no wait changed. (#13473)
  • Computer use v3: on Windows, one user's two logon sessions take turns at the computer instead of acting at once: the interaction lock is now the operating system's file lock on %LOCALAPPDATA%\OmO\ComputerControl\run\interaction.lock, in a directory only that user and SYSTEM can open, on a local NTFS or ReFS volume. A session without a runtime directory cannot act on a real machine; it can still read. (#13367)
  • The Windows live tier's ledger job now fails when a fresh measurement's verdict differs from the committed crates/omo-computer-backend-win32/probes/hosted-windows-live.tsv, naming each leg that moved, so the committed ledger only changes on purpose. (#13274, part of #13221)

Added

  • Computer use v3: the session crate's side-effect funnel lands. Every mutating call runs the gate, writes a pre-input marker before the first OS call on every transport, performs the call, reads back and writes one audit event per side effect; a crashed call is never resent and reads delivery_uncertain. Audit, lock, permits, batches and artifact retention come with it, and a typed secret is audited by length only. Nothing user-visible changes yet; the existing desktop engine is untouched. (#13367)
  • Computer use v3: the safety crate omo-computer-safety lands. It holds the fixed gate order (each gate refuses before any OS call with nothing sent), the supervisor whose stop commit releases every hold within 1 s or exits, the stop latch and persisted stop marker under the per-uid state root, the live/deaf/none stop paths with the host-relay-only policy, the per-OS protected tables and dangerous chords, and the self set. The walking skeleton's click and typeText now pass through the gate. Nothing user-visible changes yet; the existing desktop engine is untouched. (#13353)
  • Computer use v3: the walking skeleton lands. omo-computer-engine runs on the fake backend with --stdio (NDJSON JSON-RPC), --schema, --selftest and --version; the session answers session.open, capabilities and close plus the thin click and typeText verbs; refusals go through the closed COMPUTER_* table. script/qa/computer/replay.ts replays conformance cases against the engine binary and prints the JSON pointer of the first mismatch. Nothing user-visible changes yet; the existing desktop engine is untouched. (#13352)
  • Spike SP-R probe and ledger for the computer-use relay pipe: auth lifetime, hub hibernation, message-size and liveness limits measured against a local relay, plus read-only checks on a same-network pair and the hosted relay. Probe-only, no engine change. (#13166, part of #13051)
  • Computer use v3: spike SP-W adds a Wayland probe and its ledger under crates/omo-computer-backend-wayland/probes/. The probe measures, in one container per compositor (GNOME 46, KDE Plasma 6, sway), whether portal restore tokens survive restarts, which portals exist, libei regions, a second ConnectToEIS, the clipboard grant, key release when an input owner dies, the GNOME extension and accessibility behaviour, and KWin window-management access. Each fact is one cited ledger row. Nothing user-visible changes. (#13163)
  • Desktop engine: several clients run by the same user can attach to one shared --serve daemon session (SDK attach over the Unix socket), each with its own replies; the daemon checks every client's peer credentials and closes a connection from another user before any byte. (#12744)

Known issues

  • bun fmt over the whole tree still flags the generated desktop prelude (packages/senpi-desktop-prelude/src/prelude.js): its generator does not yet emit formatter-clean output. Formatting only; the shipped prelude behaves the same. The fix is tracked in #13855.

  • abf50be ci(publish): operator-built darwin artifacts for the v5.1.74 release (workflows from #13928, merge 8f8f4d1)

  • c1e657c Merge pull request #13444 from code-yeongyu/fix/thread-interrupt-background

  • c84deb9 Merge origin/dev into fix/thread-interrupt-background

  • 19b1d48 Merge pull request #13864 from code-yeongyu/follow-up/low-13743

  • 2c32fec Merge origin/dev into fix/thread-interrupt-background

  • 64049c5 chore(bundles): regenerate this PR's bundle delta after the dev merge (#13864)

  • 6daca61 Merge remote-tracking branch 'origin/dev' into follow-up/low-13743

  • bbab1bd Merge pull request #13782 from code-yeongyu/rt/13414-types-closure

  • a36d311 Merge pull request #13908 from code-yeongyu/ci/publish-child-path-tag-and-release-target

  • 3e8f0a6 chore(bundles): regenerate this PR's bundle delta after the dev merge (#13864)

  • f3c6590 Merge origin/dev into fix/thread-interrupt-background

  • 1d4b4c7 Merge remote-tracking branch 'origin/dev' into follow-up/low-13743

  • 8234828 Merge pull request #13854 from code-yeongyu/test/cu-is-row04b-fixtures-push

  • 7e74bb8 Merge pull request #13300 from code-yeongyu/test/cu-is-row04b-rects

  • 137d902 Merge pull request #13903 from code-yeongyu/ci/platform-decision-warning-cap

  • cc4c41e Merge pull request #13298 from code-yeongyu/test/cu-is-row04b-fixtures-p2

  • 7ba14fc Merge pull request #13367 from code-yeongyu/feat/cu-v3-row17-mutate-funnel

  • 7d72761 Merge pull request #13773 from code-yeongyu/fix/remote-attach-raiser-only-settled

  • a3ca2fb Merge pull request #13750 from code-yeongyu/rt/13414-u3-factories

  • fb83e53 Merge pull request #13747 from code-yeongyu/rt/13414-4a2-types

  • dd6a305 Merge remote-tracking branch 'origin/dev' into ci/publish-child-path-tag-and-release-target

  • 4859721 Merge pull request #13353 from code-yeongyu/feat/cu-v3-row16-safety

  • 60d3e14 ci(publish): tag the prepared source on the child path and target it in the release

  • f642f31 Merge pull request #13869 from code-yeongyu/ci/pr-bots-tracker-lows

  • cfed9ec Merge pull request #12744 from code-yeongyu/cu-v2/task-25-p3b-serve-concurrent-same-uid-clients-on

  • abb746d chore(bundles): regenerate this PR's bundle delta after the dev merge (#13864)

  • 244ed8d Merge remote-tracking branch 'origin/dev' into follow-up/low-13743

  • b5cba93 chore(bundles): regenerate this PR's generated files on #13747 8d1a33a

  • 83329c1 Merge branch 'rt/13414-4a2-types' into rt/13414-types-closure

  • ee55110 ci(pr-platform-decision): the truncation warning only above the API cap (#13859)

  • f3a61ff Merge origin/dev into fix/thread-interrupt-background

  • 734a4d1 Merge pull request #13876 from code-yeongyu/ci/platform-decision-truncated-files

  • ba7c85a test(platform-red-tracker): portable fake gh queue (#13840)

  • 15e2741 Merge pull request #13498 from code-yeongyu/feat/remote-l2-client-4a1

  • f65f8ef Merge pull request #13352 from code-yeongyu/feat/cu-v3-row15-engine

  • c4e0542 Merge branch 'rt/13414-4a2-types' into rt/13414-u3-factories

  • 8d1a33a chore(bundles): regenerate this PR's generated files on #13498 a777370 (dev 4b07590)

  • dba00eb Merge pull request #13738 from code-yeongyu/fix/computer-macos-pool-review-lows

  • 06b680f Merge branch 'feat/remote-l2-client-4a1' into rt/13414-4a2-types

  • 8a0399d Merge pull request #13271 from code-yeongyu/ci/computer-live-macos-pool-fixes

  • b29d4a3 Merge pull request #13166 from code-yeongyu/spike/cu-v3-row10-relay

  • d280f43 Merge pull request #13274 from code-yeongyu/ci/cu-is-row6b-ledger-drift

  • 1ab9e5a Merge pull request #13299 from code-yeongyu/test/cu-is-row4e-preflight-cleanup

  • 669e798 refactor(remote): the public types reach the local API's wire vocabulary through device-service-core/src/wire.ts

  • 996164b docs(changes): add the changelog fragment for #13864

  • 3c7e254 chore(bundles): regenerate this PR's bundle delta after the dev merge (#13864)

  • 9e3b13c Merge remote-tracking branch 'origin/dev' into follow-up/low-13743

  • a777370 chore(bundles): regenerate this PR's generated files on dev 4b07590

  • 8a2e39a Merge branch 'dev' into feat/remote-l2-client-4a1

  • 4b07590 Merge pull request #13163 from code-yeongyu/spike/cu-v3-row8-wayland

  • fdc2986 chore(bundle): regenerate the computer-use extension bundle for this PR's sources on the dev merge

  • d1b5200 test(remote): the types closure may reach thread/remote/alias-port.ts, the import-free alias port

  • 766fac7 Merge pull request #13886 from code-yeongyu/ci/publish-gate-reuse-receipt-waiver

  • 9074b57 Merge pull request #13260 from code-yeongyu/ci/cu-is-row07d-teardown-unread

  • be096f6 Merge pull request #13803 from code-yeongyu/fix/13763-device-service-pin-0139

  • 877f728 chore(bundles): regenerate service-auth.generated.js for this PR's remote-attach changes (Linux, current dev b597f10)

  • f1a1d5e Merge pull request #13824 from code-yeongyu/fix/13821-review-lows

  • df6a038 test(platform-red-tracker): skip the fake-gh writer cases on Windows (#13840)

  • 8e8f334 Merge pull request #13820 from code-yeongyu/docs/13818-client-process-exe-doc

  • 4b2887c Merge the row 4b fixtures head (525d214: dev b597f10 + shared platform decision) into this branch

  • ab9d362 Merge the row 4b fixtures head (525d214: dev b597f10 + shared platform decision) into this branch

  • 525d214 ci(conformance fixtures): the macOS and Windows fixture jobs follow the shared PR platform decision

  • 8a0452d ci(pr-platform-decision): review fixes for #13876 (#13859)

  • 62731d4 refactor(remote): the alias port's types move to an import-free leaf, thread/remote/alias-port.ts

  • 3785fee Merge dev b597f10 into test/cu-is-row04b-fixtures-p2

  • 6233158 Merge remote-tracking branch 'origin/dev' into fix/remote-attach-raiser-only-settled

  • c90b128 test(remote): the types closure may reach thread/contracts/permission-presets.ts, the leaf list of approval modes

  • f144b4a ci(publish): add a logged maintainer waiver of the gate-reuse hosted CI wait

  • b597f10 Merge pull request #13836 from code-yeongyu/fix/13819-packed-install-own-cache

  • 940c0f9 Merge branch 'rt/13414-4a2-types' into rt/13414-types-closure

  • ddcbb2c Merge branch 'rt/13414-4a2-types' into rt/13414-u3-factories

  • fa05c41 fix(remote): the remote module re-exports the approval-mode type only; the one list is the thread contracts'

  • 699b6c3 Merge pull request #13363 from code-yeongyu/fix/cu-v3-row14-guard-comment-items

  • 9d43de8 chore(bundles): regenerate this PR's bundle delta after the dev merge (#13803)

  • 5f55878 chore(bundles): regenerate this PR's generated files on #13498 b42ef25 (dev f007ca9)

  • d5193b6 Merge origin/dev into fix/thread-interrupt-background

  • 6cff722 Merge remote-tracking branch 'origin/dev' into cu-v2/task-25-p3b-serve-concurrent-same-uid-clients-on

  • d90a00d Merge branch 'feat/remote-l2-client-4a1' into rt/13414-4a2-types

  • 9eb1c36 Merge branch 'feat/remote-l2-client-4a1' into rt/13414-4a2-types

  • b42ef25 docs(remote): point BOUNDS at the local API's moved timeout constants

  • 258c171 ci(pr-platform-decision): a truncated file list fails toward the full matrix for a labelled PR (#13859)

  • bf432a5 Merge remote-tracking branch 'origin/dev' into fix/13763-device-service-pin-0139

  • 40841a5 ci(pr-platform-decision): say when the PR file list is truncated (#13859)

  • f0f60b9 chore(bundles): regenerate this PR's generated files on dev f007ca9

  • edbc8eb Merge pull request #13860 from code-yeongyu/ci/full-matrix-label-needs-reason

  • c433a72 Merge branch 'dev' into feat/remote-l2-client-4a1

  • fd905ca Merge pull request #13796 from code-yeongyu/fix/13412-remote-tui-review-lows

  • f007ca9 Merge pull request #13458 from code-yeongyu/fix/remote-devices-this-device

  • dccccf8 Merge pull request #13838 from code-yeongyu/fix/12876-bench-message-constants

  • 9c6106f Merge pull request #13865 from code-yeongyu/test/eq-golden-pin-freshness

  • dfe02bb Merge pull request #13473 from code-yeongyu/feat/remote-clock-seam-fakes

  • 4dde428 ci(pr-platform-decision): always check out the classifier sparsely; PR files from the API (#13859)

  • b1cf8eb ci(pr-bots): review LOWs from #13833 and #13852 (#13840)

  • a9fb727 docs(changes): add the changelog fragment for #13865

  • 83ed56a fix(eq): resolve a symlinked temp dir before the EQ.omo driver makes its sandboxes

  • 163065c Merge remote-tracking branch 'origin/dev' into fix/remote-attach-raiser-only-settled

  • cdbb93c chore(bundle): regenerate omo.js for #13796's /remote renderer delta on dev 7b81414 (+491 B)

  • d0aa85a chore(bundles): regenerate this PR's generated files on the current dev

  • ca37642 test(ci): the fixture workflow's concurrency is checked by the existing path-filtered workflow table

  • 52e0db2 Merge branch 'dev' into feat/remote-clock-seam-fakes

  • 20c8b53 test(cu-eval): pin the ratio and floorMs fields of compareBench's result that desktop-bench prints (#13838 review L1)

  • 2ba9dfa chore(bundles): regenerate this PR's bundle delta after the dev merge (#13803)

  • d2c4bf6 Merge remote-tracking branch 'origin/dev' into fix/13412-remote-tui-review-lows

  • 13e57bb Keep thread read wait limit internal

  • 98e2ffb test(ci): the tracked-workflow cancel guard renders the concurrency group template part by part

  • 9fae4b8 Merge remote-tracking branch 'origin/dev' into fix/13763-device-service-pin-0139

  • 779a1de Merge remote-tracking branch 'origin/dev' into fix/12876-bench-message-constants

  • 5f772ae Merge remote-tracking branch 'origin/dev' into fix/13819-packed-install-own-cache

  • 6ac3cd1 ci(conformance fixtures): a dev push run of the fixture workflow is never cancelled by another run

  • acf09bd Merge dev 871a808 (#13120 row 14 merged) into row 15 (merge only)

  • fed3171 build(remote): regenerate service-auth.generated.js for #13458's this-device row and --device check

  • fdb426d Merge origin/dev into fix/remote-devices-this-device

  • bf29b84 docs(computer): name the failed-Alt-release case where the posted route still sends WM_SYS* (#13824 review)

  • 0c928f5 ci(conformance fixtures): the fixture workflow also runs on a push to dev, and the platform-red tracker watches it

  • 3f6d3a8 chore(bundles): regenerate service-auth.generated.js for this PR's remote-attach changes (Linux, current dev b9be6e5)

  • 949afda Merge the row 4b fixtures head (85b628c: macOS/Windows fixture jobs gated on ci:full-matrix) into this branch

  • 85b628c ci(conformance fixtures): the macOS and Windows fixture jobs run on a pull request only with ci:full-matrix

  • 26682da chore(bundles): regenerate this PR's generated files on dev after 5.1.73 (#13627)

  • 2ddb5b7 Merge branch 'feat/remote-clock-seam-fakes' into feat/remote-l2-client-4a1

  • 4d681c0 chore(bundles): regenerate this PR's bundle delta (#13803)

  • a513d87 chore(bundle): regenerate omo.js for #13796's /remote renderer delta (+491 B)

  • 2397c84 chore(bundles): regenerate this PR's generated files on dev after the regen wave (#13810)

  • e1e40d9 Merge row 16 (819b4ed) into this branch (row 17 Windows poison-record fix, merge only)

  • 819b4ed Merge row 15 (d54f4cc) into this branch (row 17 Windows poison-record fix, merge only)

  • 2b47a7f Merge remote-tracking branch 'origin/dev' into fix/13763-device-service-pin-0139

  • d54f4cc Merge row 14 (bf84999) into this branch (row 17 Windows poison-record fix, merge only)

  • c971132 Merge row 14 (bf84999) into this branch (row 17 Windows poison-record fix, merge only)

  • c453761 Merge remote-tracking branch 'origin/dev' into fix/13412-remote-tui-review-lows

  • af283ea Merge remote-tracking branch 'origin/dev' into fix/remote-attach-raiser-only-settled

  • 5bb7430 Merge branch 'dev' into feat/remote-clock-seam-fakes

  • 74e550c test(omo-native): limit the isolated-cache comment to the measured rpc-mode.js rewrite; drop the fragment, test paths do not ship (#13836 review L1, L3)

  • 41bbd43 Merge remote-tracking branch 'origin/dev' into docs/13818-client-process-exe-doc

  • 1befdc6 Merge commit '0543add10fd4984883d58a687e84bdc1120b47b4' into inst-13299

  • a0becc8 Merge commit '0543add10fd4984883d58a687e84bdc1120b47b4' into inst-13274

  • afc0f53 Merge #13271 head 18c6b11 (carries dev 0543add) into fix/computer-macos-pool-review-lows

  • a004ab8 Merge the row 4b fixtures head (1f05c46, dev 0543add taken in) into this branch

  • 18c6b11 Merge dev 0543add into ci/computer-live-macos-pool-fixes

  • b515e6c Merge commit '0543add10fd4984883d58a687e84bdc1120b47b4' into inst-13260

  • a483c3e Merge remote-tracking branch 'origin/dev' into fix/13821-review-lows

  • 1f05c46 Merge dev 0543add into test/cu-is-row04b-fixtures-p2

  • 99e4a07 fix(ci): desktop-bench names eval-core.mjs as the source of its thresholds instead of restating them (#13838 review L1)

  • 99fbf85 Merge row 16 (b064887) into this branch (row 17 Windows poison-record fix, merge only)

  • b064887 Merge row 15 (e829758) into this branch (row 17 Windows poison-record fix, merge only)

  • e829758 Merge row 14 (fd4b11b) into this branch (row 17 Windows poison-record fix, merge only)

  • 652ae28 Merge row 14 (fd4b11b) into this branch (row 17 Windows poison-record fix, merge only)

  • 0e1b316 docs(changes): name the fragment after the PR's real number (#13838)

  • 9505a4f docs(changes): name the fragment after the PR's real number (#13836)

  • 47897dd chore(bundles): regenerate this PR's generated files on dev after the regen wave (#13810)

  • 6016262 Merge remote-tracking branch 'origin/dev' into fix/remote-attach-raiser-only-settled

  • 0d37441 Merge commit 'd9ccd6420db467310830403e5e45225feffde4d7' into inst-13299

  • fdf64e9 Merge #13271 head 5efe416 (carries dev d9ccd64) into fix/computer-macos-pool-review-lows

  • 5efe416 Merge dev d9ccd64 into ci/computer-live-macos-pool-fixes

  • 8f0f323 Merge commit 'd9ccd6420db467310830403e5e45225feffde4d7' into inst-13274

  • 563ba41 Merge commit 'd9ccd6420db467310830403e5e45225feffde4d7' into inst-13260

  • 79f6f82 Merge the row 4b fixtures head (b2022d2, dev d9ccd64 taken in) into this branch

  • b2022d2 Merge dev d9ccd64 into test/cu-is-row04b-fixtures-p2

  • 126c287 Merge remote-tracking branch 'origin/dev' into docs/13818-client-process-exe-doc

  • e7ccc1a Merge remote-tracking branch 'origin/dev' into fix/13821-review-lows

  • 7fb2763 Merge branch 'dev' into feat/remote-clock-seam-fakes

  • 591c029 docs(changes): fragment for the push; renamed to the real PR number after gh pr create

  • 0b5db5f test(omo-native): packed-install installs into its own Bun cache, so the engine prepare cannot rewrite the host's shared cache entry (#13819)

  • 40f3ec3 docs(changes): fragment for the push; renamed to the real PR number after gh pr create

  • 2745f15 Merge remote-tracking branch 'origin/dev' into fix/12876-bench-message-constants

  • b6a3e1f fix(remote): drop the G1 S2 rows this PR's port move retired

  • 69b37c3 Merge the row 4b fixtures head (e68cc1f, dev 70e4e49 taken in) into this branch

  • e68cc1f Merge dev 70e4e49 into test/cu-is-row04b-fixtures-p2 (knip.base.jsonc: dev's local-api ignore entry kept beside this PR's Electron ignoreIssues block)

  • 31a2375 Merge branch 'feat/remote-clock-seam-fakes' into feat/remote-l2-client-4a1

  • 69e2ad5 docs(changes): fragment for #13824

  • ec5aaa6 docs(computer): say where the posted route's alt_down still applies, and name the mouse-hold test after its real ordering (#13821)

  • 69df026 Merge remote-tracking branch 'origin/dev' into fix/remote-attach-raiser-only-settled

  • 5c0b261 Merge branch 'dev' into feat/remote-clock-seam-fakes

  • 8b2649a Merge remote-tracking branch 'origin/dev' into fix/13412-remote-tui-review-lows

  • 18500a3 Merge branch 'dev' into feat/remote-clock-seam-fakes

  • 67ba933 docs(computer): rewrap the client_process module doc; changelog fragment for #13820

  • 4692f9d docs(computer): client_process module doc names the /proc//exe read (#13818)

  • bc41155 Merge branch 'feat/remote-clock-seam-fakes' into feat/remote-l2-client-4a1

  • 7ef996e test(computer): the marker-unavailable standalone case takes the plain-file plant on Windows too: a 0500 mode is a no-op there, so the marker write went through and the click answered null (hosted windows-11-arm job 133186717397 at a685bd5: v3-r17-marker-unavailable.json step 3 left: Some(Null) right: None); test-only

  • 6d73ca3 Merge branch 'dev' into feat/remote-clock-seam-fakes

  • 02233f5 Merge branch 'rt/13414-4a2-types' into rt/13414-types-closure

  • c716053 Merge branch 'rt/13414-4a2-types' into rt/13414-u3-factories

  • 0a51de2 docs(changes): add unreleased entry for #13803

  • 620ce03 Merge branch 'feat/remote-l2-client-4a1' into rt/13414-4a2-types

  • e5d1187 fix(device-service): pin device service 0.1.39 and name an outdated service on wake

  • 2427c4b chore(changelog): undo the fold of row 14 fragment into this PR: a stacked spine PR keeps its own fragment only over its stack parent (lead ruling; ancestor fragments waived)

  • 92f61f8 fix(remote): engine code table is a Map, so no non-null assertion

  • edfcde3 docs(changes): /remote renders enable, disable and control (#13796)

  • a685bd5 fix(computer): the poison record flushes its directory through the durable helper per OS: a directory cannot be opened as a file on Windows (os error 5), so on windows-11-arm every crashed callId stayed unrecorded with its marker kept and the scope wedged (hosted job 133173784375 at 21050ac: "crashed callId c-crash not poison-recorded, marker kept: Access is denied. (os error 5)"; A-IS-65 Windows marker)

  • 74cce33 test(remote): a raiser-only dialog learned from the mirror record is dismissed after a relay gap (#13773 review LOW)

  • d5f071c style(computer): format the macOS pool up-hold files (bun fmt)

  • 05dd814 chore(changelog): fold row 14 fragment (#13120) into this PR one fragment (bun script/changelog-fragments-cli.ts fold-batch --base origin/dev --pr 13363): a stacked PR adds exactly one fragment (changelog-gate)

  • 3c6b27e fix(ci): print the bench regression thresholds from compareBench's result, not unexported names (#12876 review L1)

  • 21050ac fix(computer): refusals built from the typed Reason dev #13137 introduced (the audit record's reason is reason.wire()); and the one-file-per-hand test expects AUDIT_LOCK_FILE under the writer's own cfg(not(unix)) instead of a hard-coded 'audit.lock' under cfg!(windows) (review LOWs on #13368 and #13390)

  • 69e8d1d Merge row 16 (835fde3) into this branch (W2 D1 dev take: dev a6901c2 with #13137's typed Reason, merge only)

  • 835fde3 fix(computer): refusals built from the typed Reason dev #13137 introduced (Refusal::new takes a Reason; the wire token is reason.wire(); tests compare Some(Reason::...)); no behaviour change

  • 533e276 docs(changes): fragment for the remote public type closure (#13782)

  • ea7ec39 refactor(remote): keep the public types' import closure leaf-only

  • a59c6a8 Merge commit '3dc66046e78b0da9416abf44e7b0acb75ce12ca6' into inst-13299

  • 94047d3 Merge commit '3dc66046e78b0da9416abf44e7b0acb75ce12ca6' into inst-13274

  • 0607f5e Merge commit '3dc66046e78b0da9416abf44e7b0acb75ce12ca6' into inst-13260

  • 65f52ad Merge row 15 (9901de9) into this branch (W2 D1 dev take: dev a6901c2 with #13137's typed Reason, merge only)

  • 9901de9 fix(computer): refusals built from the typed Reason dev #13137 introduced (Refusal::new takes a Reason; the wire token is reason.wire(); tests compare Some(Reason::...)); no behaviour change

  • 79a9d9b Merge row 14 (2c4c023) into this branch (W2 D1 dev take: dev a6901c2 with #13137's typed Reason, merge only)

  • 878589d Merge row 14 (2c4c023) into this branch (W2 D1 dev take: dev a6901c2 with #13137's typed Reason, merge only)

  • 1b49ae9 docs: change fragment for #13773

  • 718939a fix(remote): a raiser-only dialog settled during a relay gap is dismissed after the re-attach

  • 2466e60 fix(remote): every connect-phase failure says phase connect

  • ff95f30 Merge branch 'feat/remote-l2-client-4a1' into rt/13414-4a2-types

  • 90de328 fix(remote): prototype-named engine codes fall back to unsupported; a type assertion replaces ts-expect-error

  • 9396f0a fix(computer): the Windows lock directory check compares SIDs in the form the read-back SDDL writes them (an alias such as LA for the local Administrator), and the one-file-per-hand test expects audit.lock beside the data file on Windows (hosted windows-11-arm at b594441: "the lock directory is not protected as A-IS-65 L3 requires: its owner is LA")

  • d33134f feat(remote): RemoteClientDeps.serveAttach: the caller injects the attach endpoint server

  • 2863583 fix(remote): an alias holder's device that misses the listing is offline, not missing; listings run outside the aliases window

  • b6e0eac Merge branch 'rt/13414-4a2-types' into rt/13414-u3-factories

  • fab9fc2 docs(remote): name the client version field, not a command, in the factory docs

  • abbd43d fix(remote): tests for the client's 4a-2 behaviour; one permission-preset list; stub and doc agree

  • 9b986e2 build(remote): regenerate service-auth.generated.js for the this-device row and the --device check (#13402)

  • 15dcf94 fix(remote): factories take the caller's version; the module reads no product config

  • a338006 feat(remote): the barrel exports the thread seams (canonicalizeThreadHandle, handleOf, describeThread)

  • 2c188bb feat(remote): RemoteClientDeps.permissionPreset and OpenResult.warnings

  • 66aaa95 fix(remote): remoteClientForService for the bootstrap entry; createRemoteClient leaves the barrel; cli is the thread SDK's principal

  • 419783f docs(changes): fragment for the remote client factories and aliases (#13750)

  • 968b346 fix(remote): aliases hold a session by its durable id, like the thread tools (#13414 U3)

  • aedd669 Merge remote-tracking branch 'origin/rt/13414-4a2-types' into rt/13414-u3-factories

  • 190be39 feat(remote): RemoteClientDeps.connect and gatewayWorkerUrl seams

  • 6740299 feat(remote): aliases.resolve/set/clear over the injected account alias registry (#13414 U3)

  • 20db46e Merge remote-tracking branch 'origin/rt/13414-4a2-types' into rt/13414-u3-factories

  • 3a7c77d feat(remote): RemoteClientDeps.aliasRegistry, injected by the caller

  • 41267c6 test(remote): format the factories test

  • dc1edbe docs(changes): fragment for the remote module contract (#13747)

  • 54f422a Merge remote-tracking branch 'origin/rt/13414-4a2-types' into rt/13414-u3-factories

  • e9935a5 feat(remote): client factories, the module barrel, and aliases answering unsupported without account access (#13414 U3)

  • a957ac4 Merge branch 'feat/remote-l2-client-4a1' into rt/13414-4a2-types

  • a8f5dc6 fix(remote): engine codes that are already remote codes keep their names; ThreadHandle is branded

  • 582167e fix(remote): take device-service values from the local-only subpath, not the barrel

  • 5c70987 feat(remote): the whole RemoteClient contract for #13414 (4a-2 types and wiring)

  • b7b2d35 Merge branch 'feat/remote-clock-seam-fakes' into feat/remote-l2-client-4a1

  • 2c68ed6 fix(remote): map dev's unknown_delivery and point dev's two new port tests at the moved port

  • 1ec74a0 fix(remote): this device's row is online only while the service is connected and carries no invented kind; --device accepts a device the listing has (review L1, L2, L3)

  • b594441 Merge row 16 (14a1590) into row 17 (cascade: dev taken once at row 14)

  • ef2c8ea style(computer): --selftest keeps its check table layout; only the interaction-lock lines are added (no behaviour change)

  • 9f81c5d test(remote): this device's row follows the service state and invents no kind; --device accepts a device the listing has (review L1, L2)

  • 5f7e02b Merge origin/dev into fix/remote-devices-this-device

  • 87190d9 fix(device-service-core): run the shared delegated-command bound on the connection's timers after the dev merge

  • 8a069fe Merge branch 'dev' into feat/remote-l2-client-4a1

  • ee7c478 fix(computer): macOS pool up holds its guest across the clone-to-run gap; sweep reports a list failure

  • 688e915 test(computer): macOS pool up must hold its guest across the clone-to-run gap; sweep must survive a list failure (red)

  • fbe7fdc Merge branch 'dev' into feat/remote-clock-seam-fakes

  • 4420da8 fix(remote): omo remote interrupt names the background jobs it stopped, failed to stop or left running (#13444 review L1)

  • c559bba Merge remote-tracking branch 'origin/dev' into rt-interrupt-13444

  • 14a1590 Merge row 15 (eea07bf) into row 16 (cascade: dev taken once at row 14)

  • c780e0e docs(computer): the lock module states the funnel's restore contract; DEFERRED-LEGS owes the hold-through-restore proof to rows 44, 57, 49 and 67; the changelog says in plain words that on Windows a user's two logon sessions take turns and a session without a runtime directory cannot act on a real machine (A-IS-65 L1, R70cn(3), R70cn(4), R70cp)

  • 7dfb9a2 fix(computer): the Windows lock directory is created with the protected DACL of the user and SYSTEM and read back; an existing directory this user owns is tightened to it, one another principal owns, a reparse point, or a DACL that cannot be set or read back refuses (A-IS-65 L3, L2)

  • fa4a4a4 test(computer): RED - the Windows lock directory is admitted only with the protected DACL of the user and SYSTEM, read back; another owner, an unprotected DACL, a missing, extra or weaker entry refuses (A-IS-65 L3, T11 pure leg)

  • c1bdcdf conformance fixtures: the GTK list lays out its new rows before the report is written, and the geometry leg cannot pass on stale rects

  • 38378b8 fix(computer): the Windows RUNTIME root is %LOCALAPPDATA%\OmO\ComputerControl\run, where interaction.lock lives, and --selftest prints the interaction lock's path from the one resolver, or why none resolves (A-IS-65 L3)

  • 2c0f58a test(computer): RED - --selftest prints where the interaction lock is taken, from the one RUNTIME-root resolver; the Windows lock directory is %LOCALAPPDATA%\OmO\ComputerControl\run (A-IS-65 L3)

  • 925a854 fix(computer): the audit writer's per-append lock goes through the one lock helper - flock on the data file on unix (unchanged), the hand's audit.lock on Windows, never the data file; a lock that cannot be created, opened or taken fails the append (A-IS-65 L7, L8)

  • 24e2cdb test(computer): RED - two processes appending to one hand's audit at once leave lines that all parse and gapless segments; on Windows an append in one process never fails a read of the audit file in another (A-IS-65 L7, T6, T12)

  • eea07bf Merge row 14 (32fe58b) into row 15 (cascade: dev taken once at row 14)

  • 97cfc97 Merge row 14 (32fe58b) into the row 14 LOW follow-up (dev taken once at row 14)

  • 8eec362 fix(computer): macOS pool sweep keeps a guest an up is cloning or starting; up survives a list failure

  • f115745 fix(computer): the volume admission reads whether the volume is remote as well as its filesystem; a Windows lock file is relied on only on a local NTFS or ReFS volume, through the same helper the markers use (A-IS-65 L9)

  • e3ad0b9 test(computer): macOS pool sweep must keep a guest an up is starting, and the up wait must survive a list failure (red)

  • 918ecae test(computer): RED - a remote volume refuses even when it reports NTFS or ReFS; the admission reads the volume, not only its filesystem name (A-IS-65 L9, T7 pure leg)

  • 59fce60 Merge the row 4b fixtures head (4addc97, dev 3975913 taken in) into the rects branch

  • 01f25a8 Merge #13259 head ae9a9a7 (carries dev 3975913) into ci/computer-live-macos-pool-fixes

  • 4addc97 Merge dev 3975913 into test/cu-is-row04b-fixtures-p2

  • daacce7 Merge remote-tracking branch 'origin/dev' into inst-13299

  • bc8971a Merge remote-tracking branch 'origin/dev' into inst-13274

  • 671b2ff Merge remote-tracking branch 'origin/dev' into inst-13260

  • cfaf7f9 Merge dev into cu-v2/task-25-p3b-serve-concurrent-same-uid-clients-on

  • af6b27a docs(changes): name the fragment for its PR (#13498)

  • b6d20db fix(computer): the interaction lock is the OS file lock on every OS through ONE helper - flock on unix (open path unchanged), std File::lock/try_lock on Windows on a fresh non-inheritable handle shared read/write only, a reparse point or non-regular file refuses, unlock then close on drop, any other target refuses; the row-60 deferral comment is removed (A-IS-65 L1, L3, L8, L10(c), L12)

  • 753e89a style(thread): format interrupt.ts and its parity test (#13444 hosted fmt:check red)

  • b9e16a1 conformance fixtures: the WKWebView fixture removes its own WebKit networking temp folders at quit, after the networking process holding them exits (bounded, a typed failure if not)

  • fecd746 conformance fixtures: the driver points each fixture's XDG homes and CFFIXED_USER_HOME into the run's profile directory (green)

  • a879503 conformance fixtures: a fixture's XDG data, cache and config homes and its CoreFoundation home must sit in the run's own directory (red)

  • 37fb78c Merge dev 83373dc into test/cu-is-row04b-fixtures-p2 (take dev's tip; merge only)

  • 5724343 test(computer): RED - the interaction lock excludes across processes, a killed holder frees it, held is not an error, a holder's child does not keep it, a link at the lock file refuses (A-IS-65 T2, T3, T4, T10, T11 reparse leg; red on the Windows legs)

  • c8ef58d fix(computer): a session with no runtime root refuses every exec on a real backend, with nothing sent; only the fake acts without the interaction lock (A-IS-65 L10(b))

  • 7a7c586 Merge branch 'dev' into ci/cu-is-row6b-ledger-drift

  • 5b9e140 test(computer): RED - a real-backend session with no runtime root refuses exec with nothing sent (A-IS-65 L10(b), T9)

  • 953195a docs(changes): name the fragment for its PR (#13473)

  • 65bb6f7 feat(device-service-core): clock seam and fake-owner failure modes

  • 744ee05 test(omo-native): the remote interrupt --json result now carries the typed background field

  • 48bfc40 test(computer): the one-file-per-hand test compares path components, not a '/' string (Windows red at row 17, R70ch(2); assertion unchanged)

  • dde6078 fix(computer): both markers make their directory entry durable through ONE shared helper, so the pre-input marker writes on Windows (Windows reds at row 17, R70ch(1))

  • 6cb592c Merge commit '83373dcbb545ae2336e58abb46af028e9ad29a9e' into test/cu-is-row4e-preflight-cleanup

  • 5c99535 changes: fragment for #13458

  • c3ef0d5 fix(remote): omo remote sessions --device names an unknown device

  • 87f1163 fix(remote): omo remote devices lists this device

  • 08496ce changes: fragment for #13444

  • 2ac1a6a fix(thread): thread_interrupt stops the interrupted turn's background jobs

  • c8a5c9d conformance fixtures: the self-test usage names the a11y sabotage target, as the README does (prose only, no test)

  • d68eaa2 docs(computer): C2's restart leg is unit-proved; the engine's hello.generation wiring is owed to row 28 (review L1 of #13367)

  • 9a13836 fix(computer): record and release take an artifact digest only as a plain path component (review L2 of #13367)

  • 84c977c test(computer): RED - an artifact digest that is not a plain path component is refused (review L2 of #13367)

  • 6bb992e Merge row 16 (f8489aa) into row 17 (R70t: rows 15 and 16's native fixes)

  • f8489aa Merge row 15 (353e2d1) into row 16 (R70t: row 15's native fix)

  • 8fc95e2 fix(computer): the native contract builds and passes on macOS and Windows (absolute() off on macOS; a write-through rename commits the stop marker on Windows)

  • faba871 conformance fixtures: README documents --app-key, --a11y, the a11y sabotage target, the hermetic leg and CONFORMANCE_FIXTURE_DATA_DIR (prose only, no test)

  • 2f92e4b conformance fixtures: the hermetic leg runs last against a baseline taken before the first leg, and checks macOS HTTPStorages, Saved Application State and Preferences (green)

  • 41fe3bf ci(computer): compare the Windows ledger after the plan is written

  • 7abe552 conformance fixtures: the hermetic check must count leftovers from a baseline taken before the first leg, and macOS bundle-id app data (red)

  • 51974df fix(computer): age.rs names SESSIONS_DIR inside its unix-only test, so the Windows build has no unused import

  • 353e2d1 fix(computer): the replay driver and its tests run on Windows (case names with '/', compiled stand-in engines)

  • 7abf0fe docs(changes): changelog fragment for #13367 (cu-v3 row 17)

  • 67506de fix(computer): the fake's source guard ends a block comment at its own close, counting nesting

  • 8b8cdfc test(computer): the fake's source guard must catch an env item after a nested block comment

  • 7b95593 docs(changes): fragment for #13363

  • d3a7e8f fix(computer): the fake's source guard skips comments inside a grouped std import

  • 17aa43a test(computer): the fake's source guard must catch an env item preceded by a comment inside the group

  • ff4ffda Merge row 16 (3fadff4) into row 17 (R70 cascade)

  • 3fadff4 Merge row 15 (494f3b8) into row 16 (R70 cascade)

  • 494f3b8 Merge row 14 (ROW 14 FINAL dbf96de) into row 15 (R70 cascade)

  • b6f8b88 Merge dev 4a26ea1 into test/cu-is-row04b-fixtures-p2

  • 7fd9449 Merge commit '4a26ea11815caf5cc983aa5b8abc4ac0a6d7670b' into test/cu-is-row4e-preflight-cleanup

  • 861feea Merge dev into 7d (take dev once, 4a26ea1)

  • bd98ab2 Merge remote-tracking branch 'origin/dev' into ci/cu-is-row6b-ledger-drift

  • db8e94e conformance fixtures: the driver owns the Electron profile directory

  • 4710785 conformance fixtures: remove the Electron profile at process exit; hide the page fully in the a11y sabotage

  • 1d0d310 conformance fixtures: hermetic web hosts, a page-scoped Windows a11y check, full trigger paths

  • 2833aad conformance fixtures: keep Mark done inside the WinForms client rect

  • 924e32c Merge the row 4b fixtures head (0fe7c2a, dev taken in) into the rects follow-up

  • f02107d fix(computer): teardown counts only the probe's Xorg and fails on a leftover probe workdir

  • cc948d4 test(computer): teardown counts only the probe's own Xorg and finds leftover probe workdirs - RED

  • 36e7c67 test(conformance): give the stdio probe test room for the worst-case Windows cleanup

  • 0fe7c2a Merge dev (2154199, carries the #13295 preflight fix) into the row 4b fixtures branch

  • 3718f50 Merge dev into 7d (takes the preflight.test.ts Windows fix, #13295)

  • 04f2758 Merge commit '21541999d72a77ca45fa7075552b4ff030ae3df7' into test/cu-is-row4e-preflight-cleanup

  • 571e571 conformance fixtures: keep the required-widget table module-private (knip)

  • c87c001 docs(changes): fragment for #13299

  • 2e987dd test(conformance): clean up the compiled stand-in engine with the shared EBUSY-retrying removeTreeSync

  • b04aaeb conformance fixtures: report screen rects for the GTK and WinForms widgets

  • c4a43fe Merge dev into 7d (104 behind)

  • b1caa8b conformance fixtures: let Chromium join AT-SPI and fix the AX probe trust call

  • 0061e53 conformance fixtures: run the fixture workflow on pull requests and dispatch only

  • 71d081f Merge dev (793cb21, #13241 landed) into the row 4b fixtures branch

  • 1ba1908 fix(computer): a failed artifact release is logged once, still in the next sweep report (gate review a6 LOW)

  • 1ea558b test(computer): the test walks fail on an entry or file they cannot read (gate review a6 LOWs)

  • 0cafb3a test(computer): v3 funnel RED - the in-process grep-zero walk fails on a file it cannot read (gate review a6 LOW)

  • 4390507 fix(computer): an artifact release that fails reaches the next sweep's warnings (planner R10)

  • 6af1081 test(computer): v3 audit GC RED - an artifact that cannot be deleted is a sweep warning, and the GC goes on (planner R10)

  • 765ed47 fix(computer): the artifact GC reports the read and delete errors it used to drop (gate review a5 LOW)

  • 04ee097 fix(computer): an audit file that cannot be read for the chain link is logged (gate review a5 LOW)

  • c959e3a test(computer): v3 audit RED - an unreadable audit file is an error, not a missing chain link (gate review a5 LOW)

  • c5f84ce fix(computer): a marker that cannot be written is logged; the refusal is unchanged (gate review a5 LOW)

  • 0a7751c test(computer): v3 funnel RED - a marker that cannot be written is logged (gate review a5 LOW)

  • 7cf35b0 fix(computer): the standalone replay plants a read-only dir as a plain file when running as root (gate review a4 LOW)

  • 45364df test(computer): v3 funnel RED - the marker-unavailable case refuses when planted as root (gate review a4 LOW)

  • 8c494f8 chore(computer): the C14 leg text says what is proved and where (gate review a4 MEDIUM)

  • 60aaf91 test(computer): the engine-process audit proof asserts its files exist, runs every fake persona, fails on unreadable entries (gate review a4)

  • 467ead2 fix(computer): v3 guards report a brace, '..' or absolute addenda path as malformed (gate review a4 LOW)

  • 0413328 test(computer): v3 guards RED - a brace, '..' or absolute addenda path is malformed (gate review a4 LOW)

  • 3a193cd test(computer): note that the zero-send clear-failure test pins the conservative a3 disposition (gate review a4 LOW)

  • 0a11138 fix(computer): the funnel logs the io error of every fail-closed step it survives (gate review a4 LOWs)

  • eb391ce test(computer): v3 funnel RED - dropped io errors of fail-closed steps are logged (gate review a4 LOWs)

  • cadf96e fix(computer): a crashed callId that cannot be poison-recorded is logged, marker kept (gate review a4 blocker)

  • e6efbf7 test(computer): v3 funnel RED - a crashed callId that cannot be poison-recorded is logged (gate review a4 blocker)

  • f8ef30f conformance fixtures: fix the AX probe compile and name what AT-SPI can see

  • 0b02470 chore(computer): C14 passed - the secure classification is proved through the engine process (R38; tests/secure_audit_engine_process.rs, 3 legs green on seoul at 1b6bb3a)

  • 1b6bb3a style(computer): bun fmt script/computer-v3-guards-addenda.test.ts

  • d4ab0d7 test(computer): the grep-zero set excludes exactly one file, the engine's fake sink by its exact path (R38, planner bound)

  • 77c335d test(computer): the engine-process grep-zero set excludes the fake keyboard sink, which must hold the text (R38)

  • 17b8bb9 conformance fixtures: the electron-linux job must install at-spi2-core

  • 126b835 feat(computer): v3 guards --scope admits exact paths from SCOPE-ADDENDA.tsv beside the plan (planner ruling R46)

  • 4abf854 test(computer): v3 guards RED - --scope reads SCOPE-ADDENDA.tsv beside the plan (planner ruling R46)

  • 731e38c test(computer): the engine-process audit proofs move to an engine integration test (R38 follow-up)

  • 4193b17 conformance fixtures: Atspi.get_desktop takes an index; only bus errors are ENVIRONMENT

  • 347dc65 conformance fixtures: the AT-SPI probe must ask for Atspi 2.0

  • 3198e9e fix(computer): the fake reports scripted Secure Input as secure_input_active (planner ruling R38, option 2)

  • 779c6fb test(computer): the engine-process grep-zero leg under Secure Input - RED (planner ruling R38, option 2)

  • bdd07e8 fix(computer): the fake reports a scripted secure focus as Secure Input, so the engine redacts it (planner ruling R38/R26)

  • 78ba665 test(computer): the engine-process secure leg runs again - RED (planner ruling on R38)

  • d437a6d test(computer): the engine-process secure leg is recorded UNOBSERVED, owed to row 26 (R38)

  • f6685be fix(computer): a marker cleanup that fails is logged; the primary error is still returned (gate review a3)

  • f7e6873 test(computer): v3 marker RED - a failed cleanup removal is reported, not dropped (gate review a3)

  • 7eacddb refactor(computer): MemoryMarkers moves to mutate/memory_markers.rs

  • f6adee6 fix(computer): the standalone replay restores narrowed dir modes on every exit (gate review a3)

  • 5881239 test(computer): v3 funnel RED - a panicking standalone replay restores its 0500 dir (gate review a3)

  • 6393536 fix(computer): any callId the wire admits can be poison-recorded, so no crash wedges a scope (gate review a3)

  • 674e7c5 test(computer): v3 funnel RED - a crashed callId that is no path component wedges its scope (gate review a3)

  • 0d69aee fix(computer): the funnel drops no audit, clear, permit or receipt result (gate review a3 B1, B2)

  • 82080fc test(computer): v3 funnel RED - no dropped audit, clear, permit or receipt result (gate review a3)

  • 17c2505 conformance fixtures: force renderer accessibility and assert page controls are visible

  • 196bb7f test(computer): the secure-field type audit is proved through the engine process - RED

  • f05d420 chore(changes): note the Windows ledger drift check

  • 4506248 Merge row 16 into row 17

  • de5a033 fix(computer): paths.rs imports Path only where it is used (unix and tests), so clippy -D warnings passes on x86_64-pc-windows-msvc

  • 4575dd8 test(computer): every fixture reports a stable app key

  • ef66042 ci(computer): fail the Windows ledger job when a verdict drifts from the committed ledger

  • 109ecc2 test(computer): Electron and web-view fixtures on every OS

  • 400cbc1 fix(computer): measure the macOS pool's teardown, up and sweep; exit 3 when nothing ran

  • 70415e5 test(computer): the committed Windows ledger must match a fresh measurement (RED)

  • a794f42 test(computer): macOS pool teardown, up, sweep and refusal must be measured (red)

  • 65e0428 test(computer): GTK fixture reads a menu path that contains spaces

  • c3459ea test(computer): fixture windows are titled Conformance Document

  • 8ff20a0 docs(computer): macOS live cells are recorded drops until a guest exists

  • 2420bac ci(computer): fail the teardown check on an unread pgrep and on a leftover probe X lock

  • e67b771 docs(computer): README for the conformance fixture apps

  • 4bb7ec8 test(computer): Electron fixture keeps the Chromium sandbox and reads control from fd 0

  • b9dfb53 chore(knip): scope the Electron fixture unlisted-import ignore to its two files

  • 3eeba49 test(computer): macOS AppKit fixture; fetch the Electron binary; read CRLF fixture lines

  • 3c2419d fix(computer): marker cases get a fresh runtime dir and plant their state

  • 20fd092 test(computer): the marker cases replay standalone - RED

  • 16844c3 Merge row 15 into row 16

  • 7b146bf Merge row 16 into row 17

  • 7dfb877 feat(computer): fresh runtime dir per replay case, planted state, and the cargo target dir

  • 20b0bfd test(computer): replay runtime dirs, planted state and the engine path - RED

  • acbbfb1 test(computer): Windows, Electron and web-view conformance fixtures; fix the GTK drag leg

  • 07c0da5 test(computer): Linux GTK conformance fixture with an engine-free self-test

  • 60efb3c Merge row 15 into row 16

  • 9a25812 Merge row 16 into row 17

  • e3a9dc4 feat(computer): replay lists and directories, and the supersedes rule

  • b26b33a test(computer): replay supersedes - RED

  • 47d369c Merge row 16 into row 17

  • 3a3a0d1 Merge row 15 into row 16

  • 568c16a style(computer): bun fmt the replay driver and its test (dev adopted bun fmt, #12225)

  • c1d2668 Merge row 16 into row 17

  • 83b0d1f Merge row 15 into row 16

  • 8e06216 Merge #13120's reviewed head 2fb19d1 into row 15

  • beaa9a8 fix(computer): SP-R labels never invalidate a reading; mutation set catches weakened rules; verdicts only from the gate

  • 7bad8ea test(computer): SP-R label fields, mutation strength and gate-only verdicts (RED, folded nits)

  • bf038c6 fix(computer): SP-R one verdict gate for every rule; numeric-FD visibility; bounded reconnect

  • c625f2d test(computer): SP-R one verdict gate, property and static checks (RED, review round 2)

  • 12cc267 Merge commit 'b7dbaf91d5835fa09978c86368b028799aabf091' into feat/cu-v3-row17-mutate-funnel

  • b7dbaf9 Merge commit 'a79bc8a1f3' into feat/cu-v3-row16-safety

  • a79bc8a fix(computer): replay defaults a missing transport to stdio only for imported oracle cases

  • 646460b test(computer): RED - a missing transport is stdio only for imported oracle cases

  • e4706ca fix(computer): row 17 conformance cases pin session.open and stopPath.start results that parse

  • 54c05d1 test(computer): SP-R L-STATE re-measured with the lsof positive control; silent-peer rows record the close code

  • 43d121e fix(computer): SP-R pair control listener cannot raise a firewall dialog; self-test isolation

  • bf91a55 fix(computer): SP-R probe failures are UNMEASURED, lsof zero needs a positive control

  • 4208919 refactor(computer): SP-W KWin registry questions in their own module

  • e18208d fix(computer): SP-W readers that see nothing are UNMEASURED; a zero-row container writes its rows UNMEASURED

  • a143477 test(computer): SP-R class self-tests for the review findings (RED)

  • 4dc7e47 test(computer): RED for SP-W readers that see nothing and for a zero-row container

  • 45ef261 test(computer): SP-W ledger re-measured with the final probe (release control legs, precondition notes)

  • 4f664cb fix(computer): SP-W answers only from readings that could come out the other way; release control leg; run-plan checks; self-test in test:computer

  • 1a37145 test(computer): RED for SP-W answer rules, release control and run plan

  • a55119b fix(computer): the poison read fails closed - an unreadable, symlinked or non-directory record refuses

  • c3da036 test(computer): v3 funnel RED - a tampered poison record still refuses the crashed callId

  • b81acab chore(changes): fragment for #13166

  • 52478d0 fix(computer): a crashed callId stays refused as delivery_uncertain/unsafe and is never resent; ledger owners

  • 4190fff test(computer): SP-R UNMEASURED reasons per the lead

  • 74fb658 test(computer): v3 funnel RED - a crashed callId is never delivered twice and reads delivery_uncertain on the wire

  • 21b9f48 test(computer): SP-W re-measures the 9 interface-presence rows

  • 423eb89 fix(computer): SP-W records the enumerated portal interface list

  • 30b48b8 fix(computer): SP-W reads portal interface presence from Introspect; 9 rows INVALID until re-measured

  • c5cdb87 test(computer): RED for SP-W portal interface enumeration

  • ea42d6c feat(computer): v3 session funnel - pre-input marker, audit, lock, permits, artifact retention

  • 0f99671 test(computer): v3 session funnel RED - marker, replay, kill, batch, audit, retention, secure field

  • 6af3339 docs(changes): changelog fragment for #13163

  • 1562920 test(computer): probe ledger for SP-W

  • 7341ffb test(computer): measure SP-R on a same-network pair and the hosted relay

  • 5aa5d5f fix(computer): v3 safety design excerpt line numbers run 381..=407

  • ab8e294 test(computer): v3 safety design excerpt refuses a renumbered copy

  • 96ea20c fix(computer): v3 safety table citations name the right design lines

  • 1dbb2eb test(computer): v3 safety table citations quote their cited design line

  • eda585c test(computer): probe ledger for SP-R

  • 38e4bd9 fix(computer): v3 safety tables carry every design entry; stop path on the wire

  • abcf56f test(computer): v3 safety tables against the design's cited list

  • d44fdc6 row 25: merge dev 04e22c0 (post bun-fmt; car sources re-formatted)

  • 406e13b feat(computer): v3 safety crate - gate order, supervisor, stop latch, protected tables

  • c406aed test(computer): v3 safety crate RED - gate order, stop path, supervisor, self set

  • aabe533 Merge branch 'feat/cu-v3-row14-fake-backend' into feat/cu-v3-row15-engine

  • 62602a3 Merge branch 'feat/cu-v3-row14-fake-backend' into feat/cu-v3-row15-engine

  • bed05e8 Merge branch 'feat/cu-v3-row14-fake-backend' into feat/cu-v3-row15-engine

  • b9f04cd Merge remote-tracking branch 'origin/dev' into feat/cu-v3-row15-engine

  • 3a9b963 Merge branch 'feat/cu-v3-row14-fake-backend' into feat/cu-v3-row15-engine

  • 8d16ee1 Merge branch 'feat/cu-v3-row13-core-contract' into feat/cu-v3-row15-engine

  • 8450d79 Merge remote-tracking branch 'origin/dev' into feat/cu-v3-row15-engine

  • 499bb23 row 25: merge dev c70f66f

  • 7de34fd fix(computer): raw-byte stdio lines, inert notifications, replay setup errors exit 2

  • 03a5aa0 feat(computer): engine refusals through the closed error table

  • b370def Merge branch 'feat/cu-v3-row14-fake-backend' into feat/cu-v3-row15-engine

  • 6f92d6f Merge branch 'feat/cu-v3-row13-core-contract' into feat/cu-v3-row15-engine

  • ff44651 Merge branch 'feat/cu-v3-row14-fake-backend' into feat/cu-v3-row15-engine

  • 149002d Merge branch 'feat/cu-v3-row13-core-contract' into feat/cu-v3-row15-engine

  • 85614a4 Merge branch 'feat/cu-v3-row12-core-protocol' into feat/cu-v3-row15-engine

  • 68e8e71 chore(computer): drop row 15's C2 deferred-leg line; row 12 re-owes it

  • 766585e feat(computer): v3 walking skeleton - engine on the fake backend and the replay driver

  • 65b0cfe test(computer): RED - replay driver and walking-skeleton engine cases

  • c02d3c4 Merge branch 'feat/cu-v3-row13-core-contract' into feat/cu-v3-row15-engine

  • 314f8ef Merge branch 'feat/cu-v3-row12-core-protocol' into feat/cu-v3-row15-engine

  • 2e9c165 Merge branch 'feat/cu-v3-row14-fake-backend' into feat/cu-v3-row15-engine

  • e8aa0e4 row 25: merge dev 6dca74b

  • 1f6d764 row 25: merge dev c40c986

  • 41443a7 fix(desktop-engine): retry a reset-before-reply only on Unix sockets (#12744 review)

  • 04b63e3 row 25: merge dev a68fc1b

  • 138b45e row 25: merge dev def0749

  • 42def63 row 25: merge dev a96a51d

  • 6953209 test(desktop-core): placeholder the conformance fixture resume tokens (match train E 18a5392)

  • 4274329 docs(changes): name the fragment for #12744

  • f4d0ce7 docs(changes): add the P3b fragment for shared-session same-user clients

Release provenance

  • v5.1.74 is published from abf50be3e7ec62019c4315acb0f087413983d583: the CI-receipted release-state head 243acc0da00a05d9480f35045cb2678ca0001925 plus #13928's operator-built darwin workflow change, which touches .github/workflows only (6 files, 0 outside). The package bytes are identical to the receipted head, and the darwin artifacts are operator-built from 243acc0da00a05d9480f35045cb2678ca0001925.
  • These workflows predate #13915: the release tag is created inside the publish-main job rather than in a separate job before any platform package is published. The dispatch was checked to have no existing tag pointing at a different sha.
bun add -g omo-ai

Don't miss a new omo-ai release

NewReleases is sending notifications on new releases.