docs: record native guest CI attempt and remaining verification gaps

README.md

@@ -173,7 +173,7 @@ The Gitea workflow runs for pull requests, pushes, and manual dispatch on the ex
 
 Ubuntu does not compile the Swift helpers or execute Apple frameworks. Tests requiring the native macOS environment report an explicit skip through `MacOsFact`; they must also be run on a supported Mac with `dotnet test MeetingAssistant.Tests/MeetingAssistant.Tests.csproj -f net10.0 -c Release -p:EnableWindowsTargeting=true`. That build compiles and signs the helpers, and the suite checks packaging, helper self-tests, and native image cropping. Real microphone/system-audio capture and privacy permissions still require the operational checks described above.
 
-[Docker-OSX](https://github.com/sickcodes/Docker-OSX) runs a macOS VM rather than providing a Wine-style compatibility layer. Its launcher supports software emulation with `KVM=accel=tcg`, so KVM is not an absolute requirement. However, its ready-made `auto` image uses Catalina; a supported .NET 10 guest needs macOS 14 or later plus the Swift build tools. No reproducible guest image or verified CI bootstrap for that environment is owned by this repository. CI validates source; it does not publish or deploy the workstation application.
+[Docker-OSX](https://github.com/sickcodes/Docker-OSX) runs a macOS VM rather than providing a Wine-style compatibility layer. Its launcher supports software emulation with `KVM=accel=tcg`, so KVM is not an absolute requirement. A supported .NET 10 guest needs macOS 14 or later plus the Swift build tools. The documented `auto` build downloads a preinstalled guest disk through `IMAGE_URL`; its documented ready-made tags and disk downloads were unavailable when checked on 2026-10-03. No verified native guest bootstrap is owned by this repository. CI validates source; it does not publish or deploy the workstation application.
 
 ## Operations And Limitations
 

docs/evidence/macos-tcg-guest-feasibility.md

@@ -0,0 +1,44 @@
+# Native macOS guest on existing Docker infrastructure — 2026-10-03
+
+Result: real macOS Recovery booted under software emulation, but the bounded experiment did not reach an installed guest or execute the native tests. This is not a passing CI result or a verified workflow recipe.
+
+## Source and execution boundary
+
+- Dockur source: [`16a5b470cdd601bae8b05b02d748d7edfb36c12e`](https://github.com/dockur/macos/tree/16a5b470cdd601bae8b05b02d748d7edfb36c12e).
+- Local environment: existing ARM64 Docker Desktop, linux/amd64 container translation, x86_64 QEMU guest with TCG. This does not establish the runtime or resources of the upstream Ubuntu-x64 runner.
+- Guest settings: `VERSION=14`, `MANUAL=N`, `KVM=N`, `NETWORK=slirp`, 4 GiB guest RAM, two vCPUs, 64 GiB sparse guest disk. Container memory was limited to 6 GiB.
+- Container inspection: `Privileged=false`, `CapAdd=null`, `Devices=[]`, ordinary bridge network. Diagnostic ports were published only on localhost. No runner registration, host configuration, device passthrough or new secret was requested.
+- Candidate archive: clean application/test source at `1164c26846686c1912fd1816cd06de80352e9504`, SHA-256 `30796a27c75792ab87d23dd4c4db991b426d9c55a0e7dc3d4dc45c41056846e1`. It was not executed in the guest.
+
+## Observed boundary failure
+
+The container built and downloaded Apple's Recovery image. Recovery mounted its installation-state share, found Sonoma's `startosinstall`, and passed the upstream unattended preflight. Native `sw_vers` and `uname` output identified:
+
+```text
+ProductVersion: 14.6.1
+BuildVersion: 23G93
+Darwin Kernel Version 23.6.0 ... RELEASE_X86_64 x86_64
+```
+
+The first target-selection attempt found no writable disk. Both virtio and SATA trials subsequently exposed `diskutil` failure: it could not use the DiskManagement framework. Its diagnostic listed unavailable DiskArbitration/single-user mode as a possible cause; that cause was not independently established. SATA also exposed an independent missing `io2` QEMU object, corrected through an ordinary QEMU argument without host capabilities.
+
+The pinned Recovery patch replaces an early `rc.cdrom.sh` block, before normal daemon startup, with a blocking `exec launch.sh`. One final causal trial changed only this padded bootstrap to launch the installer in the background. Its old/new bootstrap SHA-256 values were `73fd171ea4c889f9946f04d6866047fedd417a69a241b3e84db469c13975d897` and `c5fc86f95e9c65dd7dc59536911c81c96928d4f9541b63121c22cad5690f3582`. The launch script also waited for actual `diskutil list physical` success before the unchanged writable-disk/size guards and captured diagnostics after mounting the share.
+
+That trial still recorded 12 completed native `diskutil` failures and an incomplete thirteenth attempt:
+
+```text
+Unable to run because unable to use the DiskManagement framework.
+Common reasons include, but are not limited to, the DiskArbitration
+framework being unavailable due to being booted in single-user mode.
+[exit 1]
+```
+
+There was no successful disk enumeration, selected installation target, target erase or `startosinstall` invocation. The background change did not establish a working bootstrap; the remaining daemon/framework cause is unresolved. This result does not prove that macOS under TCG on Ubuntu is impossible.
+
+## End state and evidence
+
+Final container: `c4d5b83f5199063df75d43236d610ef9b71c3b39c14b80a649b0e904d13e87c8`. It started at `10:46:24.987543461 UTC`; the authorized final boot boundary was start plus 16 minutes. It was stopped at `11:02:33.810090465 UTC`: `Running=false`, `ExitCode=143`, `OOMKilled=false`. No experiment container remained running.
+
+Raw evidence is retained locally under `/private/tmp/meeting-assistant-macos-guest-proof-20261003/runs/c4d5b83f5199/`: `guest-disk-platform.log`, `physical-disks-ready.log`, `recovery-install.log`, their capture metadata, and `end-container-inspect.json`/`end-state.json`. Metadata identifies the full container ID, actual start time, capture time and log SHA-256; earlier flat logs are historical and were not treated as success in this run.
+
+No guest SSH session, official .NET/Apple CLT bootstrap, Swift build or TRX result was obtained. The application's workflow, production runtime and tests were not changed by this experiment. Native Ubuntu CI remains open; local Mac qualification is recorded separately in [the CI evidence](macos-ubuntu-ci-completion.md).

docs/evidence/macos-ubuntu-ci-completion.md

@@ -5,10 +5,10 @@
 Source repository: `/Users/dh/Documents/DanielsVault/_ops/meeting-assistant`.
 Delivery branch: `codex/macos-support` in `Daniel/meeting-assistant`; intended PR target: `Manuel/meeting-assistant:main`, PR #39.
 Fixed change baseline: `bd35ebc4c81bedf80caabb82481169124775fb36` (already contains main `7b2bcd36310ae6434a4ee6eeb3a47b6d1d431df4`).
-Work was isolated under `/private/tmp/meeting-assistant-macos-ci-20261003`; the original checkout was clean and no workstation service was restarted.
+Implementation and qualification were isolated under `/private/tmp/meeting-assistant-macos-ci-20261003`; the original checkout was clean. Updating that original checkout after publication unexpectedly invoked its existing post-merge deployment hook and restarted the workstation service. The immediately checked recording endpoint reported idle (`state=0`, `isRecording=false`) and health reported OK. Future checkout updates disable Git hooks explicitly. This qualification did not intentionally start a new recording.
 Review covers this CI/test/shutdown delta. Earlier macOS feature and main-sync evidence remains in the respective OpenSpec implementation receipts; this record is not a fresh review of all earlier branch changes.
 
-The evidence record and generated logs/TRX files are excluded from the substantive content identity to avoid self-referential hashes.
+The hashes below identify the substantive implementation published in `1164c26846686c1912fd1816cd06de80352e9504`. The evidence record and generated logs/TRX files are excluded to avoid self-referential hashes. Subsequent factual documentation corrections do not change the reviewed workflow, runtime or tests.
 
 | Reviewed file | SHA-256 |
 | --- | --- |
@@ -45,9 +45,11 @@ The passes ran sequentially in fresh, independent reviewer contexts against the
 
 No review repair changed the frozen substantive identity. All required structural passes are complete; runtime evidence and remaining Windows/native limitations are recorded separately above.
 
+The subsequent README/evidence corrections update only observed artifact availability, publication status, and the checkout-hook event. Structural review is `not-required` for this factual documentation delta; the workflow, runtime, tests and binding instructions are unchanged. A separate read-only factual check confirmed the revised `auto` description against Dockerfile.auto and checked the limits against the saved receipts. Whitespace and the final documentation diff were checked; existing test results remain applicable to the unchanged code.
+
 ## Delivery and remaining limits
 
 State at publication: structural review complete; current-head remote verification pending. This is not a merge-readiness or main-merge receipt.
-The user authorized preparing this branch for PR CI and merging into main after tests pass. Authenticated Gitea identity `Daniel` has write/admin on the fork and read-only access to `Manuel/meeting-assistant`; it cannot merge there. The latest published PR head at intake was `bf6e1e7560c6bdfcebf69ad0e3240b4de647adb6`, so the earlier local main-sync commits also need publishing.
+The user authorized preparing this branch for PR CI and merging into main after tests pass. Authenticated Gitea identity `Daniel` has write/admin on the fork and read-only access to `Manuel/meeting-assistant`; it cannot merge there. The earlier local main-sync commits and reviewed implementation were normally pushed to the existing fork branch as `1164c26846686c1912fd1816cd06de80352e9504`; PR #39's description was updated and read back against that exact head.
-Authorized delivery is a normal push of this reviewed delta plus the earlier local main-sync commits to the existing fork branch, followed by a PR description update. The published commit, actual workflow status, and any remaining owner action are reported on PR #39. Main must not be reported as merged before the required remote checks pass and the repository owner performs the permitted merge.
+Upstream [run 4145](https://gitea.schweigert.cloud/Manuel/meeting-assistant/actions/runs/4145) was created for this head, with both Ubuntu jobs waiting and no runner assigned. Its live job view explicitly says `Need approval to run workflows for fork pull request.` The current account cannot approve upstream workflows. This is the existing fork-workflow approval gate, not a request for additional infrastructure. Main is not merged; remote Windows and portable checks have not executed for this head.
-A native macOS guest/Swift CI run on Ubuntu is not implemented; Docker-OSX supports software emulation, but its ready-made Catalina guest does not provide a supported .NET 10 + modern Apple-framework build environment. No unverified guest bootstrap was made a required PR check.
+A native macOS guest/Swift CI run on Ubuntu is not implemented. Docker-OSX supports software emulation, but the documented installed-guest downloads and `auto` tags were unavailable when checked. An [isolated unprivileged Dockur/TCG experiment](macos-tcg-guest-feasibility.md) booted an actual macOS 14.6.1 x86_64 Recovery environment, but `diskutil` could not use the DiskManagement framework; the actual cause remains unresolved. The bounded trial ended without an installed guest or native tests, and its container was stopped. No unverified guest bootstrap was made a required PR check. Passing the two current Ubuntu jobs would still leave the user's requested native macOS coverage in Ubuntu CI unmet; the local Mac result does not close that requirement.