Merge branch 'main' into docs/freeloader-native-accuracy

Fail Frame smoke when a launch is skipped

app/updater.js

@@ -181,7 +181,6 @@ old="$stage/old.app"
 if mv "$app" "$old"; then
   if mv "$new" "$app"; then rm -rf "$old"; else mv "$old" "$app"; fi
 fi
-xattr -dr com.apple.quarantine "$app" 2>/dev/null
 rm -rf "$stage"
 open "$app"
 `;

docs/current-support-matrix.md

@@ -1,11 +1,11 @@
 # What we have tested
 
-Current published release: Frame Control **v0.4.52**, 28 October 2026. Rows identify the last direct check of each feature; an older row does not mean the latest installer was retested on that hardware. “Real Frame + physical PC” means an actual headset and a non-virtual computer. “VM only” means a Windows virtual machine, sometimes connected to the real Frame. A CI test or a simulated device is **not** a physical hardware test. See each link for the exact procedure and limits.
+Current published release: Frame Control **v0.4.53**, 28 October 2026. Rows identify the last direct check of each feature; an older row does not mean the latest installer was retested on that hardware. “Real Frame + physical PC” means an actual headset and a non-virtual computer. “VM only” means a Windows virtual machine, sometimes connected to the real Frame. A CI test or a simulated device is **not** a physical hardware test. See each link for the exact procedure and limits.
 
 | Feature | Real Frame + physical PC tested | VM only | Untested or limited | Last check and proof |
 |---|---|---|---|---|
 | Mac desktop release — tested on a real Mac | Physical MacBook Air (MacBookAir9,1), Intel Core i3, macOS 15.7.7 (24G720): official v0.4.39 x64 DMG and ZIP hashes/architectures verified; clean DMG app installed and launched. Native UI, preferences across restart, update discovery of 0.4.40, offline OBS Start/Stop and 116-entry Decky store checked. Actual Open at Login menu toggled on, persisted across Quit/relaunch, then restored off. A 54-byte relay upload/readback/delete passed under coordinator/headset locks | No | Ordinary browser quarantine, right-click Open and Open Anyway remain unverified; the synthetic quarantine probe was inconclusive. Actual login startup untested. Direct frame.local had no route and app SSH trust failed; relay transfer is not native Mac transfer acceptance. Real OBS/Framey VR unverified. Earlier shutdown race fix #284 shipped in v0.4.40, whose Intel installer was not tested here | 26 Oct, official v0.4.39 tag `fd0b0a9`: [repeat Intel acceptance evidence](evidence/intel-mac-v0.4.39-2026-10-26.md); [earlier release test](evidence/mac-v0439-2026-10-26.md) |
-| Core title smoke test | A Linux host ran the v0.4.52 versioned source candidate, whose tree matches the tagged release commit, against the real Frame under a unique lock: 12 steps passed, one optional x86_64 launch was skipped, zero failed; a separate locked cleanup audit passed | No | The headset was in standby and unworn. Steam Linux Runtime 4 was absent, so the x86_64 test executable did not launch. Worn rendering and physical-controller input were not checked | 28 Oct, v0.4.52 tag `v0.4.52`, Frame OS 0.5.6 build `20261022.6153909`: [release verification](evidence/v0452-release-2026-10-28.md), [candidate smoke](evidence/v0452-candidate-2026-10-28.md) |
+| Core title smoke test | A Linux host ran the v0.4.53 versioned source candidate against the real Frame under a unique lock; only documentation changed before the tagged release, leaving its executable source the same: 12 steps passed, one optional x86_64 launch was skipped, zero failed; a separate locked cleanup audit passed | No | The headset was in standby and unworn. Steam Linux Runtime 4 was absent, so the x86_64 test executable did not launch. Worn rendering and physical-controller input were not checked | 28 Oct, v0.4.53 tag `v0.4.53`, Frame OS 0.5.6 build `20261022.6153909`: [release verification](evidence/v0453-release-2026-10-28.md), [candidate smoke](evidence/v0453-candidate-2026-10-28.md) |
 | Windows desktop in the headset | No physical Windows PC test | Windows 11 VM sent a whole desktop to a real Frame; this used an intermediate fix, not the final native DLL | Physical GPU, capture and controller input; individual windows unavailable in bundled DXGI build | 24 Oct, v0.4.10 installer + candidate fix: [VM log](evidence/windows-vm-2026-10-24.md), [PC guide](pc-in-headset.md) |
 | Linux desktop in the headset | No | No | Native Linux PC capture is disabled; lxso2 lacked a desktop portal and PipeWire session. The separate Steam `-pipewire` helper is not Frame Control's PC capture | 24 Oct, v0.4.17: [host audit](evidence/linux-pc-host-lxso2-2026-10-24.md), [Steam helper PR #163](https://github.com/saphid/frame-control/pull/163) |
 | Mac desktop in the headset | A physical Mac sent its screen/window to a real Frame | No | Current v0.4.26 Mac installer and worn-headset controller input have not been repeated | 28 Sep, pre-v0.4 / Frame OS 0.4.1: [Mac guide and measurement](streaming.md#b-show-the-macs-desktop-inside-the-frame) |
@@ -24,7 +24,7 @@ Current published release: Frame Control **v0.4.52**, 28 October 2026. Rows iden
 | Open at login | No Frame needed; official v0.4.39 on the physical Intel Air toggled the actual native menu on, retained its checkmark after Quit/relaunch and restored it off. Linux autostart file previously toggled on a physical Linux host with an isolated config directory | No | A full OS login session and physical Windows host login remain untested | 26 Oct, v0.4.39: [native Intel toggle evidence](evidence/intel-mac-v0.4.39-2026-10-26.md); Linux v0.4.26: [PR #189](https://github.com/saphid/frame-control/pull/189) |
 | Local Frame Control extensions | Example panel displayed live Frame status in a browser preview; the real Frame served a read-only example in the experimental headset panel | No | No wearer-visible extension view; third-party extensions and physical Windows/macOS extension folders untested | 25 Oct, v0.4.29: [PR #194](https://github.com/saphid/frame-control/pull/194), [PR #201](https://github.com/saphid/frame-control/pull/201), [extension guide](extensions.md) |
 | Experimental Frame Control SteamVR panel | Frame OS 0.5.6 registered and served the opt-in overlay. A later unattended check safely woke SteamVR and found the panel in stereo compositor and mirror captures; it removed its temporary service and restored power settings | No | No person wore the headset for that check. Physical controller input, scrolling, VR keyboard, comfort and legibility remain unverified. The large status view has browser and packaging checks, not a wearer review | 26 Oct, v0.4.29–0.4.40: [panel PR #201](https://github.com/saphid/frame-control/pull/201), [unattended evidence](evidence/framey-vr-2026-10-25/README.md), [post-#292 unworn sample exercise](evidence/v0440-release-2026-10-26.md#post-tag-292-panel-regression), [earlier standby-gated exercise](evidence/framey-vr-standby-gate-2026-10-26.md), [worn checklist](framey-vr-check.md) |
-| Experimental Decky Loader manager | Earlier real Frame checks passed opt-in ARM64 user-service install/status/update/remove and managed ProtonDB install/update/remove. The current exact-version catalog records six scoped works: Notebook default drawing, Volume Mixer and Mixer silent per-app volume, Volume Boost muted sink controls, DeckySpy metrics, and Simple Timer silent in-panel timers | No | v0.4.52 classifies 116 exact-version plugins as **6 works / 54 partial / 45 broken / 11 N/A**. “Works” applies only to each documented behavior, never worn VR. Network Sentinel and Battery Tracker are partial; their managed real-Frame ARM64 behavior is unverified. Most other backends, audible output, physical-controller use and wearer-visible official panels remain unverified | 28 Oct, v0.4.52: [exact-version table](framey-compatibility.md), [release verification](evidence/v0452-release-2026-10-28.md), [DeckySpy evidence](evidence/framey-deckyspy-2026-10-26.md), [Simple Timer evidence](evidence/framey-simple-timer-2026-10-26.md), [audio evidence](evidence/framey-audio-arm64-2026-10-26.md) |
-| Experimental Framey VR plugin panels and store | An unattended Frame OS 0.5.6 exercise showed the repository sample panel in stereo compositor and mirror captures. Synthetic OpenVR clicks changed a live sample setting, read its Decky backend file, and confirmed it after a fresh iframe reopen; temporary services were removed and original power settings restored | No | This was the repository sample, not an official store plugin. Physical controllers, native scrolling, VR keyboard, wearer comfort and legibility remain unverified. Fake-Frame install/remove and local Chromium plugin initialization do not prove an official plugin is visible or usable in worn VR | 28 Oct, v0.4.52 catalog with earlier 25–26 Oct VR evidence: [unattended sample evidence](evidence/framey-vr-2026-10-25/README.md), [compatibility table](framey-compatibility.md), [worn checklist](framey-vr-check.md) |
+| Experimental Decky Loader manager | Earlier real Frame checks passed opt-in ARM64 user-service install/status/update/remove and managed ProtonDB install/update/remove. The current exact-version catalog records six scoped works: Notebook default drawing, Volume Mixer and Mixer silent per-app volume, Volume Boost muted sink controls, DeckySpy metrics, and Simple Timer silent in-panel timers | No | v0.4.53 classifies 116 exact-version plugins as **6 works / 54 partial / 45 broken / 11 N/A**. “Works” applies only to each documented behavior, never worn VR. Network Sentinel and Battery Tracker are partial; their managed real-Frame ARM64 behavior is unverified. Most other backends, audible output, physical-controller use and wearer-visible official panels remain unverified | 28 Oct, v0.4.53: [exact-version table](framey-compatibility.md), [release verification](evidence/v0453-release-2026-10-28.md), [DeckySpy evidence](evidence/framey-deckyspy-2026-10-26.md), [Simple Timer evidence](evidence/framey-simple-timer-2026-10-26.md), [audio evidence](evidence/framey-audio-arm64-2026-10-26.md) |
+| Experimental Framey VR plugin panels and store | An unattended Frame OS 0.5.6 exercise showed the repository sample panel in stereo compositor and mirror captures. Synthetic OpenVR clicks changed a live sample setting, read its Decky backend file, and confirmed it after a fresh iframe reopen; temporary services were removed and original power settings restored | No | This was the repository sample, not an official store plugin. Physical controllers, native scrolling, VR keyboard, wearer comfort and legibility remain unverified. Fake-Frame install/remove and local Chromium plugin initialization do not prove an official plugin is visible or usable in worn VR | 28 Oct, v0.4.53 catalog with earlier 25–26 Oct VR evidence: [unattended sample evidence](evidence/framey-vr-2026-10-25/README.md), [compatibility table](framey-compatibility.md), [worn checklist](framey-vr-check.md) |
 
 Release notes must preserve these limits until a newer recorded check changes them. When a feature is retested, add its date, exact app/Frame version, host type and proof here. The older [v0.3.1 matrix](support-matrix.html) is a historical snapshot and does not describe current coverage.

docs/evidence/framey-free-loader-native-2026-10-28.md

@@ -0,0 +1,25 @@
+# Free Loader 1.6.0: private Decky backend check
+
+The official store archive for ID 63 was downloaded and matched SHA-256
+`05208aee6f6ba824391d8ddc81b1168b2599a9711e7528cdaa789f6cc710b42a`.
+The published Decky 3.3.0 amd64 loader matched SHA-256
+`e80d03e641d1766a33a817cb27ab8d2e0a4633685e5d27e4de948a6687526181`.
+
+An unprivileged, disposable Debian container ran that loader and the exact
+official plugin with no network, host devices, host session or writable host
+mount. Decky discovered Free Loader as enabled. Its API 1 methods returned 14
+default settings and an empty local deals map. A private `larger_text` setting
+changed to `true` and read back; startup created only the container's private
+settings file, and `read_deals` created an empty private deals cache. No feed,
+notification, Steam store or external account method was called. The local
+probe exited successfully; its container was removed. The committed native
+Decky runtime case repeats these calls and also checks persistence after a
+loader restart on the CI runner's native architecture. The local x86_64 run
+passed all 14 runtime cases, including Free Loader and the existing negative
+regressions. The runner removed its temporary containers and image.
+
+This strengthens the **partial** label. It does not verify a live feed refresh,
+notification delivery, real Frame ARM64 service, rendered deal cards, worn VR
+or controller interaction. The earlier opaque Chromium check had denied backend
+and network calls, so it could not establish that deal cards render with live
+data. No headset was touched for this check.

docs/evidence/v0454-candidate-2026-10-28.md

@@ -0,0 +1,31 @@
+# v0.4.54 candidate hold, 28 October 2026
+
+The local, unpushed `release/v0.4.54` candidate is commit
+`3eefedd8d4dd2ce8db7d62d0dd2d7cff1fe2089b`, based on main
+`e7b037ce883e45990bf239aee29b05434a39a399`. It changes only the
+version in `app/package.json` and `app/package-lock.json`. It has not been
+tagged or published. The latest public release remains
+[v0.4.53](https://github.com/saphid/frame-control/releases/tag/v0.4.53).
+
+The locked real Frame smoke at 20:28 UTC read SteamVR activity 3 (standby),
+proximity 2.3125 and battery 100% on Frame OS build `20261022.6153909`.
+ARM64 and Windows/Proton test titles launched. The x86_64 test title did
+**not** launch: Steam reported that Steam Linux Runtime 4.0 was found but
+not installed. The result has **12 passed, 1 skipped, 0 failed**. Its final
+cleanup step passed and its unique lock was released. The private JSON result
+is `/tmp/frame-v0454-smoke-results/2026-10-28T202826Z-20261022.6153909.json`
+with SHA-256 `d86edaec0b86faf130c21aa3d7113cc1ac5b2ae0869dbd15f811bf3bb5b76b88`.
+
+A separate read-only audit acquired and released a fresh unique headset lock.
+It read activity 3, proximity 2.3125 and battery 100%, and found no reserved
+smoke title directories or owned test processes. It did not wake the Frame or
+send input. Nobody confirmed wearing the Frame, so no wearer-visible or
+physical-controller result follows.
+
+**Release decision: hold v0.4.54.** The smoke script at the candidate commit
+returned exit 0 even though it recorded a skipped launch. Under the current
+release rule, a skip cannot satisfy the smoke gate. The accompanying change
+makes a skipped smoke return nonzero while preserving the explicit skipped
+step and cleanup record. Exact release-commit CI, release verification, tagged
+installer builds, checksums and publisher dry run have not been performed for
+this local candidate.

docs/framey-compatibility.md

@@ -25,7 +25,7 @@ python3 scripts/framey-audit-browser.py --cache /tmp/framey-archives
 
 Open the browser audit URL in local Chromium, run `runAudit()` and inspect/download `results`. All backend/network actions remain denied.
 
-A [Free Loader duplicate-entry follow-up](evidence/framey-free-loader-duplicates-2026-10-26.md) accepts only byte-identical duplicate ZIP members whose modes differ at most in execute bits. The exact archive passed private local and focused fake-Frame install/remove; its frontend rendered in opaque Chromium and isolated Python settings/empty-deals reads passed with a Decky stub. The earlier full 116-plugin fake-Frame audit blocked this archive before the fix. Actual Decky, feed updates and worn VR remain unverified.
+A [Free Loader duplicate-entry follow-up](evidence/framey-free-loader-duplicates-2026-10-26.md) accepts only byte-identical duplicate ZIP members whose modes differ at most in execute bits. The exact archive passed private local and focused fake-Frame install/remove; its frontend rendered in opaque Chromium. The earlier full 116-plugin fake-Frame audit blocked this archive before the fix. A later [native Decky 3.3.0 check](evidence/framey-free-loader-native-2026-10-28.md) verified plugin discovery, default settings, empty local deals and private setting persistence after restart in a network-isolated amd64 container. Live feed updates, notifications, Steam store integration, real Frame ARM64 service and worn VR remain unverified.
 
 A [2026-10-28 static action-limit review](evidence/framey-static-action-limits-2026-10-28.md) hash-checked the exact AutoSuspend and CheatDeck archives. AutoSuspend contains Steam suspend/shutdown paths, and CheatDeck's advertised trainer flow needs a Steam game context menu and launch-option writes. Neither plugin was executed; both retain their broken labels.
 
@@ -81,7 +81,7 @@ The API column shows manifest / actual bundled frontend API. Many manifests say
 | 60: DeckyFileServer 1.2.0 | 1 / 2 | partial | Managed install supplies a native ARM64 helper and removes the exact archive's unnecessary root request. Real Decky 3.3.0 on Frame passes private read-only HTTPS listing/download and upload denial. Frontend address/picker limitations remain; use HTTPS and a preconfigured directory. Uploads, generated thumbnails and worn VR are unverified. Shared Toggle value rendering is repaired; other listed blockers remain. | static archive, local browser, fake-Frame, real Decky 3.3.0 amd64, real Frame SSH |
 | 61: TabMaster 2.15.1 | 1 / 2 | partial | Local Chromium rendered the QAM and navigable Overview/Filters documentation with an actual Decky 3.3.0 amd64 backend. Library-tab editing remains unavailable: Steam login/library patches and MobX reactions have no target. ARM64 and worn VR unverified. | static archive, local browser, fake-Frame |
 | 62: TS3 QuickAccess 1.2.0 | legacy / legacy | broken | Plugin contains a non-ARM64 Linux binary: ts3-qs4sd/bin/ts3-qs4sd.so. | static archive, local browser, fake-Frame |
-| 63: Free Loader 1.6.0 | 1 / 2 | partial | The exact archive's duplicate names contain identical bytes and differ only in execute bits; the guarded installer now accepts them, and a private local install/remove transaction passed. The current opaque Chromium host renders Free Games but cannot load deals with backend/network denied. Exact official Python settings and empty local-deals methods passed in a no-network container with a Decky stub. Real Decky, feed refresh, notifications and worn VR remain unverified. | static archive, local browser, fake-Frame pinned install/remove, local exact-archive transaction, isolated Python backend |
+| 63: Free Loader 1.6.0 | 1 / 2 | partial | The exact archive's byte-identical duplicate ZIP members pass guarded install/remove. Actual Decky 3.3.0 in a private network-isolated container discovers the plugin, reads default settings and empty local deals, and retains a settings change after restart. Live deal-feed refresh, notifications, Steam store integration, real Frame ARM64 service and worn VR remain unverified; no feed request was made. | static archive, local browser, fake-Frame pinned install/remove, local exact-archive transaction, isolated Python backend, isolated Linux backend |
 | 64: Syncthing 0.2.7 | legacy / legacy | broken | Plugin contains a non-ARM64 Linux binary: decky-syncthing/bin/decky-syncthing-watchdog. Shared Toggle value rendering is repaired; other listed blockers remain. | static archive, local browser, fake-Frame |
 | 65: Decky Terminal 0.5.3 | 1 / 2 | partial | Local routes initialize after parameter-route support; no shell commands were run. | static archive, local browser, fake-Frame |
 | 66: PlayTime 3.2.5 | 1 / 2 | partial | Statistics and routed settings render with a detached real Frame library/user snapshot and actual Decky 3.3.0 amd64 backend in isolation. User selection and empty statistics reads pass. Running snapshots are now exposed, but its private MobX observation is unsupported; automatic recording and ARM64 backend remain unverified. | static archive, local browser, fake-Frame, real Frame SSH |

docs/releasing.md

@@ -72,7 +72,9 @@ count. So a build reaches people only when you publish it, after testing it.
    that the uploaded `SHA256SUMS` digest matches a fresh list of those ten
    installer digests. It attaches
    `update.json` (the version, the notes, the release page and each
-   installer's digest), then publishes the release and marks it latest. It
+   installer's digest), verifies the uploaded asset's SHA-256 and downloads
+   its bytes for an exact readback before publishing the release and marking
+   it latest. It
    refuses an already published release before replacing its live update
    manifest; use a new tested release for changes after publication. It
    uses only the REST API: `gh release view` can't find a draft whose
@@ -84,6 +86,12 @@ count. So a build reaches people only when you publish it, after testing it.
    seconds after starting, then every 6 hours, and anyone can use **Check for
    Updates…** (the app menu on macOS, the Help menu elsewhere).
 
+   After publication, rerun `audit-release-assets.py` with a new report path.
+   For a published release it also downloads `update.json`, verifies its
+   GitHub SHA-256, and checks that the offered version, release page, notes,
+   installer names, sizes, and digests match that release. This checks the
+   updater channel's published bytes, which the draft audit cannot inspect.
+
 To pull a bad release, mark the previous one as latest
 (`gh release edit v0.3.9 --latest`) or turn the bad one back into a draft.
 Copies that already updated stay on it. Nothing downgrades them.

docs/testing.md

@@ -233,8 +233,9 @@ before (if it was, it stays, synced to this checkout as Frame Control always
 does). A cleanup that fails counts as a failed step. Results go to
 `tests/smoke/results/<time>-<BUILD_ID>.json` (not committed) with a summary on
 screen. The summary counts passed, skipped and failed steps separately. It
-exits 0 when no step failed (including a skipped optional runtime), 1 if a step
-failed, or 2 if the headset isn't reachable.
+exits 0 only when every step passed, 1 if a step failed or was skipped, or 2 if
+the headset isn't reachable. A skipped x86-64 launch blocks a release smoke
+gate even though it remains separately labeled in the result.
 
 Smoke process cleanup runs before and after title removal and during the
 initial/final leftover sweeps. It targets only the reserved `fc_smoke_arm64`,

scripts/audit-release-assets.py

@@ -111,6 +111,38 @@ def download_asset(asset, path):
     return digest
 
 
+def validate_update_manifest(path, release):
+    """Check the bytes running copies read against the published release."""
+    try:
+        manifest = json.loads(path.read_text(encoding="utf-8"))
+    except (UnicodeError, json.JSONDecodeError) as error:
+        raise AuditError(f"update.json is not valid UTF-8 JSON: {error}") from error
+    if not isinstance(manifest, dict):
+        raise AuditError("update.json is not an object")
+    tag = release.get("tag_name")
+    if not isinstance(tag, str) or not re.fullmatch(r"v\d+\.\d+\.\d+", tag):
+        raise AuditError("published release has an invalid tag")
+    if manifest.get("version") != tag[1:]:
+        raise AuditError("update.json version differs from the release tag")
+    page = f"https://github.com/{REPO}/releases/tag/{tag}"
+    if manifest.get("page") != page:
+        raise AuditError("update.json release page differs from the tag page")
+    if manifest.get("notes") != (release.get("body") or "")[:4000]:
+        raise AuditError("update.json notes differ from the release notes")
+    expected = {asset["name"]: {key: asset[key] for key in ("name", "size", "digest")}
+                for asset in release["assets"] if asset["name"] in INSTALLERS}
+    assets = manifest.get("assets")
+    if not isinstance(assets, list) or len(assets) != len(expected):
+        raise AuditError("update.json has an incomplete installer list")
+    names = [asset.get("name") for asset in assets if isinstance(asset, dict)]
+    if (len(names) != len(assets) or not all(isinstance(name, str) for name in names)
+            or len(names) != len(set(names))):
+        raise AuditError("update.json has invalid or duplicate installer entries")
+    if {asset["name"]: asset for asset in assets} != expected:
+        raise AuditError("update.json installers differ from the release assets")
+    return {"version": manifest["version"], "installers": len(assets)}
+
+
 def inspect_package(path, name, version):
     if name.endswith(".zip"):
         with zipfile.ZipFile(path) as archive:
@@ -190,6 +222,12 @@ def audit(tag, report_path, launch_x86_appimage=False):
         sums_hash = download_asset(by_name["SHA256SUMS"], sums_path)
         if sums_path.read_bytes() != expected_sums:
             raise AuditError("SHA256SUMS bytes differ from the ten GitHub asset digests")
+        manifest_detail = None
+        if not release.get("draft"):
+            manifest_path = scratch / "update.json"
+            manifest_digest = download_asset(by_name["update.json"], manifest_path)
+            manifest_detail = validate_update_manifest(manifest_path, release)
+            manifest_detail["sha256"] = manifest_digest
         rows = []
         for name in INSTALLERS:
             path = scratch / name
@@ -210,7 +248,8 @@ def audit(tag, report_path, launch_x86_appimage=False):
             print(f"verified {name}", flush=True)
     report = {"tag": tag, "releaseId": release["id"],
               "tagNameAtAudit": release.get("tag_name"), "draftAtAudit": release.get("draft"),
-              "sha256sumsSha256": sums_hash, "assets": rows}
+              "sha256sumsSha256": sums_hash, "updateManifest": manifest_detail,
+              "assets": rows}
     with report_path.open("x", encoding="utf-8") as output:
         json.dump(report, output, indent=2)
         output.write("\n")

scripts/framey-compatibility-doc.py

@@ -40,7 +40,7 @@ def render(data):
         'python3 scripts/framey-audit-browser.py --cache /tmp/framey-archives',
         '```', '',
         'Open the browser audit URL in local Chromium, run `runAudit()` and inspect/download `results`. All backend/network actions remain denied.', '',
-        'A [Free Loader duplicate-entry follow-up](evidence/framey-free-loader-duplicates-2026-10-26.md) accepts only byte-identical duplicate ZIP members whose modes differ at most in execute bits. The exact archive passed private local and focused fake-Frame install/remove; its frontend rendered in opaque Chromium and isolated Python settings/empty-deals reads passed with a Decky stub. The earlier full 116-plugin fake-Frame audit blocked this archive before the fix. Actual Decky, feed updates and worn VR remain unverified.', '',
+        'A [Free Loader duplicate-entry follow-up](evidence/framey-free-loader-duplicates-2026-10-26.md) accepts only byte-identical duplicate ZIP members whose modes differ at most in execute bits. The exact archive passed private local and focused fake-Frame install/remove; its frontend rendered in opaque Chromium. The earlier full 116-plugin fake-Frame audit blocked this archive before the fix. A later [native Decky 3.3.0 check](evidence/framey-free-loader-native-2026-10-28.md) verified plugin discovery, default settings, empty local deals and private setting persistence after restart in a network-isolated amd64 container. Live feed updates, notifications, Steam store integration, real Frame ARM64 service and worn VR remain unverified.', '',
         'A [2026-10-28 static action-limit review](evidence/framey-static-action-limits-2026-10-28.md) hash-checked the exact AutoSuspend and CheatDeck archives. AutoSuspend contains Steam suspend/shutdown paths, and CheatDeck\'s advertised trainer flow needs a Steam game context menu and launch-option writes. Neither plugin was executed; both retain their broken labels.', '',
         'A [Network Sentinel user-service check](evidence/framey-network-sentinel-2026-10-28.md) adapts only the exact 1.0.4 digest. In isolated actual Decky 3.3.0 amd64, the pinned plugin-local psutil wheel, private settings RPC and one monitoring-loop sample pass. External ping/DNS, Frame ARM64, panel controls and worn VR remain unverified; this is partial compatibility.', '',
         'A [Battery Tracker user-service check](evidence/framey-battery-tracker-2026-10-28.md) adapts only the exact 0.2.0 digest. A locked read-only Frame probe confirmed a Battery-type supply with the required fields; actual Decky 3.3.0 amd64 recorded private Frame-shaped samples and returned history/per-app power. The managed plugin has not run on the real Frame, and panel controls and worn VR remain unverified; this is partial compatibility.', '',

scripts/publish-release.sh

@@ -139,7 +139,23 @@ for asset in $old; do
   gh api -X DELETE "repos/$repo/releases/assets/$asset" >/dev/null
 done
 gh api -X POST "https://uploads.github.com/repos/$repo/releases/$id/assets?name=update.json" \
-  -H "Content-Type: application/json" --input "$tmp/update.json" >/dev/null
+  -H "Content-Type: application/json" --input "$tmp/update.json" > "$tmp/uploaded-manifest.json"
+asset_id=$(python3 - "$tmp/uploaded-manifest.json" "$tmp/update.json" <<'EOF'
+import hashlib, json, pathlib, sys
+asset = json.load(open(sys.argv[1], encoding="utf-8"))
+body = pathlib.Path(sys.argv[2]).read_bytes()
+digest = "sha256:" + hashlib.sha256(body).hexdigest()
+if (asset.get("name") != "update.json" or not isinstance(asset.get("id"), int)
+        or asset["id"] <= 0 or asset.get("size") != len(body)
+        or asset.get("digest") != digest):
+    sys.exit("not publishing: uploaded update.json metadata differs from its source bytes")
+print(asset["id"])
+EOF
+)
+gh api -H "Accept: application/octet-stream" "repos/$repo/releases/assets/$asset_id" > "$tmp/update-readback.json"
+cmp -s "$tmp/update.json" "$tmp/update-readback.json" || {
+  echo "not publishing: downloaded update.json differs from its source bytes" >&2; exit 1;
+}
 echo "ok       update.json"
 
 gh api -X PATCH "repos/$repo/releases/$id" -f tag_name="$tag" -f body="$(cat "$tmp/release-body.txt")" -F draft=false -F prerelease=false \

tests/decky-runtime/cases.json

@@ -14,6 +14,11 @@
     "shim": null,
     "probe": "toggle-settings"
   },
+  {
+    "id": 63,
+    "shim": null,
+    "probe": "free-loader"
+  },
   {
     "id": 12,
     "shim": "music-control-system-dbus-env",

tests/decky-runtime/check.py

@@ -240,6 +240,30 @@ def main():
                 report['result'] = ('original empty argument bug reproduced: thumbnail policy ignored'
                     if args.reproduce_fileserver_empty_flag_regression else
                     'discovery, API1 settings, native HTTPS listing/download, thumbnail policy and stop passed; uploads disabled')
+            elif case['probe'] == 'free-loader':
+                settings = current('settings_read')
+                assert settings['notify_on_free_games'] is True, settings
+                assert settings['larger_text'] is False, settings
+                assert current('read_deals') == {}
+                # This plugin commits defaults on startup and writes an empty
+                # deals cache on first read. Both paths are private to this
+                # disposable container; no feed, notification or store call.
+                settings_dir = base/'settings'/record['folder']
+                assert (settings_dir/'settings.json').is_file()
+                assert (settings_dir/'deal_db.json').is_file()
+                current('settings_setSetting', 'larger_text', True)
+                assert current('settings_read')['larger_text'] is True
+                os.killpg(backend.pid, signal.SIGTERM)
+                backend.wait(timeout=8)
+                backend = subprocess.Popen([str(loader)], env=env, stdout=log, stderr=log,
+                                           start_new_session=True)
+                processes.append(backend)
+                until_ready(lambda: rpc('loader/get_plugins'), 'restarted Decky discovery')
+                assert until_ready(lambda: current('settings_read')['larger_text'],
+                                   'private Free Loader setting after restart') is True
+                assert current('read_deals') == {}
+                report['result'] = ('discovery, default settings, empty local deals, '
+                    'private setting write/read and restart persistence passed; feed, notifications and VR unverified')
             elif case['probe'] == 'toggle-settings':
                 def call(method, *values):
                     result = rpc('loader/call_plugin_method', record['name'], method, *values)

tests/fakegh/gh

@@ -9,6 +9,7 @@ $FAKEGH_NEW_ID (default 9001) without adding it to the listing, like GitHub's li
 lagging a fresh draft. With jq installed, --jq filters a response the way gh does
 (raw strings, compact JSON)."""
 import json
+import hashlib
 import os
 import shutil
 import subprocess
@@ -120,10 +121,17 @@ elif method == "POST" and endpoint == repo + "releases":
     answer(new, json.dumps(new))
 elif method == "DELETE" and endpoint.startswith(repo + "releases/assets/"):
     pass
+elif method == "GET" and endpoint == repo + "releases/assets/901":
+    data = open(os.environ["FAKEGH_UPLOAD"], "rb").read()
+    if os.environ.get("FAKEGH_CORRUPT_MANIFEST_READBACK") == "1":
+        data += b" "
+    sys.stdout.buffer.write(data)
 elif method == "POST" and endpoint.startswith("https://uploads.github.com/" + repo + "releases/"):
-    with open(inp) as src, open(os.environ["FAKEGH_UPLOAD"], "w") as dst:
-        dst.write(src.read())
-    print("{}")
+    data = open(inp, "rb").read()
+    with open(os.environ["FAKEGH_UPLOAD"], "wb") as dst:
+        dst.write(data)
+    print(json.dumps({"id": 901, "name": "update.json", "size": len(data),
+                      "digest": "sha256:" + hashlib.sha256(data).hexdigest()}))
 elif method == "PATCH" and endpoint.startswith(repo + "releases/"):
     rid = endpoint[len(repo + "releases/"):]
     if not rid.isdigit():  # what GitHub said to v0.4.2's PATCH of releases/ (an empty id)

tests/smoke/frame_smoke.py

@@ -28,8 +28,8 @@ rechecked around deletion; shared Proton services are never stopped by name.
 
 Usage: python3 tests/smoke/frame_smoke.py [--pair]     (or scripts/frame-smoke.sh)
 Env:   FRAME_ALIAS (default frame), FRAME_SMOKE_RESULTS (default tests/smoke/results)
-Exit status: 0 no step failed (optional runtime may skip a launch), 1 a step
-failed, 2 the headset isn't reachable.
+Exit status: 0 every step passed, 1 a step failed or was skipped, 2 the
+headset isn't reachable. A skipped launch is not a release smoke pass.
 """
 import argparse
 import json
@@ -425,7 +425,7 @@ def run_smoke(args, before):
                                'passed': passed, 'skipped': skipped, 'failed': failed,
                                'devkit_utils_there_before': had_utils, 'steps': smoke.steps}, indent=1))
     print(f'==> {passed} passed, {skipped} skipped, {failed} failed on build {build}; results in {out}')
-    return 0 if failed == 0 else 1
+    return 0 if failed == 0 and skipped == 0 else 1
 
 
 def main():

tests/test_audit_release_assets.py

@@ -2,6 +2,7 @@
 import sandbox  # noqa: F401  (first: keeps tests off real data and services)
 import hashlib
 import importlib.util
+import json
 import plistlib
 import struct
 import tempfile
@@ -33,6 +34,34 @@ def release_assets():
 
 
 class AssetAudit(unittest.TestCase):
+    def test_published_manifest_matches_release_bytes(self):
+        assets, _ = release_assets()
+        release = {"tag_name": "v9.8.7", "body": "Release notes", "assets": assets}
+        manifest = {
+            "version": "9.8.7",
+            "page": "https://github.com/saphid/frame-control/releases/tag/v9.8.7",
+            "notes": "Release notes",
+            "assets": [{key: asset[key] for key in ("name", "size", "digest")}
+                       for asset in assets if asset["name"] in audit.INSTALLERS],
+        }
+        with tempfile.TemporaryDirectory() as directory:
+            path = Path(directory) / "update.json"
+            path.write_text(json.dumps(manifest), encoding="utf-8")
+            self.assertEqual(audit.validate_update_manifest(path, release),
+                             {"version": "9.8.7", "installers": 10})
+            for change in (
+                {"version": "9.8.6"},
+                {"page": "https://github.com/elsewhere/releases/tag/v9.8.7"},
+                {"notes": "stale notes"},
+                {"assets": manifest["assets"][:-1]},
+                {"assets": [dict(manifest["assets"][0], digest="sha256:" + "0" * 64),
+                            *manifest["assets"][1:]]},
+            ):
+                with self.subTest(change=change):
+                    path.write_text(json.dumps({**manifest, **change}), encoding="utf-8")
+                    with self.assertRaises(audit.AuditError):
+                        audit.validate_update_manifest(path, release)
+
     def test_complete_metadata_uses_release_checksum_order(self):
         assets, sums = release_assets()
         by_name, expected = audit.validate_assets({"assets": assets})
@@ -118,6 +147,26 @@ class AssetAudit(unittest.TestCase):
                     audit.audit("v9.8.7", report)
             self.assertFalse(report.exists())
 
+    def test_bad_published_manifest_creates_no_report(self):
+        assets, sums = release_assets()
+        assets.append({"id": 100, "name": "update.json", "size": 2,
+                       "digest": "sha256:" + hashlib.sha256(b"{}").hexdigest()})
+        release = {"id": 42, "tag_name": "v9.8.7", "draft": False,
+                   "body": "Release notes", "assets": assets}
+
+        def download(asset, path):
+            path.write_bytes(sums if asset["name"] == "SHA256SUMS" else b"{}")
+            return asset["digest"][7:]
+
+        with tempfile.TemporaryDirectory() as directory:
+            report = Path(directory) / "report.json"
+            with mock.patch.object(audit, "find_release", return_value=release), \
+                 mock.patch.object(audit, "download_asset", side_effect=download), \
+                 mock.patch.object(audit.shutil, "which", return_value="/fake/tool"):
+                with self.assertRaisesRegex(audit.AuditError, "version differs"):
+                    audit.audit("v9.8.7", report)
+            self.assertFalse(report.exists())
+
 
 if __name__ == "__main__":
     unittest.main()

tests/test_framey_decky_runtime.py

@@ -21,7 +21,7 @@ class RuntimeCoverage(unittest.TestCase):
             self.assertIn(case['id'], official)
             self.assertIn(case['probe'], ('mpris', 'screenshots', 'system-monitor', 'timer-lifecycle',
                                          'toggle-settings', 'fileserver', 'localsend', 'network-sentinel',
-                                         'battery-tracker'))
+                                         'battery-tracker', 'free-loader'))
 
     def test_new_shim_without_runtime_case_fails_coverage(self):
         import frame_decky_store

tests/test_publish_release.py

@@ -37,14 +37,16 @@ def draft(tag_name="untagged-c6ddfed7f75d67db2e99", name="Frame Control 9.8.7",
 
 
 class PublishRelease(unittest.TestCase):
-    def run_script(self, releases, *args, tags="v9.8.7", gates=None):
+    def run_script(self, releases, *args, tags="v9.8.7", gates=None,
+                   corrupt_manifest_readback=False):
         d = Path(tempfile.mkdtemp())
         (d / "releases.json").write_text(json.dumps(releases))
         (d / "gates.json").write_text(json.dumps(gates or {}))
         env = dict(os.environ, PATH="%s:%s" % (FAKEGH, os.environ["PATH"]),
                    FAKEGH_RELEASES=str(d / "releases.json"), FAKEGH_TAGS=tags,
                    FAKEGH_LOG=str(d / "log"), FAKEGH_UPLOAD=str(d / "upload.json"),
-                   FAKEGH_GATES=str(d / "gates.json"))
+                   FAKEGH_GATES=str(d / "gates.json"),
+                   FAKEGH_CORRUPT_MANIFEST_READBACK="1" if corrupt_manifest_readback else "0")
         p = subprocess.run(["sh", str(SCRIPT), *args], env=env, capture_output=True, text=True, timeout=30)
         log = [json.loads(line) for line in (d / "log").read_text().splitlines()] if (d / "log").exists() else []
         upload = json.loads((d / "upload.json").read_text()) if (d / "upload.json").exists() else None
@@ -60,11 +62,21 @@ class PublishRelease(unittest.TestCase):
         self.assertEqual(sorted(a["name"] for a in upload["assets"]), sorted(INSTALLERS))
         self.assertIn(["api", "-X", "DELETE", "repos/saphid/frame-control/releases/assets/7"], log)
         patch = next(c for c in log if "PATCH" in c)
+        readback = next(c for c in log if any("releases/assets/901" in arg for arg in c))
+        self.assertLess(log.index(readback), log.index(patch))
         self.assertEqual(patch[3], "repos/saphid/frame-control/releases/42")
         for f in ("tag_name=v9.8.7", "draft=false", "prerelease=false", "make_latest=true"):
             self.assertIn(f, patch)
         self.assertTrue(all(c[0] == "api" for c in log))  # never `gh release ...` (GraphQL)
 
+    def test_refuses_mismatched_manifest_readback_before_publication(self):
+        p, log, upload = self.run_script([draft()], "v9.8.7",
+                                         corrupt_manifest_readback=True)
+        self.assertNotEqual(p.returncode, 0)
+        self.assertIn("downloaded update.json differs", p.stderr)
+        self.assertIsNotNone(upload)
+        self.assertFalse(any("PATCH" in call for call in log))
+
     def test_an_untagged_draft_may_carry_a_subtitle(self):
         p, log, upload = self.run_script([draft(name="Frame Control 9.8.7: faster")], "v9.8.7")
         self.assertEqual(p.returncode, 0, p.stderr + p.stdout)

tests/test_smoke_cleanup.py

@@ -134,7 +134,7 @@ class Safety(unittest.TestCase):
         launch.assert_called_once_with(gid)
         remove.assert_called_once_with(gid)
 
-    def test_optional_runtime_skip_is_counted_separately_in_saved_result(self):
+    def test_optional_runtime_skip_fails_gate_and_is_counted_in_saved_result(self):
         with tempfile.TemporaryDirectory() as root:
             output = Path(root) / 'results'
             folder = Path(root) / 'titles'
@@ -156,7 +156,7 @@ class Safety(unittest.TestCase):
                  mock.patch.object(frame_smoke, 'title_cycle', side_effect=cycle), \
                  mock.patch.object(frame_smoke.tempfile, 'mkdtemp', return_value=str(folder)):
                 self.assertEqual(frame_smoke.run_smoke(mock.Mock(pair=False),
-                                 {'activity': 3, 'proximity': [], 'battery': [100]}), 0)
+                                 {'activity': 3, 'proximity': [], 'battery': [100]}), 1)
             result = json.loads(next(output.glob('*-fake.json')).read_text())
             self.assertEqual((result['passed'], result['skipped'], result['failed']), (6, 1, 0))
             self.assertEqual(next(s for s in result['steps'] if s['step'] == 'x86_64: launch')

ui/framey-compatibility.json

@@ -4703,14 +4703,15 @@
         "https://github.com/jwhitlow45/free-loader.git"
       ],
       "status": "partial",
-      "reason": "The exact archive's duplicate names contain identical bytes and differ only in execute bits; the guarded installer now accepts them, and a private local install/remove transaction passed. The current opaque Chromium host renders Free Games but cannot load deals with backend/network denied. Exact official Python settings and empty local-deals methods passed in a no-network container with a Decky stub. Real Decky, feed refresh, notifications and worn VR remain unverified.",
+      "reason": "The exact archive's byte-identical duplicate ZIP members pass guarded install/remove. Actual Decky 3.3.0 in a private network-isolated container discovers the plugin, reads default settings and empty local deals, and retains a settings change after restart. Live deal-feed refresh, notifications, Steam store integration, real Frame ARM64 service and worn VR remain unverified; no feed request was made.",
       "blockInstall": false,
       "verification": [
         "static archive",
         "local browser",
         "fake-Frame pinned install/remove",
         "local exact-archive transaction",
-        "isolated Python backend"
+        "isolated Python backend",
+        "isolated Linux backend"
       ],
       "browser": {
         "id": 63,