CAMPAIGN 12 — the class sweep: golden 0.208.0 baked (awaiting vouch), R-256..R-263 filed, gating ranked

Part 1. Golden 0.208.0 baked on the drill VM, published and ROUND-TRIP VERIFIED — 656 150 362 B,
sha256 ba668f59..5ffb82, and ./etc/felhom-controller-image read OUT of the downloaded archive says
felhom-controller:0.208.0. Acceptance markers all green (overlay2 x1, mount points x2 rootfs+mp0,
upload HTTP 201 x1, excluding/FATAL/mp1 x0), Result=success. Token file->file, read inside the VM;
systemctl show grep = 0; committed-log grep = 0 WITH a control proving the grep works. Bake VM
destroyed, drill disk restored to virgin. NOT VOUCHED — the campaign halts there deliberately.

golden_currency_gate.py was correctly RED on arrival and is green after the bake. No --no-verify
was needed anywhere in this session.

Parts 2-4. Seven defect classes swept for siblings by class rather than by feature. Analysis only:
no product code, nothing deployed, no machine touched beyond the bake VM.

Eight new rows R-256..R-263 (ceiling moved from R-255), grouped by class in OPEN-ITEMS.md. C1
produced no new instance and has no row. The sharpest is R-260: the agent reports
operator_key_configured every heartbeat, the hub has no field for it, so the check that answers
"can the operator get into this box" returns ok for a box with no operator key installed.

Every class states whether its method re-found the known instances, because a method that cannot
re-find them has not been shown to work: C1 2/3 (verified by replaying the pre-fix templates),
C2 2/2, C3 2/3 + 1 as fixed, C4 fix-pattern re-found, C5 re-found, C6 deadcode 0/2 and bespoke 1/2,
C7 weakest and said so. Blind spots stated per class; seven suspicions investigated and DISPROVED,
including two of my own methods.

Part 4's ranking is in ROADMAP.md as G-1..G-8. Gate C5 (cross-repo tag reachability — cheap,
--fast-eligible, would have caught every R-260 instance on the introducing commit). Do NOT gate C6:
golang.org/x/tools/cmd/deadcode was measured against a PLANTED probe and is blind to unreachable
METHODS on widely-used types, which is exactly the shape both known instances have.

R-242's untouched half is recorded, not built: this bake demonstrated it, the currency gate flipping
green the moment the evidence DIRECTORY existed, before the round trip finished and with no vouch
near it.

Correction the campaign owed its own brief: escrow_stale was described as closed; it is R-247 and
READY. The live repo is the source.

Sampled rather than swept, exactly: C7 60 of 2652 production invariant comments and NONE of the 1440
test comments (that half is owed); C2 19 of 221 refusals; C3/C4 controller only. No finding was
reproduced live. STATUS.md is 100 lines against its 93-line one screen.

documentation/backlog/OPEN-ITEMS.md

@@ -369,6 +369,26 @@ unconditional promise, which CAMPAIGN-11 §7 step 7 measured the customer-facing
 | **R-210** | **Which of the 345 local images may be deleted — 131 controller tags and 62 hub tags exist ONLY on this box and are not recoverable by `docker pull`** | **WAITING-ON-OPERATOR** — NEW 2026-08-05 | an operator ruling | **Nothing was deleted; this is a list, not an action.** The registry was queried directly: `felhom-controller` has **76** tags in Gitea vs **207** locally, `felhom-hub` **45** vs **107**. The **131 + 62 local-only tags are all OLD** — controller `0.39.0`–`0.135.0` plus `v0.35.0`–`v0.39.0`, hub `0.9.0`–`0.57.0` plus `v0.7.2`–`v0.13.0` — while everything from controller `0.136.0` and hub `0.58.0` upward IS in the registry and therefore re-pullable. **Size the prize honestly before spending a decision on it:** per-tag sizes sum to 139.29 GB, but that double-counts shared layers — `docker system df` puts the **real** dedup'd image footprint at **31.02 GB with 27.02 GB reclaimable**, i.e. an order of magnitude less than the build cache P3 already returned. `docker image prune -a` would remove 343 of 345 (only `redis:7-alpine` and `postgres:16-alpine` are held by running containers). **CC's view: not worth doing for the space** — it buys ~27 GB against 199 GB now free, and its only real benefit is dropping unrecoverable clutter | operator |
 | **R-211** | **Prometheus has no config-reloader — a rules change reaches the pod and is never read** | **READY (S) — NEW 2026-08-05** | — | Found while verifying R-205 rather than by looking for it. The `mon-system/prometheus` Deployment runs **one** container (`prom/prometheus:v3.12.0`) with **no `configmap-reload`/`prometheus-config-reloader` sidecar**. After the ArgoCD sync the updated `node-housekeeping-alerts.yml` was present **inside the pod** (`grep -c "and on(instance)"` → 3 on the mounted symlink) while the Prometheus **rules API still served the old expression** — for **4+ minutes**, with no error anywhere. It only took effect after an explicit `POST /-/reload`. **The consequence is general, not specific to R-205: every rule edit in this repo since the stack was built has silently not applied until something happened to restart the pod** — so "committed and synced" has never meant "in force", and ArgoCD reporting `Synced/Healthy` is true and beside the point. `--web.enable-lifecycle` IS already set, so the fix is small: add a reloader sidecar watching the ConfigMap, or a `checksum/config` pod annotation so a rules change rolls the pod. **Same class as the four *built-but-never-wired* seams** — the control exists, nothing walks it | CC |
 
+## CAMPAIGN 12 — the class sweep, 2026-08-08 (unattended)
+
+Eight rows, **grouped by class so the classes are visible as classes**. Full method, controls, blind
+spots and the Part-4 gating ranking: `audits/CAMPAIGN-12-class-sweep-2026-08-08.md`. Gating candidates
+are in `ROADMAP.md`, not here. **C1 produced no new instance** and has no row, deliberately.
+
+**⚠ A correction the campaign owed to its own brief:** the task described C5's `escrow_stale` instance
+as "closed individually". **It is not — it is R-247, `READY`.** The live repo is the source.
+
+| ID | What | State |
+|---|---|---|
+| **R-256** | **C2 — „A mentéskezelő nem elérhető." names no route at all.** `web/offbox_handlers.go:47` and `:181` flash this to the customer on the off-site backup surface. It states an internal component's unavailability in the operator's vocabulary („mentéskezelő" = the backup Manager object), gives no reason the customer can act on, and names no next step — not "try again in a few minutes", not "contact support", not a page to go to. **Contrast, in the same subsystem and shipped the same week:** R-252's fix reads *„Meghajtók", „Meglévő meghajtó csatolása". Utána gyere vissza ide.* — a route. **Severity is low and stated so it is not over-ranked:** the condition is a nil backup manager, which on a healthy box does not occur; this is about the copy, not a broken path. Found by the C2 sample (19 refusals on the recovery/restore/offbox surface; **~202 of the repo's 221 refusal strings were NOT examined**) | **READY** — owner Viktor |
+| **R-257** | **C2 — „Az offsite tároló nincs elárvult állapotban." puts an English loanword and an internal state name in front of a Hungarian household customer, and names no route.** `web/offbox_handlers.go:270` (the Go error it mirrors is `backup/offbox.go:343`). „Offsite" is untranslated; „elárvult állapot" is the codebase's own `OffboxOrphaned()` predicate surfacing verbatim. A customer who pressed a button and got this cannot tell whether something failed, whether they did something wrong, or what to do instead. **This is a refusal that is CORRECT and fail-closed and still a dead end** — the same shape R-241 recorded for `--recover-offsite-install`. **Fix shape, not a decision:** say what the customer tried to do, why it does not apply right now, and where to look — or, since this is a state they cannot reach deliberately, do not offer the action at all | **READY** — owner Viktor |
+| **R-258** | **C3 — the customer's per-app backup tick is green on the PRESENCE of a restore point, and its only red condition is a GLOBAL one.** `web/handlers.go:1240-1248`: `row.Tier1LastStatus = "ok"` whenever `ListRestorePoints(app)` returns ≥1 point, and `"error"` only when `status.LastDBDump != nil && !status.LastDBDump.Success` — where `LastDBDump` is the box's **single most-recent DB dump** (`backup/backup.go:1064`, `m.lastDBDump`), **not this app's**. Three consequences, in increasing order of how wrong they look to a customer: (a) an app whose own last backup failed shows a **green tick** provided any older restore point exists and some *other* app dumped successfully afterwards; (b) an app with **no database at all** takes the `nil` branch and is green on presence alone; (c) the tick asserts nothing about **recency** — a restore point from three weeks ago is as green as one from last night, while `row.Tier1LastRun` beside it carries the real (old) time. **It is customer-visible:** `templates/backups_apps.html:173-174` renders it as a check or a cross. **This is the project's own "presence is not success" rule as a UI badge** — the artifact's existence is being read as a successful result, which is what the workspace `CLAUDE.md` section of that name forbids. **Found by a COMPLETE sweep, not a sample:** all 9 success-status assignment sites in the controller were read; this is the only new one. **NOT reproduced live** — source reading only | **READY** — owner Viktor |
+| **R-259** | **C4 — a disk read that FAILS renders as „0.0 GB / 0.0 GB (0%)" in the nominal colour, on the dashboard's most-looked-at meter.** `system/info_linux.go:259-264`: `readDiskUsage` logs at DEBUG and **returns**, leaving the caller's `TotalGB`/`UsedGB`/`AvailGB`/`Percent` at their zero values. `dashboard.html:57-63` then renders `{{fmtGB .SystemInfo.DiskUsedGB}} / {{fmtGB .SystemInfo.DiskTotalGB}} ({{printf "%.0f" .SystemInfo.DiskPercent}}%)` plus a meter whose fill is `width:0%`, and `usageColor(0)` returns **`"nominal"`** (`web/funcmap.go:189-196`) — so **a failed measurement is drawn as a healthy, empty disk.** There is **no `*Known` companion for the system disk** (grep: 0). **The fix pattern already exists in this codebase, two files away, with the reasoning written out:** R-225's `StatsKnown` guards every derived off-site figure and its own comment says *"a 0%-wide bar over an unread store is a picture of emptiness, and a picture is a claim"* (`backups_remote.html:60-62`). The dashboard does not use it. **It also travels:** `report/builder.go:94` puts the same zeroed `TotalGB`/`UsedGB` into the host report, so a failed statfs reaches the hub as a 0-of-0 disk. **Observation attached, not filed separately:** `readDiskUsage` is one of several collectors in that file that return silently on error; only this one was traced to a customer-visible surface. **NOT reproduced live** | **READY** — owner Viktor |
+| **R-260** | **C5 — the agent reports at least eight decision-bearing facts the hub models NOWHERE, and the sharpest one blinds the check that answers „can the operator get into this box".** Measured 2026-08-08 by a tag-reachability test over all four wire directions (465 emitted json tags; a tag whose literal string occurs nowhere in the receiving repo cannot be decoded by any struct, named or anonymous). **THE ONE THAT MATTERS: `operator_key_configured`.** The agent emits it every heartbeat (`hub/report.go:178` — *"operator authorized_key installed"*); the hub's OOB decoder (`store/host_oob.go:38-45`) mirrors **5 of the agent's 8** OOB fields and has no field for it, nor for `wg_handshake_age_s` or `healed_at`. So `oobDegraded` (`monitor/host_oob.go:60-70`) tests config-invalid and (peer-configured AND not-active-or-not-reachable) — and **a box with felhom-sshd active, reachable, valid config, peer configured and NO OPERATOR KEY INSTALLED is `ok`.** The agent knows and says so; the hub throws it away. The checker's own doc comment claims it *"answers 'can the operator get into this box right now, and if not, why' proactively"* — see R-262's class. **Break-glass is the recovery chain (G1), which is what makes this the top row of the census.** **The rest of the census, hub-side absent (0 occurrences anywhere in `felhom.eu/hub/`):** `guest_net` — **the entire per-guest network-health object**, populated every cycle at `hub/collect.go:290`, carrying `dhclient_alive`, `has_route`, `heal_succeeded`, `heals_last_hour`, `last_heal_at`, `damped`; `selfupdate_pending` + `selfupdate_pending_version` — a staged-but-unapplied agent update is invisible to the fleet view; `healed_recently` (mgmt-plane); `operator_key_configured`, `wg_handshake_age_s`, `healed_at` (OOB); `mount_parity` + `mount_inventory` (restore-test, see R-262); `cpu_temp_c`, `loadavg`, `memory_total_bytes`, `memory_used_bytes`, `uptime_seconds` (host metrics); `model_name` (SMART); `applied_at` (PBS-DR). **Controller→hub, hub-side absent:** `reporting_disabled`, `config_hash`, `last_db_dump`, `last_integrity_check`, `migrated_to`, `catalog_ref`, `storage_bindings`, `container_path`, `subpath`, `temperature_celsius`, `load_avg_{1,5,15}`, `memory_{total,used}_mb`, `uptime_seconds`. **⚠ STATED PRECISELY, because it first looked worse than it is:** the hub stores the whole report as `report_json`, so nothing is lost from the DATABASE — every hub consumer, without exception, re-unmarshals it into a typed struct, so nothing reaches a check, an alarm, a notification or a screen. And a **restore-test mount-parity mismatch is NOT hidden**: it sets `res.Err` and returns before `res.Pass = true`, so `pass:false` travels; what is lost is the *depth* of a pass, not the failure. **The class is what this row is about, not any single field** — this is `escrow_stale` (R-247) for the fourth, fifth and sixth time, and §4 of the campaign report argues it is the cheapest high-value gate available to this project. **Blind spot of the method, recorded:** 29 generically-named tags (`name`, `state`, `status`, …) were excluded by name, so a drop of a generically-named field would be missed | **READY** — owner Viktor |
+| **R-261** | **C6 — `CountSelfBindTokens` exists so that callers can assert an invariant, and no production caller asserts it.** `hub/internal/store/selfbind.go:106-111`. Its doc comment: *"it exists so callers can assert the 'after this runs, the only live link is one we just issued — or none' invariant that the auto-mint at customer-create / RESET-completion depends on."* **Census: only its own declaration in production; the two callers are `selfbind_automint_test.go:29` and `customer_delete_test.go:510`.** Tests are not callers (the campaign's rule), so the invariant the auto-mint *depends on* is checked in the test suite and never at the moment it matters. **This is the smallest of the eight rows and is filed at its true size, because the rest of the C6 sweep found INERT dead accessors rather than defects:** `OffboxOrphanedRenamedTo` and `OffboxEscrowState` have no caller but their data reaches the card another way (the template reads the settings field directly, `backups_remote.html:80`) — **R-228 is genuinely closed, and the sweep's first reading that it had regressed was wrong.** **The more consequential C6 result is a method result and is in the report, not here:** `golang.org/x/tools/cmd/deadcode` re-finds **neither** known instance, and a planted probe measured why — it reports an unreachable exported FUNCTION and not an unreachable exported METHOD on a widely-used type, and both known instances are methods | **READY** — owner Viktor |
+| **R-262** | **C7 — a comment claims a cross-repo contract is mirrored „field-for-field" and „the key-set tests guard drift"; it is two fields short, AND THE FIXTURE THE TEST READS OMITS THE SAME TWO FIELDS.** `hub/internal/api/handler.go:682-687` covers `hostBackup` **and** `hostRestoreTest`. **It is TRUE of `hostBackup`** (verified field-for-field against `agent/internal/hub/Backup`). **It is FALSE of `hostRestoreTest`:** the agent emits `mount_parity` and `mount_inventory` (`hub/report.go:432-433`, populated in production from `reconcile/restoretest.go:277-283` via `backup/runner.go:517`), and the hub has no field for either — **0 occurrences in the entire hub repo** outside the CHANGELOG. **The guard is blind in exactly the place the drift is:** `TestHostReport_GoldenContract` reads `testdata/host-report.golden.json`, the two copies of which are byte-identical as required — and **neither contains `mount_parity` or `mount_inventory` at all**, so the key sets agree on a shape that is not the shape the agent sends. A test that cannot fail on the drift it names is the R-97b lesson (*prove the consequence, not the mechanism*) landing on a contract test. **Consequence, stated precisely:** the verdict is not lost (a parity mismatch fails the test before `Pass` is set), but the hub cannot distinguish a full-fidelity restore-test pass from a boot-only one, for any agent, ever. **Fix shape, not a decision:** add the two fields and put them in the fixture — or narrow the comment to name `hostBackup` only and say plainly that `hostRestoreTest` is a subset. **Attached observation:** the same fixture carries `cpu_temp_c` and `loadavg`, which no hub struct decodes — a fixture carrying keys the receiver cannot read is the same shape one level down | **READY** — owner Viktor |
+| **R-263** | **C7 — „This is the ONLY writer of `StoragePath.BackupTarget`" is false, and nothing pins it.** `settings/settings.go:1317-1319`, on `SetBackupTarget`. **`ClearBackupTarget` (`:1358-1363`) also writes the field**, 17 lines below, in the same file. **The GUARANTEE the comment protects is intact and that is why this is filed small:** the sentence continues *"registration must never set it (E-2 §3: a drive never acquires a role by appearing)"*, and `ClearBackupTarget` only ever writes `false`, so no path other than `SetBackupTarget` **grants** the role. **What is wrong is the claim as written, and the absence of anything holding it:** `backup_target_role_test.go` exercises the behaviour and asserts nothing about writer uniqueness, so if a third writer appeared tomorrow — one that granted — the comment would still read as settled and the suite would still be green. **This is the class's own definition:** an invariant asserted in prose with no test pinning it. **Fix shape:** one word (*"the only writer that GRANTS the role"*) plus a test that fails when a second granting writer appears. **Method honesty:** found in a sample of **60 of 2652** production invariant comments — the "is the only" form only, chosen because a uniqueness claim is the one form a grep can falsify. **~2592 production and all 1440 test invariant comments were NOT examined**, so C7 has the weakest coverage of the seven classes and the task's instruction to check the tests' own claims is **owed, not discharged** | **READY** — owner Viktor |
+
 ## Why the TOP READY rows rank this way
 
 

documentation/backlog/ROADMAP.md

@@ -128,6 +128,24 @@
 | R-109 | ~~The DR recipe records no backup target~~ | XS | **SHIPPED** — agent v0.118.1 + hub v0.83.0, 2026-07-30 | `backup_target` resolves from the PRIMARY tier of `cfg.Backup.BackupTiers()` (the scheduler's own function) + the mountpoint that separates `/mnt/hdd_1` from `/var/lib/vz`; unresolvable is recorded as unresolvable, never defaulted. Required an unscoped HUB half: `AssembleDRRecipe` allow-lists top-level keys → **R-122**, a fourth recipe-completeness defect that had already shipped (`offsite_restic` dropped for the feature's whole life). `audits/R106-R109-recipe-completeness-2026-07-30.md`; flipped `07` §4 |
 | E-2 | **Drive-role machinery around the moved vzdump target.** The 2026-07-28 runbook proved the architecture change by hand on both demo boxes; this is the machinery: a **backup-target role** on `StoragePath` beside `Schedulable`/`IsDefault`/`Kind`; **assignment in the storage wizard** (suggest by attribute, refuse the absurd, never decide by transport or `removable` — on the reference hardware demo-felhom's target IS a USB HDD and BOTH drives report `removable=0`); **unassigned drives do nothing automatically**; **stickiness** (never silently retarget); `felhom-host-install.sh` creating the target with `--is_mountpoint 1` **and** issuing the `FelhomAgentStore` ACL; **absent-target policy**; **retention/space accounting** on a drive the customer shares; the honest **single-drive label**; remaining fleet migration | M | READY — 2026-07-28 | Full scope + rationale in `runbooks/RUNBOOK-vzdump-target-move-2026-07-29.md` §7. Two traps already paid for live: the storage `path` must BE the mountpoint or the agent reports the target `disconnected` forever (`internal/storage/observe.go:321`), and the per-storage `FelhomAgentStore` grant is mandatory or every backup 403s. Absent-drive behaviour today is **fail-loudly, no silent retarget** (`is_mountpoint 1` proven live) — which is NOT the intended fall-back-and-alarm design. Flips: matrix row 4 |
 
+## Gating candidates — Campaign 12, Part 4 (2026-08-08)
+
+**Ranked by what a gate would be worth, using Campaign 12's own instance counts as the evidence.**
+Nothing here was built; §4 of `audits/CAMPAIGN-12-class-sweep-2026-08-08.md` carries the reasoning and
+the measurements. The recurring lesson these rank against: **a pattern found three times is not closed
+by looking a fourth time.**
+
+| Rank | Item | Size | Status | Notes |
+|----|------|------|--------|-------|
+| **G-1** | **Gate C5 — the cross-repo tag-reachability check.** For every json tag emitted on a real wire (agent→hub, controller→hub, agent→controller, hub→both), assert that the literal tag string occurs somewhere in the receiving repo. A tag occurring nowhere cannot be decoded by any struct, named **or anonymous** — which is what makes this definitive where struct-pair diffing is not. | S (~150 lines Python) | **candidate — recommended first** | **The cheapest high-value gate available to this project.** No network, no container runtime → `--fast`-eligible, so it runs in BOTH the pre-push hook and CI (the constraint that forced `golden_currency_gate.py` to check the bake rather than the vouch). It would have caught `escrow_stale` (R-247), `operator_key_configured` and every other row of R-260's census **on the commit that introduced them**. **Design requirement, not optional:** an allowlist with stated reasons, in the shape `secret_in_markup_gate.py` already uses — the hub deliberately ignores some collections and says so. **Known miss:** generically-named tags (`name`, `state`, `status`, …), where a repo-wide string test is meaningless; 29 were excluded by name in the sweep. → R-260 |
+| **G-2** | **Gate C3 — a success verdict may not be set where an incompleteness signal is in scope.** Assert that every literal success-status assignment either has no gap/skip/missing signal available at that point, or consults it. | S | candidate | The whole population in the controller is **9 sites** — Campaign 12 read all of them, which is why this class is the one where "no others exist" is supportable. Small enough to gate by enumeration rather than by inference. **Known miss:** verdicts expressed as booleans, enum constants, or the absence of an error — and the controller does use those elsewhere. Instances: R-240 (open), R-258 (new). |
+| **G-3** | **Gate C4 — every rendered count/size/percentage needs a `*Known` companion.** | M | candidate — **needs a convention decision first** | R-225's `StatsKnown` is the house pattern and it is exemplary; a gate can only enforce it once it is *the* house style. Today three-state is also encoded with pointers and with separate error fields, both legitimate, and a name-based gate reads those as unguarded. **The decision owed is "how does this codebase say 'we could not look'", not "should we gate it".** Instance: R-259. |
+| **G-4** | **Complete C1's runtime body assertion — 4 of 27 pages today.** | L | candidate — the expensive one, and honestly so | `secret_in_markup_gate.py` covers all 36 templates on the NAME-based check and is **blind to a secret under a neutral page-data key** — verified 2026-08-08 by replaying the three pre-fix templates through it: it convicts 2 of 3 and not the third. The runtime assertion catches all three; extending it means constructing each remaining page's data in a test, which is a per-page cost and is the real reason it has not been done. **Do NOT adopt the Go-side mirror Campaign 12 wrote as a gate on its own** — 27 candidates, 0 findings is bad signal-to-noise in front of every push. → R-255 |
+| **G-5** | **Gate C7, narrowly — uniqueness claims only.** A comment saying "X is the ONLY writer/place/caller of Y" is mechanically falsifiable; assert it. | S | candidate — narrow by construction | Covers ~60 of the 2652 production invariant comments. The other ~97% of the vocabulary (`never`, `always`, `must not`, `guarantees`) is not mechanical and a gate must not pretend otherwise. Instance: R-263. |
+| **G-6** | **C2 — NOT mechanically gateable.** | — | **recorded as a no** | "Names a route" is a judgement, not a predicate. The most a check could do is enforce a *convention* (e.g. every customer-visible refusal string ends in an imperative clause), which would be gamed rather than followed. Better served by the UI-copy review the `felhom-ui-design` skill already governs. Instances: R-256, R-257. |
+| **G-7** | **C6 — NOT gateable, and the measurement is the finding.** | — | **recorded as a no, with evidence** | The class *looks* the most mechanical of the seven and is the one where the off-the-shelf tool measurably fails: `golang.org/x/tools/cmd/deadcode` re-found **neither** known instance, and a **planted probe** showed why — it reports an unreachable exported FUNCTION but not an unreachable exported METHOD on a widely-used type, and both known instances are methods. A bespoke gate would be conservative by construction and would spend most of its output on **inert dead accessors** (4 of the 5 residual candidates were inert). **Recommendation: gate C5 instead** — it catches a strict subset of the same "the answer was available and discarded" family with none of the ambiguity. → R-261 |
+| **G-8** | **R-242's untouched half — catch a SKIPPED VOUCH.** | S then M | candidate — **the one with a live recurrence** | Measured during Campaign 12's own bake: `golden_currency_gate.py` **flipped red→green the moment the evidence DIRECTORY existed**, before the round-trip download finished and with no vouch near it. **(a) Cheapest, now:** have the bake session re-read `/configuration` after the operator's Save and write the observed `golden_version` into the evidence README as a machine-readable line; the gate then requires that line rather than the directory. Detects the FORGETTING, which is the actual failure mode. **(b) Loudest, when the hub is next touched:** a hub-side daily check comparing the vouched golden against the newest controller the fleet reports — it fires within a day and catches a silent ROLLBACK too, which nothing in git can ever see. **(c) A checklist item is not a fix** — R-242 already was a rule without a mechanism and it recurred the next day. → R-242 |
+
 ## P3 — post-alpha
 
 | ID | Item | Size | Status | Notes |