# Driver-expansion — Waves 0–2 tracking > **Living status tracker** for the driver-expansion program (Waves 0, 1, 2). One row per > deliverable, each pointing at its **design doc**, its **implementation plan** (once written), > and its current **status**. The authoritative *architecture* index is the program design doc; > this file is the authoritative *progress* index. > > **Program design (architecture / shared contract / build order):** > [`2026-07-15-driver-expansion-program-design.md`](2026-07-15-driver-expansion-program-design.md) > > Last updated: 2026-08-05. ## Legend | Status | Meaning | |---|---| | ✅ **Done** | Merged to master, tests green. | | 🟡 **Live gate open** | Code merged; a live `/run` or hardware-gated verification still outstanding. | | 📝 **Plan ready** | Executable implementation plan (`*-implementation.md` + `.tasks.json`) written; not yet built. | | 📐 **Design only** | Design doc exists; **no** implementation plan yet — writing-plans is the next step. | | ⛔ **Not started** | No design, no plan. | **Doc types** (per the writing-plans skill): a *design* states architecture, decisions, and risks; an *implementation plan* is the bite-sized, TDD, file-path-level task list the subagent-driven executor runs off, with a co-located `.tasks.json` for resume. A deliverable is only buildable once it reaches 📝. ## Summary | Wave | Deliverable | Design | Impl. plan | Status | Effort | Fixture / hardware | |---|---|---|---|---|---|---| | **0** | Universal Discover-backed browser | [design](2026-07-15-universal-discovery-browser-design.md) | [plan](2026-07-15-universal-discovery-browser-implementation.md) · [tasks](2026-07-15-universal-discovery-browser-implementation.md.tasks.json) | ✅ **Done** — merged `056887d6`; live gate (Gitea #468) CLOSED 2026-08-01 via the patched AbCip fixture (`c2c3c522`) | S–M | patched libplctag `ab_server` (serves `@tags`) at `10.100.0.35:44818` | | **1** | Modbus RTU (extend Modbus) | [design](2026-07-15-modbus-rtu-driver-design.md) | [plan](2026-07-24-modbus-rtu-driver.md) · [tasks](2026-07-24-modbus-rtu-driver.md.tasks.json) | ✅ **Done** — merged to master `0f38f486` (#495), 11 tasks, live gate PASSED | **S (lowest)** | `pymodbus` `rtu_over_tcp` — CI-simulatable, no new infra | | **1** | SQL poll | [design](2026-07-15-sql-poll-driver-design.md) | [plan](2026-07-24-sql-poll-driver.md) · [tasks](2026-07-24-sql-poll-driver.md.tasks.json) | ✅ **Done** — merged to master `4ad54037`, 22 tasks, live gate PASSED; follow-ups #496/#497/#498 closed in `28c28667` | S–M | **existing central SQL Server** `10.100.0.35,14330` + SQLite unit | | **2** | MTConnect Agent | [design](2026-07-15-mtconnect-driver-design.md) | [plan](2026-07-24-mtconnect-driver.md) · [tasks](2026-07-24-mtconnect-driver.md.tasks.json) | ✅ **COMPLETE** — P1 Agent MVP, all 23 tasks + the 5-leg live gate PASSED (branch `feat/mtconnect-driver`, PR #506) | S–M | Dockerized `mtconnect/agent:2.7.0.12` (not `cppagent`) — CI-simulatable, live at `10.100.0.35:5000` | | **2** | MQTT / Sparkplug B | [design](2026-07-15-mqtt-sparkplug-driver-design.md) | [plan](2026-07-24-mqtt-sparkplug-driver.md) · [tasks](2026-07-24-mqtt-sparkplug-driver.md.tasks.json) | ✅ **COMPLETE** — P1 + P2 shipped, both live-gated (Tasks 0–26) | M–L | Mosquitto TLS+auth **+ C# Sparkplug edge-node simulator**, live at `10.100.0.35:8883` | > Waves 3+ (BACnet/IP, Omron) and the deferred MELSEC are tracked in the program design doc §6/§8, > not here. Omron CIP is the program's sole real-hardware wire gate; BACnet's broadcast/BBMD leg is > env-gated live. Both are out of this file's scope until they're pulled forward. --- ## Executing a plan (git worktree + subagent-per-task) Each 📝 plan is executed with the **subagent-driven-development** skill: it sets up an isolated **git worktree** first (via `using-git-worktrees`), then dispatches a **fresh subagent per task**, running the classification-driven review chain (`trivial` = implement only … `high-risk` = spec-review → code-review → integration review) between tasks. The `.tasks.json` next to each plan tracks progress and lets a later session resume. > ⚠️ **All four plans below have been executed and merged.** This table is retained as the *pattern* > for executing a future driver plan — do **not** run these four commands, they would rebuild shipped > drivers in a fresh worktree. (Their `.tasks.json` files still read `pending`; that is stale > bookkeeping, not backlog — see `deferment.md` §6.1.) | Deliverable | Status | Command (pattern only — do not re-run) | |---|---|---| | Modbus RTU (Wave 1) | ✅ merged `0f38f486` | ~~`… execute docs/plans/2026-07-24-modbus-rtu-driver.md …`~~ | | SQL poll (Wave 1) | ✅ merged `4ad54037` | ~~`… execute docs/plans/2026-07-24-sql-poll-driver.md …`~~ | | MTConnect (Wave 2) | ✅ merged `90bdaa44` (#506) | ~~`… execute docs/plans/2026-07-24-mtconnect-driver.md …`~~ | | MQTT/Sparkplug (Wave 2) | ✅ merged `c3a2b0f7` | ~~`… execute docs/plans/2026-07-24-mqtt-sparkplug-driver.md …`~~ | For a **new** driver the invocation shape is: `Use superpowers-extended-cc:subagent-driven-development to execute in a new git worktree` - **Slash-command form** (equivalent): `/superpowers-extended-cc:subagent-driven-development `. - **Parallel-session / resume form** (batch execution with checkpoints, no fresh-subagent-per-task): `/superpowers-extended-cc:executing-plans ` — reads the same `.tasks.json` and continues from the first pending task. - **Order (historical):** the four were built lowest-effort-first — Modbus RTU → SQL poll → MTConnect → MQTT/Sparkplug. All are merged; nothing in this list remains to schedule. Run one plan per worktree; independent plans may run concurrently in separate worktrees. --- ## Wave 0 — Universal Discover-backed browser ✅ **What it is.** One generic `DiscoveryDriverBrowser` (+ `CapturingAddressSpaceBuilder`, `CapturedTreeBrowseSession`, `BrowserSessionService` fallback, and the `ITagDiscovery.SupportsOnlineDiscovery` gate) that turns any driver's `ITagDiscovery.DiscoverAsync` into an AdminUI browse tree. It is the **Wave-0 gate**: every browsable new driver depends on this seam, and it retrofits browse to the already-shipped AbCip / TwinCAT / FOCAS drivers for near-zero marginal cost. - **Design:** [`2026-07-15-universal-discovery-browser-design.md`](2026-07-15-universal-discovery-browser-design.md) - **Implementation plan:** [`2026-07-15-universal-discovery-browser-implementation.md`](2026-07-15-universal-discovery-browser-implementation.md) · [`.tasks.json`](2026-07-15-universal-discovery-browser-implementation.md.tasks.json) - **Status: ✅ Done — merged, live gate CLOSED.** - Merged to master — `056887d6` (*Merge feat/universal-discovery-browser — Wave-0 universal Discover-backed browser*), 2026-07-15. - Implementation plan: **19 / 19 tasks completed.** - Lit up AbCip / TwinCAT / FOCAS pickers with zero per-driver browse code. - **Live gate (Gitea #468) — CLOSED 2026-08-01** (`c2c3c522`). The fixture block was removed by teaching the libplctag `ab_server` sim to serve the Symbol Object `@tags` walk (vendored patches in `tests/Drivers/…AbCip.IntegrationTests/Docker/patches/`); all three criteria then ran green on docker-dev — populated nine-tag tree render, Refresh re-capture, and leaf commit round-trip. See "Second live-gate result" in the [implementation plan](2026-07-15-universal-discovery-browser-implementation.md). - **The gate earned its keep — three real defects fixed, none reachable by unit tests:** 1. `CipSymbolObjectDecoder` padded to even byte boundaries between packed entries, so any odd-length symbol name silently dropped every tag behind it (the test builder padded identically, so all tests agreed with each other and none with a controller). 2. **Gitea #536** — browse-commit discarded the discovered data type (a browsed REAL committed as `DInt`), fixed by carrying the driver's own type name via `NativeDataType`; affected AbCip/TwinCAT/FOCAS. 3. AbCip Test Connect misclassified the no-authored-tags probe **on real hardware too** — the reachable-exception matcher looked for native C status spellings libplctag.NET never emits (plus a sim-side 0x03/0x04 CIP status mismatch). - **Residual follow-up — CLOSED 2026-08-05.** The named `*DriverPage` shells were already deleted when v3 Batch 2 retired the routed `/clusters/{id}/drivers` flow, but the defect pattern had survived into the v3 hosts: `RawTagModal` and `RawManualTagEntryModal` passed a `"{}"` stub as `GetDriverConfigJson`, so the AbCip/TwinCAT/FOCAS pickers' "Browse controller" captured against no endpoint. Fixed by resolving `LoadMergedProbeConfigAsync` per open and forwarding a live accessor (design + outcome: `2026-08-05-raw-modal-picker-live-config-design.md`). Review of the fix surfaced and closed a further stale-open race (a superseded open could install another device's config); `RawBrowseModal` shares that pre-existing race shape and is recorded in `deferment.md` §0.1, not fixed here. Pinned by `RawTagModalMergedConfigTests` + `RawManualTagEntryModalMergedConfigTests`, and **live-gated 2026-08-05** — all three legs PASSED against the patched AbCip fixture (edit modal + manual entry both browse the nine-tag tree; an unreachable device degrades to the error chip and still saves). The gate also found **Gitea #539**: the AbCip *discovery* path drops the authored EtherNet/IP port, so a device on a non-default port browses whatever answers on 44818 — #534's defect surviving in `LibplctagTagEnumerator` + `LibplctagTemplateReader`. --- ## Wave 1 — low-effort / high-leverage pair ✅ Both are **fully CI-simulatable with zero new hardware**; Modbus RTU needs no new infra at all, and SQL poll reuses the always-on central SQL Server. **Both shipped.** ### Modbus RTU — ✅ Done (merged `0f38f486` / #495, live gate PASSED) - **Design:** [`2026-07-15-modbus-rtu-driver-design.md`](2026-07-15-modbus-rtu-driver-design.md) - **Implementation plan:** [`2026-07-24-modbus-rtu-driver.md`](2026-07-24-modbus-rtu-driver.md) · [`.tasks.json`](2026-07-24-modbus-rtu-driver.md.tasks.json) — **11 tasks** (1 trivial / 5 small / 2 standard / 3 high-risk), all complete via subagent-driven-development. - **Scope:** extend the existing Modbus driver with a `Transport` selector (RTU CRC-16 + FC-aware length, no TxId) + a `rtu_over_tcp` `pymodbus` docker profile + the AdminUI selector. **Direct-serial transport is descoped** (user, 2026-07-15) — RTU-over-TCP via a serial→Ethernet gateway is the only shipped mode, so there is no serial hardware gate. - **Fixture:** added the `rtu_over_tcp` profile (`framer=rtu`, host `:5021`, co-runs with `standard`) to the Modbus integration fixture. **Unknown confirmed:** pymodbus 3.13's simulator DOES honour `framer=rtu` on a TCP server (wire-probed a pure RTU FC03 response, no MBAP) — so the simple profile path was taken; **no stdlib fallback server was needed.** - **Verification (all green):** unit suite **344/344**; full solution build 0 errors; the 2 RTU integration tests pass **live** against the real `framer=rtu` fixture (read HR5→5, write+readback HR200←4242). Per-task review chain (spec + code reviewers) + a final whole-branch review, all approved-to-merge. - **Live `/run` gate — PASSED (2026-07-24, docker-dev rebuilt from this branch):** - AdminUI `/raw` Modbus **driver** config modal renders the new **Transport** selector (with the operator hint) — set to `RtuOverTcp`, saved. - **Device Test Connect → `OK · 23 MS`** — the probe built a `ModbusRtuOverTcpTransport` via the factory and spoke **FC03 over real RTU framing** to `10.100.0.35:5021`. - Authored raw tags HR5 (r/o) + HR200 (r/w) → **deployment `06ec1632` Sealed** (all 6 nodes acked). - Client.CLI on `opc.tcp://localhost:4840` (`multi-role`/`password`): **read** `ns=2;s=rtu-gate/Device1/HR5` → **5, Good**; **write** `ns=2;s=rtu-gate/Device1/HR200` = 4242 → readback **4242, Good**. - **Effort:** **S — the lowest on the roadmap.** Delivered first, as recommended. ### SQL poll — ✅ Done (merged `4ad54037`, live gate PASSED) - **Design:** [`2026-07-15-sql-poll-driver-design.md`](2026-07-15-sql-poll-driver-design.md) - **Implementation plan:** [`2026-07-24-sql-poll-driver.md`](2026-07-24-sql-poll-driver.md) · [`.tasks.json`](2026-07-24-sql-poll-driver.md.tasks.json) — **22 tasks, all done**. `ISqlDialect` + `SqlServerDialect` shipped. ~~Postgres/ODBC remain deferred to P2/P3 behind the same seam.~~ **Postgres + Oracle shipped 2026-08-05** via managed providers (Npgsql 10.0.3, ODP.NET Core 23.26.300) — dialect-owned parameter markers/bind names + Oracle `BindByName`, `INFORMATION_SCHEMA` vs `ALL_*` catalog SQL, provider gates widened, AdminUI dropdown, env-gated docker suites (postgres:17-alpine / gvenzl/oracle-free:23-slim). The live suites found + fixed ORA-01745 (Oracle rejects reserved-word bind names — `:table` — so the first Oracle build was structurally broken until binds were `p_`-prefixed). MySql/Odbc stay reserved. See [`2026-08-05-sql-postgres-oracle-design.md`](2026-08-05-sql-postgres-oracle-design.md). - **Post-merge follow-ups closed** in `28c28667`: #496 catalog identifier-validation gate (`4dc2ad80`), #497 wrong status-code constants across six drivers, #498 deploy-gate rejection of a persisted `connectionString` (`1919a8e5`). - **Scope:** a bespoke schema-browser driver polling a SQL table into equipment tags (SQL Server, **PostgreSQL and Oracle** shipped behind `ISqlDialect`; MySql/Odbc reserved). - **Fixture:** SQLite unit fixture (primary) + an **env-gated integration fixture against the existing central SQL Server** (`10.100.0.35,14330`) with a seeded `SqlPollFixture` DB. The blackhole/timeout live-gate pauses a **dedicated** `mssql` container — **never** the shared central SQL Server (it hosts `ConfigDb`). - **Effort:** S–M. --- ## Wave 2 — strategic telemetry / UNS pair ✅ Both CI-simulatable on the shared docker host, no hardware. ### MTConnect Agent — ✅ Done (P1 Agent MVP) - **Design:** [`2026-07-15-mtconnect-driver-design.md`](2026-07-15-mtconnect-driver-design.md) - **Implementation plan:** [`2026-07-24-mtconnect-driver.md`](2026-07-24-mtconnect-driver.md) · [`.tasks.json`](2026-07-24-mtconnect-driver.md.tasks.json) — **23 tasks; 22 built, Task 21 (live `/run` gate) tracked separately in the plan file.** - **Driver guide:** [`docs/drivers/MTConnect.md`](../drivers/MTConnect.md) — config keys, capability surface, data-plane precedence rules, quality-code mapping, and (most load-bearing) the **Known limitations** section. - **Scope shipped:** `IDriver`+`ITagDiscovery`+`IReadable`+`ISubscribable`+`IHostConnectivityProbe`+`IRediscoverable` (P1 shipped deliberately without `IWritable`; **`IWritable` added 2026-08-05, `b29623cf`** — see the Deferred line below), browse **free via the Wave-0 universal browser** (`SupportsOnlineDiscovery=true`, no browser project), typed AdminUI tag editor, `UNAVAILABLE→BadNoCommunication` mapping, CONDITION-as-`String`, ring-buffer re-baseline paging (both the sequence-gap and `OUT_OF_RANGE` legs). - **Diverged from the design:** Task 0/6/7 dropped the planned TrakHound `MTConnect.NET-Common`/`-HTTP` dependency entirely — the pinned packages ship no XML formatter and the HTTP stream client has no injectable handler to test behind the driver's seam. Parsing is hand-rolled `System.Xml.Linq`, matching on `LocalName` only (namespace-agnostic, so 1.3–2.x all parse). See the driver guide's "Built vs. planned" section. - **Test results:** MTConnect unit suite 491/491; integration suite 12/12 against a real Agent (`mtconnect/agent:2.7.0.12`, **not** `mtconnect/cppagent` — that Docker Hub repo doesn't exist); skips cleanly (12 Skipped) offline. AdminUI 749/749; full solution 0 build errors. - **Write-back — SHIPPED 2026-08-05, not via Interfaces** (`b29623cf`, branch `feature/mqtt-mtconnect-write`; design [`2026-08-05-mtconnect-writeback-design.md`](2026-08-05-mtconnect-writeback-design.md)). `IWritable` over the Agent's optional **`AllowPut`** surface (`POST {agent}/{device}?{dataItemId}={value}`) — **observation injection into the Agent's buffer, never a machine command**, gated off by default at both the Agent (`AllowPut`) and the driver (`Write.Enabled`), CONDITION items refused. **MTConnect *Interfaces* remains deferred** — design §3.6's rationale still holds for it (rare, adapter-side cooperation, a state machine rather than a value write), and it is a different mechanism, not this one. Caveat recorded in the driver guide: the OPC UA `CurrentWrite` bit still comes from the authored `Tag.AccessLevel` alone, so an operator must author **ReadWrite**. - **Deferred (documented, not built):** MTConnect *Interfaces* (design §3.6, above); P1.5 CONDITION→native Part-9 alarms, `TIME_SERIES` array materialization, EVENT→enum vocab (design §9); P2 SHDR adapter ingest (design §9). Also two pre-existing fleet-wide gaps this build surfaced rather than caused: `IRediscoverable`/`IHostConnectivityProbe` have no server-side consumer (eight drivers besides Galaxy), and no driver form blocks Save on a required-field validation error. - **Fixture:** canned XML unit fixtures (bulk of coverage) + a dockerized `mtconnect/agent` + stdlib-SHDR-adapter integration fixture (env-gated), deployed at `/opt/otopcua-mtconnect/` on the shared docker host. See the fixture's own [`Docker/README.md`](../../tests/Drivers/ZB.MOM.WW.OtOpcUa.Driver.MTConnect.IntegrationTests/Docker/README.md). - **Effort:** S–M, hand-rolled (the TrakHound estimate in the design doc did not apply once that path was ruled out). ### MQTT / Sparkplug B — ✅ **COMPLETE** (P1 + P2, both live-gated) - **Design:** [`2026-07-15-mqtt-sparkplug-driver-design.md`](2026-07-15-mqtt-sparkplug-driver-design.md) - **Implementation plan:** [`2026-07-24-mqtt-sparkplug-driver.md`](2026-07-24-mqtt-sparkplug-driver.md) · [`.tasks.json`](2026-07-24-mqtt-sparkplug-driver.md.tasks.json) — **27 tasks, all done** (P1 plain MQTT = Tasks 0–14; P2 Sparkplug B = Tasks 15–26). - **Scope delivered:** P1 plain MQTT (MQTTnet-5 connect/TLS/auth + hand-rolled reconnect, subscribe→ `OnDataChange`, retained last-value read, `#`-observation browser); P2 Sparkplug B ingest (vendored Tahu proto + `Grpc.Tools` codegen, birth/alias/seq-gap/rebirth state machine, death→STALE, birth-driven browse tree, scoped Request-rebirth NCMD, typed tag editor). **P3 write-through (`IWritable`) shipped 2026-07-31** — `72be1922` ([#508](https://gitea.dohertylan.com/dohertj2/lmxopcua/issues/508), live-gated): plain `writeTopic` publish + Sparkplug NCMD/DCMD, with a node discovered `ViewOnly` when the encoder would refuse the write. Superseded the "every MQTT node materializes read-only" line this entry carried; see `docs/drivers/Mqtt.md` §Write-through. - **Fixture:** Mosquitto TLS+auth broker + JSON publisher sidecar, **live** at `10.100.0.35:8883` (`:1883` plaintext-but-authenticated); stack `/opt/otopcua-mqtt`, compose in `tests/Drivers/…​.Driver.Mqtt.IntegrationTests/Docker/`. Env-gated live suite (`MQTT_FIXTURE_ENDPOINT`, see `infra/README.md` §3). The **project-owned C# Sparkplug edge-node simulator** (`--profile sparkplug`, group `OtOpcUaSim`, nodes `EdgeA`/`EdgeB`, `EdgeA` device `Filler1`) shipped with P2 and answers rebirth NCMDs. - **P2 status (2026-07-25):** Tasks 15–26 complete. **Offline 1528 passed / 0 failed** (581 driver · 809 AdminUI · 138 Core.Abstractions); **live 15/15** against the broker + simulator. - **P1 live gate found two AdminUI authoring gaps** (the gate's whole point — both invisible to green unit tests, because AdminUI has no bUnit and nothing tied these hand-maintained lists to `DriverTypeNames`): 1. **FIXED —** `RawDriverTypeDialog`'s option array never got an MQTT row, so **no operator could create an MQTT driver at all**. Fixed, plus a reflection parity guard (`RawDriverTypeDialogParityTests`) that fails for *any* future `DriverTypeNames` entry missing from the picker. 2. **FIXED —** `MqttDriverForm` / `MqttDeviceForm` did not exist, so the broker connection was unauthorable from the AdminUI. Both shipped after the P1 gate. - **P2 live gate found a third one, of the same family — FIXED:** `MqttDriverForm` still carried its P1 Sparkplug **placeholder**. Switching Mode to `SparkplugB` rendered a *"not available yet"* notice and **no Group ID field** — so once Sparkplug ingest shipped there was still **no way to author a Sparkplug driver from the AdminUI**, and `Sparkplug.GroupId` is the driver's entire subscription filter (`spBv1.0/{GroupId}/#`): blank ⇒ connected, `Healthy`, ingesting nothing. Fixed (all five Sparkplug keys, merge-not-replace on save, group-id segment validation) + 8 falsifiable `MqttDriverFormModelTests` cases. **Three gates, three instances of the same defect class: a hand-maintained AdminUI surface left behind by a driver-side feature.** - **P2 live gate — four UI gaps recorded, ALL FOUR NOW CLOSED 2026-07-31** (`7f53bd96`; see `docs/drivers/Mqtt.md` §Known gaps, which lists the closures). Recorded as found, with what actually fixed each: - **A completely empty browse tree cannot arm a rebirth** ([#514](https://gitea.dohertylan.com/dohertj2/lmxopcua/issues/514)). The scope comes from a tree click, so on a plant that has not birthed since the window opened `Request rebirth…` stays disabled — the very case the affordance exists for. `MqttBrowseSession.ResolveRebirthTargets` already accepts a bare `{group}/{edgeNode}` ("the prime rebirth target"); only a UI path to enter one was missing. Mitigation shipped at the time: a **Refresh** button on the browse tree (it re-reads the same session, which previously rendered exactly once at open and never again). **FIXED —** `RawBrowseModal` now offers a `group/edgeNode` scope input whenever no tree scope is armed; edge-node only, never a typed group broadcast. - **A metric name containing `/` cannot be browse-committed** ([#509](https://gitea.dohertylan.com/dohertj2/lmxopcua/issues/509)). The derived raw **tag Name** is a RawPath segment and may not contain `/`, while `metricName` legitimately may — so the spec-mandatory `Node Control/Rebirth` was refused at commit (`Row N: Name must not contain '/'`). The refusal was loud and all-or-nothing, never a silently mis-bound tag, and **Manual entry** was the working path. Fixing it needed a name-sanitisation policy with a collision answer, so it was recorded rather than guessed at. **FIXED —** `RawBrowseCommitMapper.ToRawSegment` sanitises the segment and refuses a collision explicitly. - **`_canRebirth` is captured at browse-open**, so a reaped session still drew the button ([#510](https://gitea.dohertylan.com/dohertj2/lmxopcua/issues/510)). **FIXED.** - **The tag editor injects the Plain-only `payloadFormat` key into a Sparkplug blob** ([#513](https://gitea.dohertylan.com/dohertj2/lmxopcua/issues/513)) — cosmetic; the factory routes on the Sparkplug tuple and ignored it. **FIXED.** - **P2 live gate also found a PRE-EXISTING, cross-driver AdminUI defect — FIXED:** every node label in the shared `DriverBrowseTree` was an ``. In a Blazor Web App, `blazor.web.js`'s enhanced-navigation click interceptor resolves a bare `#` against ``, so **clicking any browse-tree node label navigated the whole AdminUI to `/`**, tore down the circuit and destroyed the hosting modal — losing the browse session and the tag selection. `@onclick:preventDefault` does not help: it suppresses the browser's default action, not Blazor's own interceptor. The labels are now `