NODEDC_MISSION_CORE/docs/07_MISSION_CORE_MONOREPO.md

17 KiB

NODEDC MISSION CORE monorepo

This document maps the current repository to the target Mission Core architecture without claiming that deferred package and process boundaries already exist. The device-lifecycle decision is recorded in ADR 0004, and the bounded laboratory runtime seam in ADR 0011.

Current layout

Path Current responsibility Target responsibility
apps/control-station/ React shell, model catalog, plugin host slots, spatial scene Vendor-neutral operator application
apps/control-station/src/core/device-plugins/frontendSdk.ts Public React host surface for reviewed frontend contributions Versioned frontend Plugin SDK package surface
packages/plugin-sdk/ Installable v0alpha2 Pydantic contracts and JSON Schema export Portable host/plugin identity, lifecycle, stream and evidence boundary
plugins/xgrids-k1/ v1alpha2 manifest, exact profile/loader and plugin-owned React connection/acquisition UI Independently versioned XGRIDS device plugin
plugins/xgrids-k1/frontend/ XGRIDS provisioning, project/acquisition/replay, optional spatial controls, diagnostics, metrics, runtime mapping and scoped CSS Plugin-owned reviewed frontend contribution
src/k1link/web/ Local FastAPI host, fail-closed runtime handshake/transport seam, in-memory operation/acquisition lifecycle and generic session API Generic host APIs plus isolated plugin supervisor
src/k1link/device_plugins/xgrids_k1/ Physically isolated K1 BLE/MQTT/protobuf/LZ4/camera/replay/CLI compatibility implementation Independently built XGRIDS device plugin process
src/k1link/data_plane/ Transport-neutral decoded in-process consumer views Local projections hydrated from portable SDK envelopes
src/k1link/viewer/ Vendor-neutral Rerun consumer, metrics and recorded blueprint Replaceable canonical scene sink and presentation adapters
src/k1link/sessions/ Host-owned SQLite catalog, evidence discovery/recovery, background preparation and derived-cache lifecycle Device-neutral observation archive service
src/k1link/web/session_api.py Opaque saved-session, immutable RRD/media and workspace-layout API Versioned Control/Edge observation contract
apps/control-station/src/core/observation/ Session selection, replay admission, recorded-source and layout state Device-neutral observation UI runtime
docs/domain-model/ Versioned experimental Mission Core vocabulary Promotion source for meanings proven across devices/services

Implemented boundary

The current K1 live/replay data path is:

raw K1 transport
  -> raw-first evidence capture
  -> injected K1 ModelingReport observer (device time/distance/speed)
  -> bounded four-message visual preview queue
  -> XGRIDS normalizer injected by the plugin facade
  -> DecodedPointCloudView / DecodedPoseView
  -> Rerun bridge
  -> embedded local viewer

The host-owned recorded path is separate from the disposable live preview:

sealed or recovered native observation session
  -> SQLite catalog and bounded single-worker preparation
  -> capture-clock-bound atomic RRD cache v9 + immutable recorded-media manifest v2
  -> generation-bound same-origin HTTP
  -> aggregate RRD/camera admission
  -> native Rerun HTTP receiver + recorded fMP4 player

New K1 source generations durably publish captures/mqtt_live/mqtt.timeline.origin.json before camera production. MQTT shutdown adds provisional transport envelope mqtt.timeline.json; after camera and MQTT/runtime shutdown, the acquisition owner publishes content-addressed mqtt.timeline.session-<sha256>.json and atomically switches the capture-summary pointer while still holding the session lease. Only that session scope is eligible to advertise combined media replay. Cache v9 materializes both session envelope endpoints as real RRD rows and logs archived device-reported scan time, route distance and speed as scalar time series. A crash-recovered origin-only point/pose prefix falls back to its last validated message and cannot advertise camera media. Older valid sessions without the clock contract keep their first/last-message fallback and cannot acquire camera evidence retroactively.

Rerun does not import K1 protocol code, inspect MQTT topics or receive raw payloads. VisualizationRuntime cannot choose a vendor decoder implicitly; its composition owner injects one. The local Decoded*View classes are in-process consumer projections, not the portable Plugin SDK wire envelopes.

The Plugin SDK v0alpha2 separately defines model/device/session identity, operation events, acquisition-related session state, canonical stream envelopes, payload handles, evidence lineage and compatibility assessments. It is importable through the root editable path dependency and independently buildable from packages/plugin-sdk. A runtime-owned descriptor must match the manifest ID, version, host API and exact action set, then complete the plugin-runtime/v0alpha1 handshake before its lifecycle health becomes ready. Backend actions instantiate immutable SDK RuntimeActionInvocation and RuntimeActionResult envelopes through the replaceable transport seam, and the host rejects uncorrelated results. The current transport is still in-process. The live spatial path does not yet instantiate portable SDK stream envelopes or use the SDK EvidenceStore protocol.

Frontend device workflows follow the same ownership rule. The generic host selects a manifest model and mounts its device.connection component. Concrete XGRIDS source lives under plugins/xgrids-k1/frontend, imports the host only via @mission-core/plugin-sdk, and is connected by the single reviewed import in apps/control-station/src/composition/devicePlugins.ts. BLE/Wi-Fi provisioning, exact-profile confirmation, project metadata, acquisition/replay UI and the optional K1 spatial-scene control block are not Core components. The host owns only the optional slot contract and generic actions such as opening the scene or releasing a prior manual/archive selection after a plugin source successfully starts.

A generic fail-closed guard blocks manual source input/apply/reset, saved-session switching and persisted replay reattach whenever the runtime reports a nonterminal acquisition. Unknown future acquisition states block by default. A plugin clears the previous host-selected source only after its start returns success, preserving the current scene on failure. That internal success-result transition is distinct from an operator switch. The guard does not stop equipment or invent a lifecycle transition; the acquisition must finish through its owning plugin.

Experimental vocabulary, not Platform Ontology

The local vocabulary distinguishes:

model != device != transport alias != device session
acquisition != operation != source/channel != evidence
observation session != preparation job != replay launch != viewer instance
workspace layout != sensor evidence
acknowledgement != completion
raw evidence != decoded data != viewer state

These meanings remain versioned under docs/domain-model/ and do not mutate or create a runtime dependency on NODE.DC Platform Ontology. Promotion requires a second device family or another platform service using the same stable meaning, an owner, lifecycle, compatibility policy and migration tests.

Plugin manifest and compatibility

The catalog accepts existing v1alpha1 manifests and additive v1alpha2 manifests. v1alpha1 remains a one-model contract; v1alpha2 accepts one or more models and requires every model to be covered by a reviewed, plugin-local, path-confined compatibility profile. The current XGRIDS profile matches exactly:

  • XGRIDS LixelKity K1;
  • firmware 3.0.2;
  • direct-LAN topology;
  • one retained physical laboratory evidence scope.

Unknown or unattested firmware fails closed. The current runtime does not read firmware from the device: it activates this limited, read-only profile only after explicit operator attestation of firmware 3.0.2 and direct-LAN topology, and records the basis as operator-attested. That claim is not device-derived proof. The profile independently records observation, decode, replay, physical-verification and write evidence. Loading it cannot authorize a transport mutation. BLE Wi-Fi provisioning and application control retain separate explicit operator gates.

An exact K1 application-control substrate now lives inside the vendor plugin. It byte-matches the retained ten-request pre-START dialogue plus START/STOP, correlates live DeviceInfo/modeling responses and maps bounded device-status values into a fail-closed state machine. The OpenAPI value is one private application-level authority; vendor ID and serial come from the BLE-selected live K1 and are cross-checked against status. Plugin v0.5.0 owns a single interactive MQTT session from connection through STOP/standby, with one explicit UI action per recorded lifecycle boundary and no automatic retry. A fixed Keychain loader and dormant shadow coordinator remain present. Stable physical proof after STOP remains unresolved; write capability is exposed only while that interactive socket owner is active.

Semantic lifecycle

The transitional facade now creates separate provisional device, device-session, acquisition and operation IDs. Acquisition can be prepared, wait for receiver readiness, then wait for an external physical start, and become acquiring only after real point data, wait for an externally confirmed stop, finalize and complete. Capture-only stop reports the K1 physical state as unknown.

Preparation requires a project name. The frontend and backend both apply NFKC normalization plus surrounding-whitespace trimming, reject control/surrogate characters and names above 96 Unicode characters, and store the result as session/catalog display metadata. It is not a path component. Because command publishing is off, the current integration does not claim that the name reaches K1.

The K1 contribution may mount a SpatialControlsView beside the host-owned viewport. It presents plugin lifecycle wording, the local stop action and live ModelingReport scan time, route distance and speed. Under the current capture-only profile the stop action seals local reception; it does not claim a device stop. ModelingReport is consumed before the visual preview queue through an injected observer, preserving the generic normalizer/Rerun boundary.

The operation journal is bounded and in memory. It records IDs, idempotency, declared deadlines, progress and terminal results without action parameters or secrets. It is not durable, distributed or recoverable after process restart.

Semantic acquisition actions coexist with legacy stream.* actions for v1alpha1/UI compatibility. On this profile, start and stop operate the local receiver and operator workflow only. K1 scanning is still started and stopped by physical double-click; no modeling request is published.

Left/right camera preview transport, endpoint paths and H.264 framing are now observed under the exact compatibility profile. The local read-only adapter copy-remuxes one selected RTSP producer into bounded fMP4/WebSocket delivery for the generic MSE UI. New acquisitions archive selected-camera init/segments/index independently of browser delivery. The generic recorded player, manifest-v2 validation and shared session_time controls are connected; historical sessions created before this archive contract contain no recoverable video, so physical recorded-camera acceptance remains open. Portable FFmpeg packaging, disk-backed browser buffering, fan-out and remote delivery also remain open. Device calibration command and sensor-to-vehicle extrinsics are unavailable.

The 2026-07-17 physical acceptance gate confirmed continuously updating point clouds, a matching live trajectory, and both camera selections in the same Mission Core acquisition. The embedded Rerun blueprint owns live-edge following on stream_time; the React shell selects the timeline once and does not drive it with a timer. Raw captures, camera frames, device identity and network details remain outside Git. The bounded acquisition subset of DeviceStatusReport and ModelingReport route telemetry are decoded for the exact profile; heartbeat and nested system/RTK status payloads remain raw observed evidence. See ADR 0007 for the measured gate and remaining limits.

Remaining extraction order

Native replay parity, exact-profile physical point/pose regression, immutable SDK runtime-action envelopes, fail-closed runtime handshake and physical extraction of K1 transports/codecs are complete. Portable SDK stream/evidence envelopes remain separate from the allocation-conscious in-process preview views.

  1. When a real deployment or second execution target requires it, implement a subprocess transport behind the current seam, then add heartbeat timeout, restart policy and resource limits as separately accepted behavior.
  2. Replace compatibility routes and singleton state with multi-device session routing.
  3. Split Edge execution from the Control Station behind authenticated transport.
  4. Physically accept a newly archived left/right K1 session, then package the read-only RTSP/H.264 adapter for each target OS, add disk-backed sealed media caching and evolve same-host MSE delivery toward an authenticated Edge media plane. Keep the modeling-command publisher disabled until operator-owned Keychain item provisioning, physical acceptance of the reviewed uninstalled transport and durable-save confirmation gate.

Complex equipment will likely be assembled from separately useful component plugins into configured hardware packs, while standalone equipment remains available for engineering work. That direction is intentionally not a current domain schema: pack ownership, roles, compatibility and lifecycle must be derived from real additional component families rather than inferred from the single K1 vertical.

Invariants

  • Raw bytes are appended and OS-flushed before preview work; aligned raw and metadata become durable together at the bounded group-commit boundary and are never overwritten by a decode result.
  • Mission Core starts when the XGRIDS manifest is absent. Formal installed plugin quarantine remains a future supervisor feature.
  • Core code does not branch on XGRIDS IDs, fields, topics or firmware.
  • A plugin cannot add global navigation or arbitrary CSS.
  • A transport alias never becomes a stable device identity merely because it appears in a vendor header.
  • Acknowledgement, first data, operator confirmation and physical completion remain distinct evidence levels.
  • Vendor writes require a declared capability, exact compatibility evidence, host authorization and an operator gate. Manifest metadata and profiles alone grant none of these.
  • The visualization engine can be replaced without changing vendor transports or codecs.
  • Real captures, credentials, device identities, maps and router metadata remain outside Git.

Deferred, not implemented

  • isolated/signed plugin processes and crash containment;
  • durable operation journal, restart recovery, vault and RBAC;
  • persistent device identity and concurrent multi-device routing;
  • complete SDK-envelope/EvidenceStore hot-path integration;
  • remote Edge split, authenticated WAN relay and fleet orchestration;
  • automatic K1 start/stop/calibration commands;
  • device-reported camera capability discovery beyond the reviewed K1 profile;
  • physical acceptance of shared-timeline point-cloud plus one selected-camera playback on a newly archived K1 session;
  • frame-accurate camera/LiDAR calibration, panoramic stitching, disk-backed browser media cache and remote multi-consumer delivery;
  • production retention, replication, encryption and long-run WebViewer/WASM memory acceptance for large observation sessions.

Workspace-repeatable locked architecture gate

From the repository root:

uv sync --frozen --group dev
uv run pytest \
  tests/test_plugin_sdk_v0alpha2_contracts.py \
  tests/test_plugin_runtime.py \
  tests/test_plugin_catalog.py \
  tests/test_xgrids_compatibility_profile.py \
  tests/test_device_lifecycle.py \
  tests/test_xgrids_acquisition_lifecycle.py \
  tests/test_canonical_pipeline.py \
  tests/test_rerun_bridge.py \
  tests/test_web_validation_security.py
uv run ruff check .
uv run mypy
uv run mypy --strict packages/plugin-sdk/python/missioncore_plugin_sdk
uv run python plugins/xgrids-k1/profile_loader.py
uv run python -m missioncore_plugin_sdk.v0alpha2 \
  > /tmp/missioncore-plugin-sdk-v0alpha2.schemas.json

cd apps/control-station
npm ci
npm run test:unit
npm run typecheck
npm run build

These checks perform no BLE scan, network capture, router mutation or K1 application write. The next physical and communication gates are specified in ADR 0004.

This gate is repeatable in the current workspace, not standalone reproducible. The frontend consumes mutable sibling file: dependencies from NODEDC_DESIGN_GUIDELINE; its lockfile does not pin the donor checkout's Git revision or content hash. CI and portable packaging remain blocked until those packages are published or vendored, or an immutable donor revision and content verification are enforced. The current root Python wheel also omits the plugin manifests/profile loader and built frontend, so it is not a deployment bundle.