ui(raw): surface WiFi-CSI breathing + heart rate pills

ADR-021 already publishes `vital_signs` inside SensingUpdate but the
raw calibration console had no readout — the operator had to curl
/api/v1/vital-signs to see breathing/HR. Add two pills (🫁 + 💓)
next to the mmWave one and update them on every WS tick.

Confidence < 20 % dims the pill so noise-floor estimates don't read
as real values. Missing/zero rates fall back to "— BPM".

Mirrored ui/raw.html → static/raw.html so both deployment paths
serve the same console.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

.gitignore

@@ -258,3 +258,12 @@ v2/crates/rvcsi-node/*.node
 v2/crates/rvcsi-node/binding.js
 v2/crates/rvcsi-node/binding.d.ts
 v2/crates/rvcsi-node/npm/
+
+# ADR-117/118/119/120: deployment-specific training artifacts.
+# - recordings/ are large (175 MB each) and room/operator-specific
+# - adaptive_model.json is regenerated by `POST /api/v1/adaptive/train`
+# - baseline.json is regenerated by `POST /api/v1/baseline/calibrate`
+v2/data/recordings/
+v2/data/adaptive_model.json
+v2/data/baseline.json
+.claude/launch.json

CHECKLIST.md

@@ -5,15 +5,25 @@ at the end of every session. Pair with
 [`docs/references/espectre-gap-analysis.md`](docs/references/espectre-gap-analysis.md)
 for the technical detail behind each line.
 
-Last sweep: **2026-05-17**, branch `feat/ota-rssi-mobile`, head `0ec1e4b0`.
-Status: 47 Done / 0 Open in-scope. Deferred items (out of session scope,
-each with explicit reason) listed at the bottom.
+Last sweep: **2026-05-18**, branch `feat/ota-rssi-mobile`, head `12e1cf9d`.
+Status: **50 Done / 0 Open in-scope**. Deferred listed at the bottom.
 
-This count includes the ADR-100..114 carry-in from the prior agent + this
-session's ADR-115 (FW set-target REST), ADR-116 (WiFlow-v1 Rust loader),
-ADR-116 cosmetic (UI dropdown), and ADR-117 (process hygiene + audit
-follow-ups). ADR-111 is intentionally absent (folded into ADR-109 during
-the AP-MAC tracking work).
+ADR-100..114 carry-in from the prior agent; ADR-115..120 are this
+session. ADR-111 intentionally absent (folded into ADR-109).
+
+Adaptive classifier accuracy trajectory this session — full detail in
+ADR-118/119/120:
+
+| Stage | Acc |
+|---|---|
+| 2-node, 15-feat LogReg (baseline) | 40.4% |
+| 6-node, 15-feat LogReg | 44.4% |
+| 6-node, 22-feat LogReg (ADR-118) | 49.58% |
+| 6-node, 22-feat MLP (ADR-119) | 53.53% |
+| 6-node, 22-feat W-MLP (ADR-120) | **90.40%** (training-set) |
+
+W-MLP 90.40% is training-set accuracy; held-out test + cleaner
+per-class re-records are the recommended next step.
 
 ---
 
@@ -91,6 +101,19 @@ the AP-MAC tracking work).
       runtime classifier; sensing tab container restored; multi-node
       test guards external :5005; docs/typo/range sweep.
 
+### Adaptive Classifier (data pipeline + model)
+
+- [x] **ADR-118** Feature decorrelation + multi-node extractor (22 feats
+      = 4 global + 6 nodes × 3 with z-score). Accuracy 44.4% → 49.58%
+      (`e86f6506`).
+- [x] **ADR-119** Frame-level MLP (22→32→6 ReLU+softmax), manual
+      backprop, ~3k weights. Accuracy 49.58% → 53.53% (`94330708`).
+- [x] **ADR-120** Windowed temporal W-MLP (440→64→6, 20×22 stack) —
+      captures walking / sit-stand / gesture cadence. Accuracy 53.53%
+      → 90.40% training; held-out TBD. Hybrid priority (rule-based owns
+      4 base, W-MLP owns waving/transition) + two-layer label smoothing
+      + `/api/v1/adaptive/debug` (`da4c123d`..`12e1cf9d`, 7 commits).
+
 ### Tests / fixtures
 
 - [x] **ADR-114** `tests/fixtures/replay_idle.jsonl` +
@@ -112,7 +135,7 @@ the AP-MAC tracking work).
 
 ### Documentation
 
-- [x] **ADR-100..117** all written (ADR-111 intentionally absent), each ≤ 200 lines
+- [x] **ADR-100..120** all written (ADR-111 intentionally absent), each ≤ 200 lines
 - [x] `docs/references/espectre-techniques.md` — Pace technique catalogue
 - [x] `docs/references/espectre-gap-analysis.md` — section-by-section gap
 - [x] Documentation actualization sweep — every Open Items section
@@ -151,26 +174,15 @@ ADR-113, see Done above)
 
 ### Deferred — out of session scope
 
-Marked here so future sessions don't re-litigate; each line carries
-an explicit reason. Bring them back only if scope changes.
+Each line carries an explicit reason; revisit if scope changes.
 
-- **HA via MQTT** — new integration. Excluded by current session brief
-  (no new integrations on current hardware).
-- **ESPHome native component** — same reason as HA/MQTT.
-- **Web Serial calibration game** — explicitly excluded.
-- **Boot-time NBVI freeze in FW** — explicitly excluded.
-- **Per-channel NVS cache for gain-lock** — explicitly excluded; only
-  matters if channel hopping is reactivated, which is also excluded.
-- **DensePose model train + load** — explicitly excluded.
-- **AETHER contrastive pretrain on live data** — explicitly excluded.
-- **MERIDIAN domain generalization** — explicitly excluded.
-- **Channel hopping (ADR-029)** — explicitly excluded.
-- **Multi-antenna support (`n_antennas` > 1)** — explicitly excluded.
-- **README.md trim (542 lines)** — explicitly excluded.
-- **CLAUDE.md trim (407 lines)** — explicitly excluded.
-- **Tailscale-target in NVS** — Mac stable on TP-Link this session,
-  low ROI. Not blocking. (ADR-100 follow-up; bring back if Mac
-  network swap becomes routine.)
+* New integrations (excluded by session brief): HA/MQTT, ESPHome,
+  Web Serial game, DensePose train, AETHER pretrain, MERIDIAN.
+* FW changes excluded: boot-time NBVI freeze, per-channel NVS cache
+  for gain-lock, channel hopping (ADR-029), multi-antenna support
+  (`n_antennas > 1`).
+* **Tailscale-target in NVS** — Mac stable on TP-Link this session;
+  low ROI. ADR-100 follow-up if Mac network swap becomes routine.
 
 ---
 
@@ -178,7 +190,7 @@ an explicit reason. Bring them back only if scope changes.
 
 | Doc | Purpose |
 |---|---|
-| [`docs/adr/`](docs/adr) | All ADRs 001-117 (111 absent); 100-117 are this session |
+| [`docs/adr/`](docs/adr) | All ADRs 001-120 (111 absent); 100-120 are this session |
 | [`docs/references/espectre-techniques.md`](docs/references/espectre-techniques.md) | Pace technique catalogue + RuView adoption |
 | [`docs/references/espectre-gap-analysis.md`](docs/references/espectre-gap-analysis.md) | Section-by-section gap with priority table |
 | [`docs/references/ota-pipeline.md`](docs/references/ota-pipeline.md) | OTA recipe — port 8032, three FW prereqs |

CLAUDE.md

@@ -1,223 +1,47 @@
-# Claude Code Configuration — WiFi-DensePose + Claude Flow V3
+# Claude Code Configuration — WiFi-DensePose
 
 ## Project: wifi-densepose
 
-WiFi-based human pose estimation using Channel State Information (CSI).
-Dual codebase: Python v1 (`v1/`) and Rust port (`v2/`).
-### Key Rust Crates
-| Crate | Description |
-|-------|-------------|
-| `wifi-densepose-core` | Core types, traits, error types, CSI frame primitives |
-| `wifi-densepose-signal` | SOTA signal processing + RuvSense multistatic sensing (14 modules) |
-| `wifi-densepose-nn` | Neural network inference (ONNX, PyTorch, Candle backends) |
-| `wifi-densepose-train` | Training pipeline with ruvector integration + ruview_metrics |
-| `wifi-densepose-mat` | Mass Casualty Assessment Tool — disaster survivor detection |
-| `wifi-densepose-hardware` | ESP32 aggregator, TDM protocol, channel hopping firmware |
-| `wifi-densepose-ruvector` | RuVector v2.0.4 integration + cross-viewpoint fusion (5 modules) |
-| `wifi-densepose-api` | REST API (Axum) |
-| `wifi-densepose-db` | Database layer (Postgres, SQLite, Redis) |
-| `wifi-densepose-config` | Configuration management |
-| `wifi-densepose-wasm` | WebAssembly bindings for browser deployment |
-| `wifi-densepose-cli` | CLI tool (`wifi-densepose` binary) |
-| `wifi-densepose-sensing-server` | Lightweight Axum server for WiFi sensing UI |
-| `wifi-densepose-wifiscan` | Multi-BSSID WiFi scanning (ADR-022) |
-| `wifi-densepose-vitals` | ESP32 CSI-grade vital sign extraction (ADR-021) |
-| `nvsim` | Deterministic NV-diamond magnetometer pipeline simulator (ADR-089) — standalone leaf, WASM-ready |
-| `vendor/rvcsi` (submodule) | **rvCSI** — edge RF sensing runtime (ADR-095/096): 9 crates (`rvcsi-core`/`-dsp`/`-events`/`-adapter-file`/`-adapter-nexmon`/`-ruvector`/`-runtime`/`-node`/`-cli`). Lives in its own repo ([github.com/ruvnet/rvcsi](https://github.com/ruvnet/rvcsi)), vendored here under `vendor/rvcsi`, published to crates.io as `rvcsi-* 0.3.x` and to npm as `@ruv/rvcsi`. Not a `v2/` workspace member — depend on the published crates (or the submodule's `crates/rvcsi-*` paths). Normalized `CsiFrame`/`CsiWindow`/`CsiEvent` schema, validate-before-FFI, reusable DSP, typed confidence-scored events, the napi-c Nexmon shim (real nexmon_csi `.pcap` from a Raspberry Pi 5 / 4 / 3B+ — BCM43455c0), the napi-rs SDK, the `rvcsi` CLI, a Claude Code plugin. |
+WiFi-based human pose estimation from Channel State Information (CSI).
+Dual codebase: Python v1 (`archive/v1/`) and Rust port (`v2/`).
 
-### RuvSense Modules (`signal/src/ruvsense/`)
-| Module | Purpose |
-|--------|---------|
-| `multiband.rs` | Multi-band CSI frame fusion, cross-channel coherence |
-| `phase_align.rs` | Iterative LO phase offset estimation, circular mean |
-| `multistatic.rs` | Attention-weighted fusion, geometric diversity |
-| `coherence.rs` | Z-score coherence scoring, DriftProfile |
-| `coherence_gate.rs` | Accept/PredictOnly/Reject/Recalibrate gate decisions |
-| `pose_tracker.rs` | 17-keypoint Kalman tracker with AETHER re-ID embeddings |
-| `field_model.rs` | SVD room eigenstructure, perturbation extraction |
-| `tomography.rs` | RF tomography, ISTA L1 solver, voxel grid |
-| `longitudinal.rs` | Welford stats, biomechanics drift detection |
-| `intention.rs` | Pre-movement lead signals (200-500ms) |
-| `cross_room.rs` | Environment fingerprinting, transition graph |
-| `gesture.rs` | DTW template matching gesture classifier |
-| `adversarial.rs` | Physically impossible signal detection, multi-link consistency |
+See **[`CHECKLIST.md`](CHECKLIST.md)** for current implementation status
+(50 Done / 0 Open in-scope; ADR-100..120 are this operational session).
 
-### Cross-Viewpoint Fusion (`ruvector/src/viewpoint/`)
-| Module | Purpose |
-|--------|---------|
-| `attention.rs` | CrossViewpointAttention, GeometricBias, softmax with G_bias |
-| `geometry.rs` | GeometricDiversityIndex, Cramer-Rao bounds, Fisher Information |
-| `coherence.rs` | Phase phasor coherence, hysteresis gate |
-| `fusion.rs` | MultistaticArray aggregate root, domain events |
+### Detailed handbooks (extracted to keep this file ≤200 lines)
 
-### RuVector v2.0.4 Integration (ADR-016 complete, ADR-017 proposed)
-All 5 ruvector crates integrated in workspace:
-- `ruvector-mincut` → `metrics.rs` (DynamicPersonMatcher) + `subcarrier_selection.rs`
-- `ruvector-attn-mincut` → `model.rs` (apply_antenna_attention) + `spectrogram.rs`
-- `ruvector-temporal-tensor` → `dataset.rs` (CompressedCsiBuffer) + `breathing.rs`
-- `ruvector-solver` → `subcarrier.rs` (sparse interpolation 114→56) + `triangulation.rs`
-- `ruvector-attention` → `model.rs` (apply_spatial_attention) + `bvp.rs`
-
-### Architecture Decisions
-43 ADRs in `docs/adr/` (ADR-001 through ADR-043). Key ones:
-- ADR-014: SOTA signal processing (Accepted)
-- ADR-015: MM-Fi + Wi-Pose training datasets (Accepted)
-- ADR-016: RuVector training pipeline integration (Accepted — complete)
-- ADR-017: RuVector signal + MAT integration (Proposed — next target)
-- ADR-024: Contrastive CSI embedding / AETHER (Accepted)
-- ADR-027: Cross-environment domain generalization / MERIDIAN (Accepted)
-- ADR-028: ESP32 capability audit + witness verification (Accepted)
-- ADR-029: RuvSense multistatic sensing mode (Proposed)
-- ADR-030: RuvSense persistent field model (Proposed)
-- ADR-031: RuView sensing-first RF mode (Proposed)
-- ADR-032: Multistatic mesh security hardening (Proposed)
-
-### Supported Hardware
-
-| Device | Port | Chip | Role | Cost |
-|--------|------|------|------|------|
-| ESP32-S3 (8MB flash) | COM7 | Xtensa dual-core | WiFi CSI sensing node | ~$9 |
-| ESP32-S3 SuperMini (4MB) | — | Xtensa dual-core | WiFi CSI (compact) | ~$6 |
-| ESP32-C6 + Seeed MR60BHA2 | COM4 | RISC-V + 60 GHz FMCW | mmWave HR/BR/presence | ~$15 |
-| HLK-LD2410 | — | 24 GHz FMCW | Presence + distance | ~$3 |
-
-**Not supported:** ESP32 (original), ESP32-C3 — single-core, can't run CSI DSP pipeline.
-
-### Build & Test Commands (this repo)
-```bash
-# Rust — full workspace tests (1,031+ tests, ~2 min)
-cd v2
-cargo test --workspace --no-default-features
-
-# Rust — single crate check (no GPU needed)
-cargo check -p wifi-densepose-train --no-default-features
-
-# Python — deterministic proof verification (SHA-256)
-python archive/v1/data/proof/verify.py
-
-# Python — test suite
-cd archive/v1 && python -m pytest tests/ -x -q
-```
-
-### ESP32 Firmware Build (Windows — Python subprocess required)
-```bash
-# Build 8MB firmware (real WiFi CSI mode, no mocks)
-# See CLAUDE.local.md for the full Python subprocess command
-# Key: must strip MSYSTEM env vars for ESP-IDF v5.4 on Git Bash
-
-# Build 4MB firmware
-cp sdkconfig.defaults.4mb sdkconfig.defaults
-# then same build process
-
-# Flash to COM7
-# [python, idf_py, '-p', 'COM7', 'flash']
-
-# Provision WiFi
-python firmware/esp32-csi-node/provision.py --port COM7 \
-  --ssid "YourWiFi" --password "secret" --target-ip 192.168.1.20
-
-# Monitor serial
-python -m serial.tools.miniterm COM7 115200
-```
-
-### Firmware Release Process
-1. Build 8MB from `sdkconfig.defaults.template` (no mock)
-2. Build 4MB from `sdkconfig.defaults.4mb` (no mock)
-3. Save 6 binaries: `esp32-csi-node.bin`, `bootloader.bin`, `partition-table.bin`, `ota_data_initial.bin`, `esp32-csi-node-4mb.bin`, `partition-table-4mb.bin`
-4. Tag: `git tag v0.X.Y-esp32 && git push origin v0.X.Y-esp32`
-5. Release: `gh release create v0.X.Y-esp32 <binaries> --title "..." --notes-file ...`
-6. Verify on real hardware (COM7) before publishing
-7. **CRITICAL:** Always test with real WiFi CSI, not mock mode — mock missed the Kconfig threshold bug
-
-### Crate Publishing Order
-Crates must be published in dependency order:
-1. `wifi-densepose-core` (no internal deps)
-2. `wifi-densepose-vitals` (no internal deps)
-3. `wifi-densepose-wifiscan` (no internal deps)
-4. `wifi-densepose-hardware` (no internal deps)
-5. `wifi-densepose-config` (no internal deps)
-6. `wifi-densepose-db` (no internal deps)
-7. `wifi-densepose-signal` (depends on core)
-8. `wifi-densepose-nn` (no internal deps, workspace only)
-9. `wifi-densepose-ruvector` (no internal deps, workspace only)
-10. `wifi-densepose-train` (depends on signal, nn)
-11. `wifi-densepose-mat` (depends on core, signal, nn)
-12. `wifi-densepose-api` (no internal deps)
-13. `wifi-densepose-wasm` (depends on mat)
-14. `wifi-densepose-sensing-server` (depends on wifiscan)
-15. `wifi-densepose-cli` (depends on mat)
-
-### Validation & Witness Verification (ADR-028)
-
-**After any significant code change, run the full validation:**
-
-```bash
-# 1. Rust tests — must be 1,031+ passed, 0 failed
-cd v2
-cargo test --workspace --no-default-features
-
-# 2. Python proof — must print VERDICT: PASS
-cd ..
-python archive/v1/data/proof/verify.py
-
-# 3. Generate witness bundle (includes both above + firmware hashes)
-bash scripts/generate-witness-bundle.sh
-
-# 4. Self-verify the bundle — must be 7/7 PASS
-cd dist/witness-bundle-ADR028-*/
-bash VERIFY.sh
-```
-
-**If the Python proof hash changes** (e.g., numpy/scipy version update):
-```bash
-# Regenerate the expected hash, then verify it passes
-python archive/v1/data/proof/verify.py --generate-hash
-python archive/v1/data/proof/verify.py
-```
-
-**Witness bundle contents** (`dist/witness-bundle-ADR028-<sha>.tar.gz`):
-- `WITNESS-LOG-028.md` — 33-row attestation matrix with evidence per capability
-- `ADR-028-esp32-capability-audit.md` — Full audit findings
-- `proof/verify.py` + `expected_features.sha256` — Deterministic pipeline proof
-- `test-results/rust-workspace-tests.log` — Full cargo test output
-- `firmware-manifest/source-hashes.txt` — SHA-256 of all 7 ESP32 firmware files
-- `crate-manifest/versions.txt` — All 15 crates with versions
-- `VERIFY.sh` — One-command self-verification for recipients
-
-**Key proof artifacts:**
-- `archive/v1/data/proof/verify.py` — Trust Kill Switch: feeds reference signal through production pipeline, hashes output
-- `archive/v1/data/proof/expected_features.sha256` — Published expected hash
-- `archive/v1/data/proof/sample_csi_data.json` — 1,000 synthetic CSI frames (seed=42)
-- `docs/WITNESS-LOG-028.md` — 11-step reproducible verification procedure
-- `docs/adr/ADR-028-esp32-capability-audit.md` — Complete audit record
-
-### Branch
-Default branch: `main`
-Active feature branch: `ruvsense-full-implementation` (PR #77)
+| File | Covers |
+|---|---|
+| [`docs/dev-handbook.md`](docs/dev-handbook.md) | Rust crate map (15 crates), RuvSense modules (14), Cross-Viewpoint fusion (5), Architecture Decisions list, supported hardware, build & test commands, ESP32 firmware build + provision, release process, crate publishing order, witness verification (ADR-028) |
+| [`docs/claude-swarm.md`](docs/claude-swarm.md) | V3 CLI commands, available agents (60+ types), memory commands reference, Claude Code vs CLI tools |
+| [`docs/architecture.md`](docs/architecture.md) | End-to-end pipeline diagram from CSI capture to pose / vital signs / room fingerprint |
+| [`docs/use-cases.md`](docs/use-cases.md) | Full deployment-tier catalogue (Everyday / Specialized / Robotics / Extreme) + all 60 ADR-041 edge modules + self-learning system (ADR-024) |
+| [`docs/adr/`](docs/adr/) | All 120 ADRs (ADR-111 intentionally absent); session-specific records ADR-100..120 |
 
 ---
 
 ## Behavioral Rules (Always Enforced)
 
 - Do what has been asked; nothing more, nothing less
-- NEVER create files unless they're absolutely necessary for achieving your goal
+- NEVER create files unless absolutely necessary for the goal
 - ALWAYS prefer editing an existing file to creating a new one
 - NEVER proactively create documentation files (*.md) or README files unless explicitly requested
-- NEVER save working files, text/mds, or tests to the root folder
-- Never continuously check status after spawning a swarm — wait for results
+- NEVER save working files, tests or markdown to the root folder
 - ALWAYS read a file before editing it
 - NEVER commit secrets, credentials, or .env files
+- Never continuously check status after spawning a swarm — wait for results
+- README.md and CLAUDE.md must each stay ≤ 200 lines; move detail to `docs/` and link
 
 ## File Organization
 
 - NEVER save to root folder — use the directories below
-- `docs/adr/` — Architecture Decision Records (43 ADRs)
+- `docs/adr/` — Architecture Decision Records (currently 120, ADR-111 absent)
 - `docs/ddd/` — Domain-Driven Design models
-- `v2/crates/` — Rust workspace crates (15 crates)
-- `v2/crates/wifi-densepose-signal/src/ruvsense/` — RuvSense multistatic modules (14 files)
-- `v2/crates/wifi-densepose-ruvector/src/viewpoint/` — Cross-viewpoint fusion (5 files)
+- `v2/crates/` — Rust workspace crates (15+ crates)
+- `v2/crates/wifi-densepose-signal/src/ruvsense/` — RuvSense multistatic modules
+- `v2/crates/wifi-densepose-ruvector/src/viewpoint/` — Cross-viewpoint fusion
 - `v2/crates/wifi-densepose-hardware/src/esp32/` — ESP32 TDM protocol
-- `firmware/esp32-csi-node/main/` — ESP32 C firmware (channel hopping, NVS config, TDM)
+- `firmware/esp32-csi-node/main/` — ESP32 C firmware (CSI capture, NVS config, OTA, channel hopping)
 - `archive/v1/src/` — Python source (core, hardware, services, api)
 - `archive/v1/data/proof/` — Deterministic CSI proof bundles
 - `.claude-flow/` — Claude Flow coordination state (committed for team sharing)
@@ -226,7 +50,7 @@ Active feature branch: `ruvsense-full-implementation` (PR #77)
 ## Project Architecture
 
 - Follow Domain-Driven Design with bounded contexts
-- Keep files under 500 lines
+- Keep files under 500 lines; ADRs ≤ 200 lines; README.md and CLAUDE.md ≤ 200 lines
 - Use typed interfaces for all public APIs
 - Prefer TDD London School (mock-first) for new code
 - Use event sourcing for state changes
@@ -244,39 +68,38 @@ Active feature branch: `ruvsense-full-implementation` (PR #77)
 
 Before merging any PR, verify each item applies and is addressed:
 
-1. **Rust tests pass** — `cargo test --workspace --no-default-features` (1,031+ passed, 0 failed)
+1. **Rust tests pass** — `cargo test --workspace --no-default-features`
 2. **Python proof passes** — `python archive/v1/data/proof/verify.py` (VERDICT: PASS)
-3. **README.md** — Update platform tables, crate descriptions, hardware tables, feature summaries if scope changed
-4. **CLAUDE.md** — Update crate table, ADR list, module tables, version if scope changed
-5. **CHANGELOG.md** — Add entry under `[Unreleased]` with what was added/fixed/changed
-6. **User guide** (`docs/user-guide.md`) — Update if new data sources, CLI flags, or setup steps were added
-7. **ADR index** — Update ADR count in README docs table if a new ADR was created
+3. **README.md** — Update if scope changed; verify ≤ 200 lines
+4. **CLAUDE.md** — Update if scope changed; verify ≤ 200 lines; move detail into `docs/`
+5. **CHANGELOG.md** — Add entry under `[Unreleased]`
+6. **User guide** (`docs/user-guide.md`) — Update if new data sources, CLI flags, or setup steps
+7. **ADR index** — Update ADR count + range in CHECKLIST and reference tables when a new ADR is created
 8. **Witness bundle** — Regenerate if tests or proof hash changed: `bash scripts/generate-witness-bundle.sh`
-9. **Docker Hub image** — Only rebuild if Dockerfile, dependencies, or runtime behavior changed
+9. **Docker Hub image** — Rebuild only if Dockerfile / dependencies / runtime behavior changed
 10. **Crate publishing** — Only needed if a crate is published to crates.io and its public API changed
-11. **`.gitignore`** — Add any new build artifacts or binaries
-12. **Security audit** — Run security review for new modules touching hardware/network boundaries
+11. **`.gitignore`** — Add any new build artifacts or large deployment-specific data files
+12. **Security audit** — Run a security review for new modules touching hardware/network boundaries
 
 ## Build & Test
 
 ```bash
-# Build
-npm run build
+# Rust — full workspace tests
+cargo test --workspace --no-default-features
 
-# Test
-npm test
-
-# Lint
-npm run lint
+# Python — deterministic proof
+python archive/v1/data/proof/verify.py
 ```
 
-- ALWAYS run tests after making code changes
+- ALWAYS run tests after code changes
 - ALWAYS verify build succeeds before committing
 
+Full per-crate commands and firmware flash recipe: **[`docs/dev-handbook.md`](docs/dev-handbook.md)**.
+
 ## Security Rules
 
 - NEVER hardcode API keys, secrets, or credentials in source files
-- NEVER commit .env files or any file containing secrets
+- NEVER commit `.env` files or any file containing secrets
 - Always validate user input at system boundaries
 - Always sanitize file paths to prevent directory traversal
 - Run `npx @claude-flow/cli@latest security scan` after security-related changes
@@ -294,114 +117,33 @@ npm run lint
 
 - MUST initialize the swarm using CLI tools when starting complex tasks
 - MUST spawn concurrent agents using Claude Code's Task tool
-- Never use CLI tools alone for execution — Task tool agents do the actual work
+- Never use CLI tools alone for execution — Task-tool agents do the actual work
 - MUST call CLI tools AND Task tool in ONE message for complex work
 
-### 3-Tier Model Routing (ADR-026)
-
-| Tier | Handler | Latency | Cost | Use Cases |
-|------|---------|---------|------|-----------|
-| **1** | Agent Booster (WASM) | <1ms | $0 | Simple transforms (var→const, add types) — Skip LLM |
-| **2** | Haiku | ~500ms | $0.0002 | Simple tasks, low complexity (<30%) |
-| **3** | Sonnet/Opus | 2-5s | $0.003-0.015 | Complex reasoning, architecture, security (>30%) |
-
-- Always check for `[AGENT_BOOSTER_AVAILABLE]` or `[TASK_MODEL_RECOMMENDATION]` before spawning agents
-- Use Edit tool directly when `[AGENT_BOOSTER_AVAILABLE]`
-
-## Swarm Configuration & Anti-Drift
-
-- ALWAYS use hierarchical topology for coding swarms
-- Keep maxAgents at 6-8 for tight coordination
-- Use specialized strategy for clear role boundaries
-- Use `raft` consensus for hive-mind (leader maintains authoritative state)
-- Run frequent checkpoints via `post-task` hooks
-- Keep shared memory namespace for all agents
-
-```bash
-npx @claude-flow/cli@latest swarm init --topology hierarchical --max-agents 8 --strategy specialized
-```
+Full CLI command reference, agent type catalogue, memory operations and
+3-tier model routing: **[`docs/claude-swarm.md`](docs/claude-swarm.md)**.
 
 ## Swarm Execution Rules
 
-- ALWAYS use `run_in_background: true` for all agent Task calls
-- ALWAYS put ALL agent Task calls in ONE message for parallel execution
-- After spawning, STOP — do NOT add more tool calls or check status
-- Never poll TaskOutput or check swarm status — trust agents to return
-- When agent results arrive, review ALL results before proceeding
+1. **Spawn in background** — use `run_in_background: true` for all agent Task calls
+2. **Spawn all at once** — put ALL agent Task calls in ONE message for parallel execution
+3. **Tell the user** — after spawning, list what each agent is doing
+4. **Stop and wait** — after spawning, STOP; do NOT add more tool calls or check status
+5. **No polling** — never poll TaskOutput or swarm status; trust agents to return
+6. **Synthesize** — when agent results arrive, review ALL before proceeding
+7. **No confirmation** — don't ask "should I check?"; just wait for results
 
-## V3 CLI Commands
+---
 
-### Core Commands
+## Branch
 
-| Command | Subcommands | Description |
-|---------|-------------|-------------|
-| `init` | 4 | Project initialization |
-| `agent` | 8 | Agent lifecycle management |
-| `swarm` | 6 | Multi-agent swarm coordination |
-| `memory` | 11 | AgentDB memory with HNSW search |
-| `task` | 6 | Task creation and lifecycle |
-| `session` | 7 | Session state management |
-| `hooks` | 17 | Self-learning hooks + 12 workers |
-| `hive-mind` | 6 | Byzantine fault-tolerant consensus |
-
-### Quick CLI Examples
-
-```bash
-npx @claude-flow/cli@latest init --wizard
-npx @claude-flow/cli@latest agent spawn -t coder --name my-coder
-npx @claude-flow/cli@latest swarm init --v3-mode
-npx @claude-flow/cli@latest memory search --query "authentication patterns"
-npx @claude-flow/cli@latest doctor --fix
-```
-
-## Available Agents (60+ Types)
-
-### Core Development
-`coder`, `reviewer`, `tester`, `planner`, `researcher`
-
-### Specialized
-`security-architect`, `security-auditor`, `memory-specialist`, `performance-engineer`
-
-### Swarm Coordination
-`hierarchical-coordinator`, `mesh-coordinator`, `adaptive-coordinator`
-
-### GitHub & Repository
-`pr-manager`, `code-review-swarm`, `issue-tracker`, `release-manager`
-
-### SPARC Methodology
-`sparc-coord`, `sparc-coder`, `specification`, `pseudocode`, `architecture`
-
-## Memory Commands Reference
-
-```bash
-# Store (REQUIRED: --key, --value; OPTIONAL: --namespace, --ttl, --tags)
-npx @claude-flow/cli@latest memory store --key "pattern-auth" --value "JWT with refresh" --namespace patterns
-
-# Search (REQUIRED: --query; OPTIONAL: --namespace, --limit, --threshold)
-npx @claude-flow/cli@latest memory search --query "authentication patterns"
-
-# List (OPTIONAL: --namespace, --limit)
-npx @claude-flow/cli@latest memory list --namespace patterns --limit 10
-
-# Retrieve (REQUIRED: --key; OPTIONAL: --namespace)
-npx @claude-flow/cli@latest memory retrieve --key "pattern-auth" --namespace patterns
-```
-
-## Quick Setup
-
-```bash
-claude mcp add claude-flow -- npx -y @claude-flow/cli@latest
-npx @claude-flow/cli@latest daemon start
-npx @claude-flow/cli@latest doctor --fix
-```
-
-## Claude Code vs CLI Tools
-
-- Claude Code's Task tool handles ALL execution: agents, file ops, code generation, git
-- CLI tools handle coordination via Bash: swarm init, memory, hooks, routing
-- NEVER use CLI tools as a substitute for Task tool agents
+Default branch: `main`.
+Current operator branch (this session series): `feat/ota-rssi-mobile` —
+PR [#596](https://github.com/ruvnet/RuView/pull/596) on the upstream fork.
 
 ## Support
 
-- Documentation: https://github.com/ruvnet/claude-flow
-- Issues: https://github.com/ruvnet/claude-flow/issues
+- GitHub Issues: <https://github.com/ruvnet/RuView/issues>
+- ADR index: [`docs/adr/`](docs/adr/)
+- Implementation status: [`CHECKLIST.md`](CHECKLIST.md)
+- Detailed dev handbook: [`docs/dev-handbook.md`](docs/dev-handbook.md)

README.md

@@ -36,14 +36,14 @@ RuView also supports pose estimation (17 COCO keypoints via the WiFlow architect
 
 ### Built for low-power edge applications
 
-[Edge modules](#edge-intelligence-adr-041) are small programs that run directly on the ESP32 sensor — no internet needed, no cloud fees, instant response.
+[Edge modules](docs/use-cases.md) are small programs that run directly on the ESP32 sensor — no internet needed, no cloud fees, instant response.
 
 [![Rust 1.85+](https://img.shields.io/badge/rust-1.85+-orange.svg)](https://www.rust-lang.org/)
 [![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
 [![Tests: 1463](https://img.shields.io/badge/tests-1463%20passed-brightgreen.svg)](https://github.com/ruvnet/RuView)
 [![Docker: multi-arch](https://img.shields.io/badge/docker-amd64%20%2B%20arm64-blue.svg)](https://hub.docker.com/r/ruvnet/wifi-densepose)
-[![Vital Signs](https://img.shields.io/badge/vital%20signs-breathing%20%2B%20heartbeat-red.svg)](#vital-sign-detection)
-[![ESP32 Ready](https://img.shields.io/badge/ESP32--S3-CSI%20streaming-purple.svg)](#esp32-s3-hardware-pipeline)
+[![Vital Signs](https://img.shields.io/badge/vital%20signs-breathing%20%2B%20heartbeat-red.svg)](docs/use-cases.md)
+[![ESP32 Ready](https://img.shields.io/badge/ESP32--S3-CSI%20streaming-purple.svg)](docs/use-cases.md)
 [![crates.io](https://img.shields.io/crates/v/wifi-densepose-ruvector.svg)](https://crates.io/crates/wifi-densepose-ruvector)
 
  
@@ -97,7 +97,6 @@ node scripts/mincut-person-counter.js --port 5006  # Correct person counting
 >
 ---
 
-
   <a href="https://ruvnet.github.io/RuView/">
     <img src="assets/v2-screen.png" alt="WiFi DensePose — Live pose detection with setup guide" width="800">
   </a>
@@ -110,378 +109,35 @@ node scripts/mincut-person-counter.js --port 5006  # Correct person counting
   &nbsp;|&nbsp;
   <a href="https://ruvnet.github.io/RuView/pointcloud/"><strong>▶ Live 3D Point Cloud</strong></a>
 
-> The [server](#-quick-start) is optional for visualization and aggregation — the ESP32 [runs independently](#esp32-s3-hardware-pipeline) for presence detection, vital signs, and fall alerts.
+> The [server](docs/dev-handbook.md) is optional for visualization and aggregation — the ESP32 [runs independently](docs/architecture.md) for presence detection, vital signs, and fall alerts.
 >
-> **Live ESP32 pipeline**: Connect an ESP32-S3 node → run the [sensing server](#sensing-server) → open the [pose fusion demo](https://ruvnet.github.io/RuView/pose-fusion.html) for real-time dual-modal pose estimation (webcam + WiFi CSI). See [ADR-059](docs/adr/ADR-059-live-esp32-csi-pipeline.md).
-
+> **Live ESP32 pipeline**: Connect an ESP32-S3 node → run the [sensing server](docs/dev-handbook.md) → open the [pose fusion demo](https://ruvnet.github.io/RuView/pose-fusion.html) for real-time dual-modal pose estimation (webcam + WiFi CSI). See [ADR-059](docs/adr/ADR-059-live-esp32-csi-pipeline.md).
 
 ## 🔬 How It Works
 
-WiFi routers flood every room with radio waves. When a person moves — or even breathes — those waves scatter differently. WiFi DensePose reads that scattering pattern and reconstructs what happened:
+WiFi routers flood every room with radio waves. RuView's ESP32 mesh
+captures CSI from those waves, fuses it across channels and nodes, and
+feeds a coherence-gated signal pipeline into an attention-graph neural
+network that outputs pose keypoints, vital signs, and room fingerprints.
 
-```
-WiFi Router → radio waves pass through room → hit human body → scatter
-    ↓
-ESP32 mesh (4-6 nodes) captures CSI on channels 1/6/11 via TDM protocol
-    ↓
-Multi-Band Fusion: 3 channels × 56 subcarriers = 168 virtual subcarriers per link
-    ↓
-Multistatic Fusion: N×(N-1) links → attention-weighted cross-viewpoint embedding
-    ↓
-Coherence Gate: accept/reject measurements → stable for days without tuning
-    ↓
-Signal Processing: Hampel, SpotFi, Fresnel, BVP, spectrogram → clean features
-    ↓
-AI Backbone (RuVector): attention, graph algorithms, compression, field model
-    ↓
-Signal-Line Protocol (CRV): 6-stage gestalt → sensory → topology → coherence → search → model
-    ↓
-Neural Network: processed signals → 17 body keypoints + vital signs + room model
-    ↓
-Output: real-time pose, breathing, heart rate, room fingerprint, drift alerts
-```
-
-No training cameras required — the [Self-Learning system (ADR-024)](docs/adr/ADR-024-contrastive-csi-embedding-model.md) bootstraps from raw WiFi data alone. [MERIDIAN (ADR-027)](docs/adr/ADR-027-cross-environment-domain-generalization.md) ensures the model works in any room, not just the one it trained in.
+→ **Full pipeline diagram + module-by-module breakdown:**
+  [`docs/architecture.md`](docs/architecture.md)
 
 ---
 
 ## 🏢 Use Cases & Applications
 
-WiFi sensing works anywhere WiFi exists. No new hardware in most cases — just software on existing access points or a $8 ESP32 add-on. Because there are no cameras, deployments avoid privacy regulations (GDPR video, HIPAA imaging) by design.
+RuView serves four deployment tiers: **Everyday** (healthcare, retail,
+office), **Specialized** (events, fitness, education), **Robotics &
+Industrial** (cobots, AMRs, manufacturing) and **Extreme** (search &
+rescue, defense, underground). Each one comes with a concrete hardware
+BOM, expected accuracy, and pointer to the matching ADR-041 edge module.
 
-**Scaling:** Each AP distinguishes ~3-5 people (56 subcarriers). Multi-AP multiplies linearly — a 4-AP retail mesh covers ~15-20 occupants. No hard software limit; the practical ceiling is signal physics.
+Also covers the [Self-Learning WiFi AI (ADR-024)](docs/adr/ADR-024-contrastive-csi-embedding-model.md)
+(128-dim fingerprint, 55 KB on ESP32) and the full 60-module ADR-041
+Edge Intelligence catalogue.
 
-| | Why WiFi sensing wins | Traditional alternative |
-|---|----------------------|----------------------|
-| 🔒 | **No video, no GDPR/HIPAA imaging rules** | Cameras require consent, signage, data retention policies |
-| 🧱 | **Works through walls, shelving, debris** | Cameras need line-of-sight per room |
-| 🌙 | **Works in total darkness** | Cameras need IR or visible light |
-| 💰 | **$0-$8 per zone** (existing WiFi or ESP32) | Camera systems: $200-$2,000 per zone |
-| 🔌 | **WiFi already deployed everywhere** | PIR/radar sensors require new wiring per room |
-
-<details>
-<summary><strong>🏥 Everyday</strong> — Healthcare, retail, office, hospitality (commodity WiFi)</summary>
-
-| Use Case | What It Does | Hardware | Key Metric | Edge Module |
-|----------|-------------|----------|------------|-------------|
-| **Elderly care / assisted living** | Fall detection, nighttime activity monitoring, breathing rate during sleep — no wearable compliance needed | 1 ESP32-S3 per room ($8) | Fall alert <2s | [Sleep Apnea](docs/edge-modules/medical.md), [Gait Analysis](docs/edge-modules/medical.md) |
-| **Hospital patient monitoring** | Continuous breathing + heart rate for non-critical beds without wired sensors; nurse alert on anomaly | 1-2 APs per ward | Breathing: 6-30 BPM | [Respiratory Distress](docs/edge-modules/medical.md), [Cardiac Arrhythmia](docs/edge-modules/medical.md) |
-| **Emergency room triage** | Automated occupancy count + wait-time estimation; detect patient distress (abnormal breathing) in waiting areas | Existing hospital WiFi | Occupancy accuracy >95% | [Queue Length](docs/edge-modules/retail.md), [Panic Motion](docs/edge-modules/security.md) |
-| **Retail occupancy & flow** | Real-time foot traffic, dwell time by zone, queue length — no cameras, no opt-in, GDPR-friendly | Existing store WiFi + 1 ESP32 | Dwell resolution ~1m | [Customer Flow](docs/edge-modules/retail.md), [Dwell Heatmap](docs/edge-modules/retail.md) |
-| **Office space utilization** | Which desks/rooms are actually occupied, meeting room no-shows, HVAC optimization based on real presence | Existing enterprise WiFi | Presence latency <1s | [Meeting Room](docs/edge-modules/building.md), [HVAC Presence](docs/edge-modules/building.md) |
-| **Hotel & hospitality** | Room occupancy without door sensors, minibar/bathroom usage patterns, energy savings on empty rooms | Existing hotel WiFi | 15-30% HVAC savings | [Energy Audit](docs/edge-modules/building.md), [Lighting Zones](docs/edge-modules/building.md) |
-| **Restaurants & food service** | Table turnover tracking, kitchen staff presence, restroom occupancy displays — no cameras in dining areas | Existing WiFi | Queue wait ±30s | [Table Turnover](docs/edge-modules/retail.md), [Queue Length](docs/edge-modules/retail.md) |
-| **Parking garages** | Pedestrian presence in stairwells and elevators where cameras have blind spots; security alert if someone lingers | Existing WiFi | Through-concrete walls | [Loitering](docs/edge-modules/security.md), [Elevator Count](docs/edge-modules/building.md) |
-
-</details>
-
-<details>
-<summary><strong>🏟️ Specialized</strong> — Events, fitness, education, civic (CSI-capable hardware)</summary>
-
-| Use Case | What It Does | Hardware | Key Metric | Edge Module |
-|----------|-------------|----------|------------|-------------|
-| **Smart home automation** | Room-level presence triggers (lights, HVAC, music) that work through walls — no dead zones, no motion-sensor timeouts | 2-3 ESP32-S3 nodes ($24) | Through-wall range ~5m | [HVAC Presence](docs/edge-modules/building.md), [Lighting Zones](docs/edge-modules/building.md) |
-| **Fitness & sports** | Rep counting, posture correction, breathing cadence during exercise — no wearable, no camera in locker rooms | 3+ ESP32-S3 mesh | Pose: 17 keypoints | [Breathing Sync](docs/edge-modules/exotic.md), [Gait Analysis](docs/edge-modules/medical.md) |
-| **Childcare & schools** | Naptime breathing monitoring, playground headcount, restricted-area alerts — privacy-safe for minors | 2-4 ESP32-S3 per zone | Breathing: ±1 BPM | [Sleep Apnea](docs/edge-modules/medical.md), [Perimeter Breach](docs/edge-modules/security.md) |
-| **Event venues & concerts** | Crowd density mapping, crush-risk detection via breathing compression, emergency evacuation flow tracking | Multi-AP mesh (4-8 APs) | Density per m² | [Customer Flow](docs/edge-modules/retail.md), [Panic Motion](docs/edge-modules/security.md) |
-| **Stadiums & arenas** | Section-level occupancy for dynamic pricing, concession staffing, emergency egress flow modeling | Enterprise AP grid | 15-20 per AP mesh | [Dwell Heatmap](docs/edge-modules/retail.md), [Queue Length](docs/edge-modules/retail.md) |
-| **Houses of worship** | Attendance counting without facial recognition — privacy-sensitive congregations, multi-room campus tracking | Existing WiFi | Zone-level accuracy | [Elevator Count](docs/edge-modules/building.md), [Energy Audit](docs/edge-modules/building.md) |
-| **Warehouse & logistics** | Worker safety zones, forklift proximity alerts, occupancy in hazardous areas — works through shelving and pallets | Industrial AP mesh | Alert latency <500ms | [Forklift Proximity](docs/edge-modules/industrial.md), [Confined Space](docs/edge-modules/industrial.md) |
-| **Civic infrastructure** | Public restroom occupancy (no cameras possible), subway platform crowding, shelter headcount during emergencies | Municipal WiFi + ESP32 | Real-time headcount | [Customer Flow](docs/edge-modules/retail.md), [Loitering](docs/edge-modules/security.md) |
-| **Museums & galleries** | Visitor flow heatmaps, exhibit dwell time, crowd bottleneck alerts — no cameras near artwork (flash/theft risk) | Existing WiFi | Zone dwell ±5s | [Dwell Heatmap](docs/edge-modules/retail.md), [Shelf Engagement](docs/edge-modules/retail.md) |
-
-</details>
-
-<details>
-<summary><strong>🤖 Robotics & Industrial</strong> — Autonomous systems, manufacturing, android spatial awareness</summary>
-
-WiFi sensing gives robots and autonomous systems a spatial awareness layer that works where LIDAR and cameras fail — through dust, smoke, fog, and around corners. The CSI signal field acts as a "sixth sense" for detecting humans in the environment without requiring line-of-sight.
-
-| Use Case | What It Does | Hardware | Key Metric | Edge Module |
-|----------|-------------|----------|------------|-------------|
-| **Cobot safety zones** | Detect human presence near collaborative robots — auto-slow or stop before contact, even behind obstructions | 2-3 ESP32-S3 per cell | Presence latency <100ms | [Forklift Proximity](docs/edge-modules/industrial.md), [Perimeter Breach](docs/edge-modules/security.md) |
-| **Warehouse AMR navigation** | Autonomous mobile robots sense humans around blind corners, through shelving racks — no LIDAR occlusion | ESP32 mesh along aisles | Through-shelf detection | [Forklift Proximity](docs/edge-modules/industrial.md), [Loitering](docs/edge-modules/security.md) |
-| **Android / humanoid spatial awareness** | Ambient human pose sensing for social robots — detect gestures, approach direction, and personal space without cameras always on | Onboard ESP32-S3 module | 17-keypoint pose | [Gesture Language](docs/edge-modules/exotic.md), [Emotion Detection](docs/edge-modules/exotic.md) |
-| **Manufacturing line monitoring** | Worker presence at each station, ergonomic posture alerts, headcount for shift compliance — works through equipment | Industrial AP per zone | Pose + breathing | [Confined Space](docs/edge-modules/industrial.md), [Gait Analysis](docs/edge-modules/medical.md) |
-| **Construction site safety** | Exclusion zone enforcement around heavy machinery, fall detection from scaffolding, personnel headcount | Ruggedized ESP32 mesh | Alert <2s, through-dust | [Panic Motion](docs/edge-modules/security.md), [Structural Vibration](docs/edge-modules/industrial.md) |
-| **Agricultural robotics** | Detect farm workers near autonomous harvesters in dusty/foggy field conditions where cameras are unreliable | Weatherproof ESP32 nodes | Range ~10m open field | [Forklift Proximity](docs/edge-modules/industrial.md), [Rain Detection](docs/edge-modules/exotic.md) |
-| **Drone landing zones** | Verify landing area is clear of humans — WiFi sensing works in rain, dust, and low light where downward cameras fail | Ground ESP32 nodes | Presence: >95% accuracy | [Perimeter Breach](docs/edge-modules/security.md), [Tailgating](docs/edge-modules/security.md) |
-| **Clean room monitoring** | Personnel tracking without cameras (particle contamination risk from camera fans) — gown compliance via pose | Existing cleanroom WiFi | No particulate emission | [Clean Room](docs/edge-modules/industrial.md), [Livestock Monitor](docs/edge-modules/industrial.md) |
-
-</details>
-
-<details>
-<summary><strong>🔥 Extreme</strong> — Through-wall, disaster, defense, underground</summary>
-
-These scenarios exploit WiFi's ability to penetrate solid materials — concrete, rubble, earth — where no optical or infrared sensor can reach. The WiFi-Mat disaster module (ADR-001) is specifically designed for this tier.
-
-| Use Case | What It Does | Hardware | Key Metric | Edge Module |
-|----------|-------------|----------|------------|-------------|
-| **Search & rescue (WiFi-Mat)** | Detect survivors through rubble/debris via breathing signature, START triage color classification, 3D localization | Portable ESP32 mesh + laptop | Through 30cm concrete | [Respiratory Distress](docs/edge-modules/medical.md), [Seizure Detection](docs/edge-modules/medical.md) |
-| **Firefighting** | Locate occupants through smoke and walls before entry; breathing detection confirms life signs remotely | Portable mesh on truck | Works in zero visibility | [Sleep Apnea](docs/edge-modules/medical.md), [Panic Motion](docs/edge-modules/security.md) |
-| **Prison & secure facilities** | Cell occupancy verification, distress detection (abnormal vitals), perimeter sensing — no camera blind spots | Dedicated AP infrastructure | 24/7 vital signs | [Cardiac Arrhythmia](docs/edge-modules/medical.md), [Loitering](docs/edge-modules/security.md) |
-| **Military / tactical** | Through-wall personnel detection, room clearing confirmation, hostage vital signs at standoff distance | Directional WiFi + custom FW | Range: 5m through wall | [Perimeter Breach](docs/edge-modules/security.md), [Weapon Detection](docs/edge-modules/security.md) |
-| **Border & perimeter security** | Detect human presence in tunnels, behind fences, in vehicles — passive sensing, no active illumination to reveal position | Concealed ESP32 mesh | Passive / covert | [Perimeter Breach](docs/edge-modules/security.md), [Tailgating](docs/edge-modules/security.md) |
-| **Mining & underground** | Worker presence in tunnels where GPS/cameras fail, breathing detection after collapse, headcount at safety points | Ruggedized ESP32 mesh | Through rock/earth | [Confined Space](docs/edge-modules/industrial.md), [Respiratory Distress](docs/edge-modules/medical.md) |
-| **Maritime & naval** | Below-deck personnel tracking through steel bulkheads (limited range, requires tuning), man-overboard detection | Ship WiFi + ESP32 | Through 1-2 bulkheads | [Structural Vibration](docs/edge-modules/industrial.md), [Panic Motion](docs/edge-modules/security.md) |
-| **Wildlife research** | Non-invasive animal activity monitoring in enclosures or dens — no light pollution, no visual disturbance | Weatherproof ESP32 nodes | Zero light emission | [Livestock Monitor](docs/edge-modules/industrial.md), [Dream Stage](docs/edge-modules/exotic.md) |
-
-</details>
-
-<details>
-<summary><strong>🧩 Edge Intelligence (<a href="docs/adr/ADR-041-wasm-module-collection.md">ADR-041</a>)</strong> — 60 WASM modules across 13 categories, all implemented (609 tests)</summary>
-
-Small programs that run directly on the ESP32 sensor — no internet needed, no cloud fees, instant response. Each module is a tiny WASM file (5-30 KB) that you upload to the device over-the-air. It reads WiFi signal data and makes decisions locally in under 10 ms. [ADR-041](docs/adr/ADR-041-wasm-module-collection.md) defines 60 modules across 13 categories — all 60 are implemented with 609 tests passing.
-
-| | Category | Examples |
-|---|----------|---------|
-| 🏥 | [**Medical & Health**](docs/edge-modules/medical.md) | Sleep apnea detection, cardiac arrhythmia, gait analysis, seizure detection |
-| 🔐 | [**Security & Safety**](docs/edge-modules/security.md) | Intrusion detection, perimeter breach, loitering, panic motion |
-| 🏢 | [**Smart Building**](docs/edge-modules/building.md) | Zone occupancy, HVAC control, elevator counting, meeting room tracking |
-| 🛒 | [**Retail & Hospitality**](docs/edge-modules/retail.md) | Queue length, dwell heatmaps, customer flow, table turnover |
-| 🏭 | [**Industrial**](docs/edge-modules/industrial.md) | Forklift proximity, confined space monitoring, structural vibration |
-| 🔮 | [**Exotic & Research**](docs/edge-modules/exotic.md) | Sleep staging, emotion detection, sign language, breathing sync |
-| 📡 | [**Signal Intelligence**](docs/edge-modules/signal-intelligence.md) | Cleans and sharpens raw WiFi signals — focuses on important regions, filters noise, fills in missing data, and tracks which person is which |
-| 🧠 | [**Adaptive Learning**](docs/edge-modules/adaptive-learning.md) | The sensor learns new gestures and patterns on its own over time — no cloud needed, remembers what it learned even after updates |
-| 🗺️ | [**Spatial Reasoning**](docs/edge-modules/spatial-temporal.md) | Figures out where people are in a room, which zones matter most, and tracks movement across areas using graph-based spatial logic |
-| ⏱️ | [**Temporal Analysis**](docs/edge-modules/spatial-temporal.md) | Learns daily routines, detects when patterns break (someone didn't get up), and verifies safety rules are being followed over time |
-| 🛡️ | [**AI Security**](docs/edge-modules/ai-security.md) | Detects signal replay attacks, WiFi jamming, injection attempts, and flags abnormal behavior that could indicate tampering |
-| ⚛️ | [**Quantum-Inspired**](docs/edge-modules/autonomous.md) | Uses quantum-inspired math to map room-wide signal coherence and search for optimal sensor configurations |
-| 🤖 | [**Autonomous & Exotic**](docs/edge-modules/autonomous.md) | Self-managing sensor mesh — auto-heals dropped nodes, plans its own actions, and explores experimental signal representations |
-
-All implemented modules are `no_std` Rust, share a [common utility library](v2/crates/wifi-densepose-wasm-edge/src/vendor_common.rs), and talk to the host through a 12-function API. Full documentation: [**Edge Modules Guide**](docs/edge-modules/README.md). See the [complete implemented module list](#edge-module-list) below.
-
-</details>
-
-<details id="edge-module-list">
-<summary><strong>🧩 Edge Intelligence — <a href="docs/edge-modules/README.md">All 65 Modules Implemented</a></strong> (ADR-041 complete)</summary>
-
-All 60 modules are implemented, tested (609 tests passing), and ready to deploy. They compile to `wasm32-unknown-unknown`, run on ESP32-S3 via WASM3, and share a [common utility library](v2/crates/wifi-densepose-wasm-edge/src/vendor_common.rs). Source: [`crates/wifi-densepose-wasm-edge/src/`](v2/crates/wifi-densepose-wasm-edge/src/)
-
-**Core modules** (ADR-040 flagship + early implementations):
-
-| Module | File | What It Does |
-|--------|------|-------------|
-| Gesture Classifier | [`gesture.rs`](v2/crates/wifi-densepose-wasm-edge/src/gesture.rs) | DTW template matching for hand gestures |
-| Coherence Filter | [`coherence.rs`](v2/crates/wifi-densepose-wasm-edge/src/coherence.rs) | Phase coherence gating for signal quality |
-| Adversarial Detector | [`adversarial.rs`](v2/crates/wifi-densepose-wasm-edge/src/adversarial.rs) | Detects physically impossible signal patterns |
-| Intrusion Detector | [`intrusion.rs`](v2/crates/wifi-densepose-wasm-edge/src/intrusion.rs) | Human vs non-human motion classification |
-| Occupancy Counter | [`occupancy.rs`](v2/crates/wifi-densepose-wasm-edge/src/occupancy.rs) | Zone-level person counting |
-| Vital Trend | [`vital_trend.rs`](v2/crates/wifi-densepose-wasm-edge/src/vital_trend.rs) | Long-term breathing and heart rate trending |
-| RVF Parser | [`rvf.rs`](v2/crates/wifi-densepose-wasm-edge/src/rvf.rs) | RVF container format parsing |
-
-**Vendor-integrated modules** (24 modules, ADR-041 Category 7):
-
-**📡 Signal Intelligence** — Real-time CSI analysis and feature extraction
-
-| Module | File | What It Does | Budget |
-|--------|------|-------------|--------|
-| Flash Attention | [`sig_flash_attention.rs`](v2/crates/wifi-densepose-wasm-edge/src/sig_flash_attention.rs) | Tiled attention over 8 subcarrier groups — finds spatial focus regions and entropy | S (<5ms) |
-| Coherence Gate | [`sig_coherence_gate.rs`](v2/crates/wifi-densepose-wasm-edge/src/sig_coherence_gate.rs) | Z-score phasor gating with hysteresis: Accept / PredictOnly / Reject / Recalibrate | L (<2ms) |
-| Temporal Compress | [`sig_temporal_compress.rs`](v2/crates/wifi-densepose-wasm-edge/src/sig_temporal_compress.rs) | 3-tier adaptive quantization (8-bit hot / 5-bit warm / 3-bit cold) | L (<2ms) |
-| Sparse Recovery | [`sig_sparse_recovery.rs`](v2/crates/wifi-densepose-wasm-edge/src/sig_sparse_recovery.rs) | ISTA L1 reconstruction for dropped subcarriers | H (<10ms) |
-| Person Match | [`sig_mincut_person_match.rs`](v2/crates/wifi-densepose-wasm-edge/src/sig_mincut_person_match.rs) | Hungarian-lite bipartite assignment for multi-person tracking | S (<5ms) |
-| Optimal Transport | [`sig_optimal_transport.rs`](v2/crates/wifi-densepose-wasm-edge/src/sig_optimal_transport.rs) | Sliced Wasserstein-1 distance with 4 projections | L (<2ms) |
-
-**🧠 Adaptive Learning** — On-device learning without cloud connectivity
-
-| Module | File | What It Does | Budget |
-|--------|------|-------------|--------|
-| DTW Gesture Learn | [`lrn_dtw_gesture_learn.rs`](v2/crates/wifi-densepose-wasm-edge/src/lrn_dtw_gesture_learn.rs) | User-teachable gesture recognition — 3-rehearsal protocol, 16 templates | S (<5ms) |
-| Anomaly Attractor | [`lrn_anomaly_attractor.rs`](v2/crates/wifi-densepose-wasm-edge/src/lrn_anomaly_attractor.rs) | 4D dynamical system attractor classification with Lyapunov exponents | H (<10ms) |
-| Meta Adapt | [`lrn_meta_adapt.rs`](v2/crates/wifi-densepose-wasm-edge/src/lrn_meta_adapt.rs) | Hill-climbing self-optimization with safety rollback | L (<2ms) |
-| EWC Lifelong | [`lrn_ewc_lifelong.rs`](v2/crates/wifi-densepose-wasm-edge/src/lrn_ewc_lifelong.rs) | Elastic Weight Consolidation — remembers past tasks while learning new ones | S (<5ms) |
-
-**🗺️ Spatial Reasoning** — Location, proximity, and influence mapping
-
-| Module | File | What It Does | Budget |
-|--------|------|-------------|--------|
-| PageRank Influence | [`spt_pagerank_influence.rs`](v2/crates/wifi-densepose-wasm-edge/src/spt_pagerank_influence.rs) | 4x4 cross-correlation graph with power iteration PageRank | L (<2ms) |
-| Micro HNSW | [`spt_micro_hnsw.rs`](v2/crates/wifi-densepose-wasm-edge/src/spt_micro_hnsw.rs) | 64-vector navigable small-world graph for nearest-neighbor search | S (<5ms) |
-| Spiking Tracker | [`spt_spiking_tracker.rs`](v2/crates/wifi-densepose-wasm-edge/src/spt_spiking_tracker.rs) | 32 LIF neurons + 4 output zone neurons with STDP learning | S (<5ms) |
-
-**⏱️ Temporal Analysis** — Activity patterns, logic verification, autonomous planning
-
-| Module | File | What It Does | Budget |
-|--------|------|-------------|--------|
-| Pattern Sequence | [`tmp_pattern_sequence.rs`](v2/crates/wifi-densepose-wasm-edge/src/tmp_pattern_sequence.rs) | Activity routine detection and deviation alerts | S (<5ms) |
-| Temporal Logic Guard | [`tmp_temporal_logic_guard.rs`](v2/crates/wifi-densepose-wasm-edge/src/tmp_temporal_logic_guard.rs) | LTL formula verification on CSI event streams | S (<5ms) |
-| GOAP Autonomy | [`tmp_goap_autonomy.rs`](v2/crates/wifi-densepose-wasm-edge/src/tmp_goap_autonomy.rs) | Goal-Oriented Action Planning for autonomous module management | S (<5ms) |
-
-**🛡️ AI Security** — Tamper detection and behavioral anomaly profiling
-
-| Module | File | What It Does | Budget |
-|--------|------|-------------|--------|
-| Prompt Shield | [`ais_prompt_shield.rs`](v2/crates/wifi-densepose-wasm-edge/src/ais_prompt_shield.rs) | FNV-1a replay detection, injection detection (10x amplitude), jamming (SNR) | L (<2ms) |
-| Behavioral Profiler | [`ais_behavioral_profiler.rs`](v2/crates/wifi-densepose-wasm-edge/src/ais_behavioral_profiler.rs) | 6D behavioral profile with Mahalanobis anomaly scoring | S (<5ms) |
-
-**⚛️ Quantum-Inspired** — Quantum computing metaphors applied to CSI analysis
-
-| Module | File | What It Does | Budget |
-|--------|------|-------------|--------|
-| Quantum Coherence | [`qnt_quantum_coherence.rs`](v2/crates/wifi-densepose-wasm-edge/src/qnt_quantum_coherence.rs) | Bloch sphere mapping, Von Neumann entropy, decoherence detection | S (<5ms) |
-| Interference Search | [`qnt_interference_search.rs`](v2/crates/wifi-densepose-wasm-edge/src/qnt_interference_search.rs) | 16 room-state hypotheses with Grover-inspired oracle + diffusion | S (<5ms) |
-
-**🤖 Autonomous Systems** — Self-governing and self-healing behaviors
-
-| Module | File | What It Does | Budget |
-|--------|------|-------------|--------|
-| Psycho-Symbolic | [`aut_psycho_symbolic.rs`](v2/crates/wifi-densepose-wasm-edge/src/aut_psycho_symbolic.rs) | 16-rule forward-chaining knowledge base with contradiction detection | S (<5ms) |
-| Self-Healing Mesh | [`aut_self_healing_mesh.rs`](v2/crates/wifi-densepose-wasm-edge/src/aut_self_healing_mesh.rs) | 8-node mesh with health tracking, degradation/recovery, coverage healing | S (<5ms) |
-
-**🔮 Exotic (Vendor)** — Novel mathematical models for CSI interpretation
-
-| Module | File | What It Does | Budget |
-|--------|------|-------------|--------|
-| Time Crystal | [`exo_time_crystal.rs`](v2/crates/wifi-densepose-wasm-edge/src/exo_time_crystal.rs) | Autocorrelation subharmonic detection in 256-frame history | S (<5ms) |
-| Hyperbolic Space | [`exo_hyperbolic_space.rs`](v2/crates/wifi-densepose-wasm-edge/src/exo_hyperbolic_space.rs) | Poincare ball embedding with 32 reference locations, hyperbolic distance | S (<5ms) |
-
-**🏥 Medical & Health** (Category 1) — Contactless health monitoring
-
-| Module | File | What It Does | Budget |
-|--------|------|-------------|--------|
-| Sleep Apnea | [`med_sleep_apnea.rs`](v2/crates/wifi-densepose-wasm-edge/src/med_sleep_apnea.rs) | Detects breathing pauses during sleep | S (<5ms) |
-| Cardiac Arrhythmia | [`med_cardiac_arrhythmia.rs`](v2/crates/wifi-densepose-wasm-edge/src/med_cardiac_arrhythmia.rs) | Monitors heart rate for irregular rhythms | S (<5ms) |
-| Respiratory Distress | [`med_respiratory_distress.rs`](v2/crates/wifi-densepose-wasm-edge/src/med_respiratory_distress.rs) | Alerts on abnormal breathing patterns | S (<5ms) |
-| Gait Analysis | [`med_gait_analysis.rs`](v2/crates/wifi-densepose-wasm-edge/src/med_gait_analysis.rs) | Tracks walking patterns and detects changes | S (<5ms) |
-| Seizure Detection | [`med_seizure_detect.rs`](v2/crates/wifi-densepose-wasm-edge/src/med_seizure_detect.rs) | 6-state machine for tonic-clonic seizure recognition | S (<5ms) |
-
-**🔐 Security & Safety** (Category 2) — Perimeter and threat detection
-
-| Module | File | What It Does | Budget |
-|--------|------|-------------|--------|
-| Perimeter Breach | [`sec_perimeter_breach.rs`](v2/crates/wifi-densepose-wasm-edge/src/sec_perimeter_breach.rs) | Detects boundary crossings with approach/departure | S (<5ms) |
-| Weapon Detection | [`sec_weapon_detect.rs`](v2/crates/wifi-densepose-wasm-edge/src/sec_weapon_detect.rs) | Metal anomaly detection via CSI amplitude shifts | S (<5ms) |
-| Tailgating | [`sec_tailgating.rs`](v2/crates/wifi-densepose-wasm-edge/src/sec_tailgating.rs) | Detects unauthorized follow-through at access points | S (<5ms) |
-| Loitering | [`sec_loitering.rs`](v2/crates/wifi-densepose-wasm-edge/src/sec_loitering.rs) | Alerts when someone lingers too long in a zone | S (<5ms) |
-| Panic Motion | [`sec_panic_motion.rs`](v2/crates/wifi-densepose-wasm-edge/src/sec_panic_motion.rs) | Detects fleeing, struggling, or panic movement | S (<5ms) |
-
-**🏢 Smart Building** (Category 3) — Automation and energy efficiency
-
-| Module | File | What It Does | Budget |
-|--------|------|-------------|--------|
-| HVAC Presence | [`bld_hvac_presence.rs`](v2/crates/wifi-densepose-wasm-edge/src/bld_hvac_presence.rs) | Occupancy-driven HVAC control with departure countdown | S (<5ms) |
-| Lighting Zones | [`bld_lighting_zones.rs`](v2/crates/wifi-densepose-wasm-edge/src/bld_lighting_zones.rs) | Auto-dim/off lighting based on zone activity | S (<5ms) |
-| Elevator Count | [`bld_elevator_count.rs`](v2/crates/wifi-densepose-wasm-edge/src/bld_elevator_count.rs) | Counts people entering/leaving with overload warning | S (<5ms) |
-| Meeting Room | [`bld_meeting_room.rs`](v2/crates/wifi-densepose-wasm-edge/src/bld_meeting_room.rs) | Tracks meeting lifecycle: start, headcount, end, availability | S (<5ms) |
-| Energy Audit | [`bld_energy_audit.rs`](v2/crates/wifi-densepose-wasm-edge/src/bld_energy_audit.rs) | Tracks after-hours usage and room utilization rates | S (<5ms) |
-
-**🛒 Retail & Hospitality** (Category 4) — Customer insights without cameras
-
-| Module | File | What It Does | Budget |
-|--------|------|-------------|--------|
-| Queue Length | [`ret_queue_length.rs`](v2/crates/wifi-densepose-wasm-edge/src/ret_queue_length.rs) | Estimates queue size and wait times | S (<5ms) |
-| Dwell Heatmap | [`ret_dwell_heatmap.rs`](v2/crates/wifi-densepose-wasm-edge/src/ret_dwell_heatmap.rs) | Shows where people spend time (hot/cold zones) | S (<5ms) |
-| Customer Flow | [`ret_customer_flow.rs`](v2/crates/wifi-densepose-wasm-edge/src/ret_customer_flow.rs) | Counts ins/outs and tracks net occupancy | S (<5ms) |
-| Table Turnover | [`ret_table_turnover.rs`](v2/crates/wifi-densepose-wasm-edge/src/ret_table_turnover.rs) | Restaurant table lifecycle: seated, dining, vacated | S (<5ms) |
-| Shelf Engagement | [`ret_shelf_engagement.rs`](v2/crates/wifi-densepose-wasm-edge/src/ret_shelf_engagement.rs) | Detects browsing, considering, and reaching for products | S (<5ms) |
-
-**🏭 Industrial & Specialized** (Category 5) — Safety and compliance
-
-| Module | File | What It Does | Budget |
-|--------|------|-------------|--------|
-| Forklift Proximity | [`ind_forklift_proximity.rs`](v2/crates/wifi-densepose-wasm-edge/src/ind_forklift_proximity.rs) | Warns when people get too close to vehicles | S (<5ms) |
-| Confined Space | [`ind_confined_space.rs`](v2/crates/wifi-densepose-wasm-edge/src/ind_confined_space.rs) | OSHA-compliant worker monitoring with extraction alerts | S (<5ms) |
-| Clean Room | [`ind_clean_room.rs`](v2/crates/wifi-densepose-wasm-edge/src/ind_clean_room.rs) | Occupancy limits and turbulent motion detection | S (<5ms) |
-| Livestock Monitor | [`ind_livestock_monitor.rs`](v2/crates/wifi-densepose-wasm-edge/src/ind_livestock_monitor.rs) | Animal presence, stillness, and escape alerts | S (<5ms) |
-| Structural Vibration | [`ind_structural_vibration.rs`](v2/crates/wifi-densepose-wasm-edge/src/ind_structural_vibration.rs) | Seismic events, mechanical resonance, structural drift | S (<5ms) |
-
-**🔮 Exotic & Research** (Category 6) — Experimental sensing applications
-
-| Module | File | What It Does | Budget |
-|--------|------|-------------|--------|
-| Dream Stage | [`exo_dream_stage.rs`](v2/crates/wifi-densepose-wasm-edge/src/exo_dream_stage.rs) | Contactless sleep stage classification (wake/light/deep/REM) | S (<5ms) |
-| Emotion Detection | [`exo_emotion_detect.rs`](v2/crates/wifi-densepose-wasm-edge/src/exo_emotion_detect.rs) | Arousal, stress, and calm detection from micro-movements | S (<5ms) |
-| Gesture Language | [`exo_gesture_language.rs`](v2/crates/wifi-densepose-wasm-edge/src/exo_gesture_language.rs) | Sign language letter recognition via WiFi | S (<5ms) |
-| Music Conductor | [`exo_music_conductor.rs`](v2/crates/wifi-densepose-wasm-edge/src/exo_music_conductor.rs) | Tempo and dynamic tracking from conducting gestures | S (<5ms) |
-| Plant Growth | [`exo_plant_growth.rs`](v2/crates/wifi-densepose-wasm-edge/src/exo_plant_growth.rs) | Monitors plant growth, circadian rhythms, wilt detection | S (<5ms) |
-| Ghost Hunter | [`exo_ghost_hunter.rs`](v2/crates/wifi-densepose-wasm-edge/src/exo_ghost_hunter.rs) | Environmental anomaly classification (draft/insect/wind/unknown) | S (<5ms) |
-| Rain Detection | [`exo_rain_detect.rs`](v2/crates/wifi-densepose-wasm-edge/src/exo_rain_detect.rs) | Detects rain onset, intensity, and cessation via signal scatter | S (<5ms) |
-| Breathing Sync | [`exo_breathing_sync.rs`](v2/crates/wifi-densepose-wasm-edge/src/exo_breathing_sync.rs) | Detects synchronized breathing between multiple people | S (<5ms) |
-
-</details>
-
----
-
-<details>
-<summary><strong>🧠 Self-Learning WiFi AI (ADR-024)</strong> — Adaptive recognition, self-optimization, and intelligent anomaly detection</summary>
-
-Every WiFi signal that passes through a room creates a unique fingerprint of that space. WiFi-DensePose already reads these fingerprints to track people, but until now it threw away the internal "understanding" after each reading. The Self-Learning WiFi AI captures and preserves that understanding as compact, reusable vectors — and continuously optimizes itself for each new environment.
-
-**What it does in plain terms:**
-- Turns any WiFi signal into a 128-number "fingerprint" that uniquely describes what's happening in a room
-- Learns entirely on its own from raw WiFi data — no cameras, no labeling, no human supervision needed
-- Recognizes rooms, detects intruders, identifies people, and classifies activities using only WiFi
-- Runs on an $8 ESP32 chip (the entire model fits in 55 KB of memory)
-- Produces both body pose tracking AND environment fingerprints in a single computation
-
-**Key Capabilities**
-
-| What | How it works | Why it matters |
-|------|-------------|----------------|
-| **Self-supervised learning** | The model watches WiFi signals and teaches itself what "similar" and "different" look like, without any human-labeled data | Deploy anywhere — just plug in a WiFi sensor and wait 10 minutes |
-| **Room identification** | Each room produces a distinct WiFi fingerprint pattern | Know which room someone is in without GPS or beacons |
-| **Anomaly detection** | An unexpected person or event creates a fingerprint that doesn't match anything seen before | Automatic intrusion and fall detection as a free byproduct |
-| **Person re-identification** | Each person disturbs WiFi in a slightly different way, creating a personal signature | Track individuals across sessions without cameras |
-| **Environment adaptation** | MicroLoRA adapters (1,792 parameters per room) fine-tune the model for each new space | Adapts to a new room with minimal data — 93% less than retraining from scratch |
-| **Memory preservation** | EWC++ regularization remembers what was learned during pretraining | Switching to a new task doesn't erase prior knowledge |
-| **Hard-negative mining** | Training focuses on the most confusing examples to learn faster | Better accuracy with the same amount of training data |
-
-**Architecture**
-
-```
-WiFi Signal [56 channels] → Transformer + Graph Neural Network
-                                  ├→ 128-dim environment fingerprint (for search + identification)
-                                  └→ 17-joint body pose (for human tracking)
-```
-
-**Quick Start**
-
-```bash
-# Step 1: Learn from raw WiFi data (no labels needed)
-cargo run -p wifi-densepose-sensing-server -- --pretrain --dataset data/csi/ --pretrain-epochs 50
-
-# Step 2: Fine-tune with pose labels for full capability
-cargo run -p wifi-densepose-sensing-server -- --train --dataset data/mmfi/ --epochs 100 --save-rvf model.rvf
-
-# Step 3: Use the model — extract fingerprints from live WiFi
-cargo run -p wifi-densepose-sensing-server -- --model model.rvf --embed
-
-# Step 4: Search — find similar environments or detect anomalies
-cargo run -p wifi-densepose-sensing-server -- --model model.rvf --build-index env
-```
-
-**Training Modes**
-
-| Mode | What you need | What you get |
-|------|--------------|-------------|
-| Self-Supervised | Just raw WiFi data | A model that understands WiFi signal structure |
-| Supervised | WiFi data + body pose labels | Full pose tracking + environment fingerprints |
-| Cross-Modal | WiFi data + camera footage | Fingerprints aligned with visual understanding |
-
-**Fingerprint Index Types**
-
-| Index | What it stores | Real-world use |
-|-------|---------------|----------------|
-| `env_fingerprint` | Average room fingerprint | "Is this the kitchen or the bedroom?" |
-| `activity_pattern` | Activity boundaries | "Is someone cooking, sleeping, or exercising?" |
-| `temporal_baseline` | Normal conditions | "Something unusual just happened in this room" |
-| `person_track` | Individual movement signatures | "Person A just entered the living room" |
-
-**Model Size**
-
-| Component | Parameters | Memory (on ESP32) |
-|-----------|-----------|-------------------|
-| Transformer backbone | ~28,000 | 28 KB |
-| Embedding projection head | ~25,000 | 25 KB |
-| Per-room MicroLoRA adapter | ~1,800 | 2 KB |
-| **Total** | **~55,000** | **55 KB** (of 520 KB available) |
-
-The self-learning system builds on the [AI Backbone (RuVector)](#ai-backbone-ruvector) signal-processing layer — attention, graph algorithms, and compression — adding contrastive learning on top.
-
-See [`docs/adr/ADR-024-contrastive-csi-embedding-model.md`](docs/adr/ADR-024-contrastive-csi-embedding-model.md) for full architectural details.
-
-</details>
+→ **Full catalogue + tables: [`docs/use-cases.md`](docs/use-cases.md)**
 
 ---
 

assets/sensors/README.md

@@ -0,0 +1,63 @@
+# Sensor & antenna hardware inventory
+
+Photos of the deployment's 6-node sensor mesh + external-antenna stock,
+captured 2026-05-18. The fleet splits cleanly into two roles: two
+**camera-bearing** nodes (1, 2) that can collect ground-truth keypoints
+on-device, and four **antenna-upgradeable** nodes (3-6) that supply
+spatial coverage.
+
+## Active sensor mesh (6 ESP32-S3 nodes)
+
+| node | IP | Board | Camera | u.FL | Photos |
+|---|---|---|---|---|---|
+| 1 | 192.168.0.101 | ESP32-S3 + OV camera + microSD + USB-OTG (FFC ribbon) | ✅ | — | [`sensor_06`](sensor_06.jpeg), [`sensor_07`](sensor_07.jpeg) |
+| 2 | 192.168.0.100 | same as node 1 | ✅ | — | [`sensor_06`](sensor_06.jpeg), [`sensor_07`](sensor_07.jpeg) |
+| 3 | 192.168.0.102 | YD-ESP32-23 V1.3, ESP32-S3-N16R8 + FTDI USB-serial + dual USB-C | — | ✅ | [`sensor_08`](sensor_08.jpeg), [`sensor_09`](sensor_09.jpeg) |
+| 4 | 192.168.0.104 | same as node 3 | — | ✅ | same |
+| 5 | 192.168.0.105 | same as node 3 | — | ✅ | same |
+| 6 | 192.168.0.106 | same as node 3 | — | ✅ | same |
+
+## External antenna stock (for nodes 3-6)
+
+| File | What | Use |
+|---|---|---|
+| [`sensor_01.jpeg`](sensor_01.jpeg) | 5× u.FL (IPEX-1) pigtail antennas, bare cable | Direct feed via the u.FL connector on YD-ESP32-23 boards (nodes 3-6). Adds gain over the chip antenna; supports polarisation diversity if mounted perpendicular pairs. |
+| [`sensor_02.jpeg`](sensor_02.jpeg) | 4× flat PCB-strip 2.4 GHz antennas with 3M double-sided tape backing + u.FL pigtails | Stick-on external antennas — better directivity than the bare pigtail. One per node 3-6 = full set. |
+
+## Auxiliary modality — mmWave radar (vitals ground truth)
+
+| File | What | Use |
+|---|---|---|
+| [`sensor_03.jpeg`](sensor_03.jpeg) | HLK-LD2402 24 GHz mmWave radar (V1.0, chip `S1KM0008` 2438 batch), TX/RX patch antennas | New sensing modality. mmWave gives sub-mm range to a moving target — ideal for vitals (breathing / pulse) ground truth alongside WiFi CSI. UART output @ 256000 8N1. |
+| [`sensor_04.jpeg`](sensor_04.jpeg) | CP2102 USB-to-UART bridge (AMS1117-3.3 LDO, USB-A) | Powers + reads the HLK-LD2402 from the Mac. Pin map: GND / RXT / TXD / 3.3V / RTS / CTS. |
+| [`sensor_05.jpeg`](sensor_05.jpeg) | HLK-LD2402 + USB-UART wired together | Plug-and-play setup; module ships with factory firmware, no flashing required. |
+
+## Suggested next moves
+
+* **External antennas for nodes 3-6** — the 4 PCB-strip antennas in
+  `sensor_02` map 1:1 to the 4 YD-ESP32-23 boards. Power-cycle each
+  to attach. Per the ADR-118 audit, nodes near the AP currently sit
+  at sep_ratio ~0.05 — external antennas perpendicular to the body
+  axis should pull more body modulation into the signal.
+* **HLK-LD2402 as vitals ground truth** — connect via USB-UART, log
+  breathing rate alongside the WiFi vitals detector, compare bias.
+  Later fuse via `MultistaticFuser` if accuracy delta is material.
+  Would warrant a fresh ADR.
+* **On-device camera capture for WiFlow Pack E.2 retrain** — nodes
+  1 and 2 already have OV-cameras. Path is:
+  1. Extend `firmware/esp32-csi-node/` with a parallel
+     `camera_capture.c` that grabs frames @ ~10 Hz and streams them
+     to the server as MJPEG over a new UDP port (or HTTP `multipart/x-mixed-replace`).
+  2. Run MediaPipe Pose on the server (we already have it in
+     `~/.venv/ruview-train` from this session).
+  3. Time-align CSI + keypoints via the existing
+     `scripts/align-ground-truth.js` infrastructure.
+  4. Train via `scripts/train-wiflow-supervised.js --scale lite`.
+  This is the cleanest replacement for the awkward "laptop is the
+  camera AND the server AND in the sensing zone" workaround.
+
+---
+
+These are reference photos. Linked from `docs/use-cases.md` and
+`CHECKLIST.md` so future sessions see the available hardware at
+a glance.

docs/adr/ADR-116-wiflow-v1-supervised-pose-loader.md

@@ -121,23 +121,15 @@ false`.
 
 ### D7 — Honest about output quality
 
-The loaded model produces **17 keypoints**, but the **numerical values
-are saturated** (most x/y near 0 or 1) — sigmoid extremes meaning the
-network has no learned response to our specific deployment's CSI
-distribution. This is expected: the model was trained on a different
-ESP32 setup, different room, different person, with camera ground truth
-we don't have here. **The integration is correct; the model needs
-deployment-specific fine-tune to produce useful keypoints.**
+The loaded model emits 17 keypoints, but values saturate near 0/1
+(sigmoid extremes) — the network was trained on a different ESP32
+deployment and has no learned response to ours. Integration is correct;
+production-grade output needs deployment-specific fine-tune.
 
-Two paths to usable output, left as follow-ups (Pack E):
-
-1. **Apply `node-1.json` / `node-2.json` LoRA adapters** (ADR-117 candidate)
-   — they're shipped alongside `wiflow-v1.json` in the same HuggingFace
-   repo, rank=8, alpha=16, target the encoder + task heads. Loader stub +
-   forward fold ~2 h.
-2. **Re-train via `scripts/train-wiflow-supervised.js` with new ground-
-   truth capture** (~30 min capture + 19 min training per the model card).
-   Operator-side work.
+Follow-ups (Pack E): apply `node-1`/`node-2` LoRA adapters from the
+same HuggingFace repo (~2h), or re-train via
+`scripts/train-wiflow-supervised.js` against fresh camera ground-truth
+(~30 min capture + 19 min training).
 
 ## Files Touched
 
@@ -159,28 +151,12 @@ Binary size delta: 3.0 MB → 3.1 MB.
 
 ## Verified Acceptance
 
-Live test on the operator's TP-Link deployment (.103, both nodes
-192.168.0.100/.101):
-
-```
-$ ./target/release/sensing-server --source esp32 --csi-keepalive-pps 25 \
-    --wiflow-model data/models/ruview/wiflow-v1/wiflow-v1.json
-  ...
-  ADR-116 wiflow-v1 loaded from data/models/ruview/wiflow-v1/wiflow-v1.json
-                                              (lite scale, 186946 params)
-  keepalive: learned address for node 2 = 192.168.0.100:63940
-  keepalive: learned address for node 1 = 192.168.0.101:63844
-
-$ curl :8080/api/v1/info  → "pose_estimation": true
-$ curl :8080/api/v1/pose/stats  → "model_loaded": true, frames_processed: 2699
-$ curl :8080/api/v1/pose/current
-  { persons: [{id: 1, keypoints: [17 × {name, x, y, z, confidence}], ...}],
-    total_persons: 1, model_loaded: true }
-```
-
-End-to-end: model on disk → loader → forward pass → 17 keypoints → REST &
-WS payload. UI's pose canvas (un-gated by ADR-105 D4) now draws what the
-model emits.
+Live on the operator's TP-Link deployment (Mac .103, nodes .100/.101):
+sensing-server log shows `ADR-116 wiflow-v1 loaded ... (lite scale,
+186946 params)` + `keepalive: learned address for node 2/1`; `curl
+/api/v1/info` returns `"pose_estimation": true`; `curl /api/v1/pose/current`
+returns 17 named COCO keypoints under one `persons[0]`. End-to-end:
+model on disk → loader → forward pass → 17 keypoints → REST + WS payload.
 
 ## Cargo tests
 
@@ -194,23 +170,14 @@ model emits.
 
 ## Open Items
 
-* **Pack E.1 — LoRA adapter loader.** `node-1.json` / `node-2.json` rank-8
-  adapters from the same HF repo, ~21 KB each. The trainer encodes them
-  in the same custom format as `wiflow-v1.json` (different `format` tag),
-  so the loader plumbing is small. ~2 h.
-* **Pack E.2 — Camera-supervised retraining for this room.** Run
-  `scripts/collect-ground-truth.py` against this Mac's webcam +
-  TP-Link/.100/.101 CSI for 5 min, then `scripts/train-wiflow-
-  supervised.js --scale lite`. Should drop sigmoid saturation and produce
-  spatially-coherent keypoints. ~1 h operator + 19 min train.
-* **Inference rate-limiting.** Currently runs every tick (10 fps). If
-  multiple WS clients connect, each tick computes once and the result is
-  reused — fine. If model size grows to small/medium scale (~200K/800K
-  params), should cache the result per tick instead of computing per-client.
-* **Per-node pose tracks.** Right now a single virtual person is emitted;
-  the broadcaster places it in `zone_1` with a fixed bbox. If/when LoRA
-  adapters disambiguate per-node viewpoints, fan out to one
-  `PersonDetection` per node (left/right of the room).
+* **Pack E.1 — LoRA adapter loader.** Apply `node-1`/`node-2` rank-8
+  adapters from the same HF repo (~2 h).
+* **Pack E.2 — Camera-supervised retrain for this room.**
+  `scripts/collect-ground-truth.py` + `scripts/train-wiflow-supervised.js
+  --scale lite` — should drop sigmoid saturation (~1 h + 19 min train).
+* **Inference rate-limit / per-node pose tracks** — currently single
+  virtual person emitted with fixed `zone_1` bbox; future LoRA-per-node
+  could fan out to one `PersonDetection` per sensor viewpoint.
 
 ## References
 

docs/adr/ADR-117-process-hygiene-and-audit-followups.md

@@ -10,70 +10,32 @@
 
 ## Context
 
-A deep audit pass (4 parallel auditors covering sensors, server, UI, docs)
-surfaced two operational fires and a stack of correctness/honesty issues
-that had accumulated across ADR-100..116. This ADR collects the immediate
-fixes.
+A 4-auditor pass (sensors, server, UI, docs) over ADR-100..116
+surfaced two operational fires and a stack of correctness/honesty
+issues. This ADR collects the immediate fixes.
 
-### Fire 1 — Runaway ping zombies
-
-Live `ps` showed **250+ `/sbin/ping -i 0.040` processes** on the Mac, most
-parented to PID 1 (orphans from prior server lifetimes) and **8 fresh
-pings to `127.0.0.1` parented to the current server**.
-
-Root cause: a `cargo test --workspace` run sent UDP packets to
-`127.0.0.1:5005` from `tests/multi_node_test.rs::test_multi_node_udp_send`
-while the production server was bound to `0.0.0.0:5005`. The integration
-test injects 55 synthetic frames with `node_ids = [1, 2, 3, 5, 7]`. Each
-distinct `node_id` byte in a CSI magic packet triggered a fresh entry in
-`NODE_ADDRS`, and the keepalive task spawned exactly one `ping` child
-per entry. Combined with macOS not propagating parent death to children
-(killed servers leave ping orphans), the count accumulated rapidly.
-
-### Fire 2 — Per-node feature divergence on node 2
-
-Node 2 (192.168.0.100) showed `dominant_freq_hz: 0.05` vs node 1 (.101)
-`6.30` — a 126× split in the same room. Pointed to stale gain-lock on
-node 2 from a different AP/orientation. Cleared via
-`POST /ota/recalibrate` (ADR-109) — sensor re-runs the 300-packet
-calibration sampler at next boot.
-
-### Correctness issues (server auditor)
-
-* `run_wiflow_inference` hardcoded keypoint `confidence: 1.0` — lied about
-  data quality. Real signal: the runtime classifier's `confidence`.
-* `wiflow_v1.rs` zero-pad path duplicated subcarrier index 0 instead of
-  zero-padding when < 35 finite subcarriers — comment said "zero the
-  rest", code did the opposite.
-* `nbvi_history.clone()` cloned the entire 600-deep VecDeque (≈270 KB) on
-  every inference, while only the last 20 frames are used.
-* `run_wiflow_inference` picked the node with longest history regardless
-  of recency — stale data from a dead sensor would keep producing pose.
-
-### UI issues (UI auditor)
-
-* `/` served a static API-index HTML page; users typing `localhost:8080`
-  never reached the SPA at `/ui/index.html`.
-* `<section id="sensing">` was empty; `app.js::SensingTab.mount` queried
-  `#sensing-container` and rendered into nothing — the Sensing tab was
-  permanently blank.
-* `LiveDemoTab.fetchModels` unconditionally overwrote `activeModelId =
-  'wiflow-v1'` whenever `/api/v1/info` reported `pose_estimation: true`,
-  even when the operator had just loaded an RVF model. Dropdown silently
-  flipped back to WiFlow on every refresh.
-
-### Docs issues (docs auditor)
-
-* `CHECKLIST.md` header: `head c827cde6`, count `43 Done` — stale
-  by 4 commits and 2 ADRs.
-* `ADR-115 References` cited "ADR-100 — TP-Link WISP" (it's ADR-110)
-  and "ADR-108 / ADR-111" (ADR-111 doesn't exist — folded into ADR-109).
-* `espectre-gap-analysis.md::Still open` table listed 8 items as open
-  that had already shipped (ADR-104, ADR-109, ADR-112, ADR-114).
-* `ota-pipeline.md` documented OTA flashing but never mentioned
-  `/ota/set-target` (ADR-115) or `/ota/recalibrate` (ADR-109) — operator
-  hitting the "Mac moved networks" scenario wouldn't find the recovery
-  path.
+* **Fire 1 — Ping zombies.** `ps` showed 250+ `/sbin/ping -i 0.040`
+  processes — orphans from prior server lifetimes + 8 fresh pings to
+  `127.0.0.1` parented to the current server. Root cause:
+  `cargo test --workspace` sent UDP frames to `127.0.0.1:5005` from
+  `tests/multi_node_test.rs` with `node_ids = [1,2,3,5,7]` — each
+  unique nid registered in `NODE_ADDRS`, keepalive spawned one `ping`
+  child per nid, macOS doesn't propagate parent death.
+* **Fire 2 — Node 2 feature divergence.** `dominant_freq_hz` 0.05 (n2)
+  vs 6.30 (n1), same room, 126×. Stale gain-lock from prior AP geometry.
+  Fixed via `POST /ota/recalibrate` (ADR-109).
+* **Correctness:** hardcoded keypoint `confidence: 1.0`, `wiflow_v1.rs`
+  zero-pad path duplicated subcarrier 0, `nbvi_history.clone()` copied
+  full 600-deep deque per tick, `run_wiflow_inference` ignored node
+  staleness.
+* **UI:** `/` served static API index (SPA was at `/ui/index.html`),
+  `<section id="sensing">` was empty (no `sensing-container` div),
+  `LiveDemoTab.fetchModels` overrode the operator's RVF selection on
+  every poll.
+* **Docs:** `CHECKLIST.md` header stale by 4 commits / 2 ADRs;
+  `ADR-115` cited wrong sister ADRs ("ADR-100" → ADR-110, "ADR-111" → ADR-109);
+  `espectre-gap-analysis.md` listed 8 shipped items as open;
+  `ota-pipeline.md` never documented the post-flash REST endpoints.
 
 ## Decisions
 
@@ -212,25 +174,17 @@ $ curl http://localhost:8080/api/v1/pose/current | jq '.persons[0].keypoints[0]'
 
 ## Out of Scope (intentional non-fixes)
 
-* **Health endpoint fake constants** (cpu:2.5, mem:1.8, disk:15.0) —
-  flagged by the auditor as critical. Replacing with `sysinfo` crate
-  would add a dependency for low-value telemetry; the orchestrator
-  readiness probe today is only used by Docker compose, not Kubernetes
-  liveness. Deferred. Real fix: `/health/ready` only reports
-  `model_loaded` + `node_count > 0`.
-* **`derive_pose_from_sensing` call-site cleanup** — function returns
-  `Vec::new()` since ADR-105; removing the 5 call sites is a no-op
-  refactor with no behaviour change. Skipped to keep diff focused.
-* **`tracker_bridge:10` unused imports warning** — module is integrated
-  via `tracker_bridge::tracker_update` (4 callers), the import list
-  just has dead names. Cosmetic. `cargo fix` deferred.
+* **Health endpoint fake constants** (cpu/mem/disk hardcoded) — adding
+  `sysinfo` crate just for orchestrator telemetry is heavy. Deferred.
+* **`derive_pose_from_sensing` call-site cleanup** — already returns
+  `Vec::new()` (ADR-105); removing 5 call sites is no-op refactor.
+* **`tracker_bridge:10` unused-imports warning** — module is integrated
+  via `tracker_update` (4 callers); import list has dead names. Cosmetic.
 * **CLI training flags** (`--train`, `--dataset`, `--epochs`,
   `--checkpoint-dir`, `--pretrain*`) — silent no-ops; training is via
-  REST. Removing the flags would break any operator script that passes
-  them harmlessly. Deferred to a separate flag-audit pass.
-* **OTA PSK provisioning** — operator workflow change, not a code
-  change. Note added to ADR-115 open items. Operator can set
-  `security/ota_psk` via USB provision.py whenever convenient.
+  REST. Removing flags would break operator scripts. Deferred audit.
+* **OTA PSK provisioning** — workflow change, not a code change.
+  Logged in ADR-115 open items.
 
 ## References
 

docs/adr/ADR-120-windowed-temporal-classifier.md

@@ -176,27 +176,18 @@ docs/adr/ADR-120-windowed-temporal-classifier.md  (this)
 
 ## Out of Scope / Follow-ups
 
-* **Held-out test set** — must record fresh data and evaluate the saved
-  model cold. Critical to confirm 90% is not training-set memorisation.
-* **TCN replacing stacked-MLP** — true 1D convolutions over time would
-  use weights more efficiently (~5k vs 28k) and generalise better.
-  Stack-MLP works but is parameter-heavy. Worth a follow-up if data
-  scales 10×.
-* **Sliding output smoothing** — `classify_window` emits one decision
-  per tick (~10 Hz). Adjacent windows are 19/20 identical, so adjacent
-  predictions should agree. They mostly do (98%+) but flicker at class
-  boundaries — could apply a 3-tick majority filter.
-* **`sitting` vs `standing` split** — both currently merge into
-  `present_still`. The W-MLP gets them both right at 100% as a combined
-  class. Splitting them would test whether temporal RF signatures
-  differ between sitting (chair anchor) and standing (free body).
-* **Class imbalance** — `present_still` has 2× the windows of other
-  classes (sitting + standing both contribute). Acceptable since it's
-  the "neutral" class, but oversampling minority classes might lift
-  accuracy 1-2 pts further.
-* **Smaller window size experiments** — 20 frames = 2 sec at ~10 Hz.
-  Could try 10 frames (1 sec, faster reaction) or 30 (3 sec, more
-  context). 20 was a reasonable first guess.
+* **Held-out test set** — record fresh data, evaluate cold to confirm
+  90% isn't memorisation.
+* **TCN instead of stacked-MLP** — 1D conv over time would use weights
+  more efficiently (~5k vs 28k). Worth pursuing if dataset scales 10×.
+* **Output smoothing** — shipped via two-layer mode+confirm filter on the
+  adaptive output, see ADR-120 follow-up commits.
+* **Split `sitting`/`standing`** — currently merged into `present_still`;
+  separating them would test whether the temporal RF signatures differ.
+* **Class imbalance** — `present_still` has 2× windows; oversampling
+  minority classes might lift accuracy 1-2 pts.
+* **Window size experiments** — 20 frames is a reasonable first guess;
+  10 (faster) or 30 (more context) untested.
 
 ## References
 

docs/adr/ADR-121-mmwave-hlk-ld2402.md

@@ -0,0 +1,159 @@
+# ADR-121 — HLK-LD2402 24 GHz mmWave Radar (auxiliary modality)
+
+**Status**: Accepted (single-modality readout). Fusion deferred.
+**Date**: 2026-05-18
+**Scope**: `v2/crates/wifi-densepose-sensing-server/src/mmwave.rs` (new),
+`Cargo.toml` (serialport dep), `main.rs` (CLI flags `--mmwave-port` /
+`--mmwave-baud`, spawn reader, `mmwave_latest` REST handler, route),
+`ui/components/SensingTab.js` (new card, poll integration).
+
+## Context
+
+The operator has an HLK-LD2402 24 GHz mmWave radar module attached via
+a CP2102 USB-to-UART bridge. Factory firmware emits ASCII
+`distance:<cm>\r\n` lines at 115200 8N1, ~6 Hz, in Normal Mode.
+
+This module is a useful **auxiliary modality**: sub-mm range to a
+moving target, very different physical principle than WiFi CSI, runs
+fully independent. Concrete uses:
+
+1. **Live readout in the UI** — easiest. Operator sees the radar
+   distance alongside the WiFi sensing.
+2. **Vitals ground-truth** — at 6 Hz the data is too slow for HR but
+   captures breathing rate (0.2-0.4 Hz). Compare against the WiFi-CSI
+   vitals detector (ADR-021) for calibration.
+3. **Multi-modal fusion** — feed the mmWave distance + WiFi features
+   into a future classifier. Different physics, very different
+   confusion set — high-value addition.
+
+This ADR ships #1 only. #2 and #3 are follow-ups.
+
+## Decisions
+
+### D1 — Dedicated blocking reader thread, not async
+
+`serialport` is a sync API. Wrapping it with `tokio::spawn_blocking`
+adds overhead for a single-port reader running indefinitely. A
+plain `std::thread` named `mmwave-reader` reads the port, parses
+lines, and writes the latest reading into a global
+`OnceLock<Mutex<Option<MmwaveReading>>>`.
+
+### D2 — Graceful absence
+
+`--mmwave-port` is **optional**. When unset, the server runs as
+before. When set but the port can't be opened, the reader thread
+logs a single warning and exits — server keeps running with WiFi
+sensing only. No retries, no panics. (Operator can hot-plug; if
+auto-reconnect is wanted we add it later.)
+
+### D3 — Stale-after policy
+
+`mmwave::current(staleness)` returns `None` if the most recent
+reading is older than `staleness`. The REST endpoint uses 2 seconds
+— at the module's 6 Hz cadence, 2 s = ~12 missed frames, plenty of
+slack for a brief USB hiccup but tight enough to flag a dead module.
+
+### D4 — Single new REST endpoint, no SensingUpdate change
+
+`GET /api/v1/mmwave/latest` returns:
+
+```json
+{ "available": true, "distance_cm": 152, "age_ms": 90 }
+```
+
+or
+
+```json
+{ "available": false }
+```
+
+Not embedded in `SensingUpdate` because:
+
+* The WS stream is already busy with per-tick CSI broadcasts; a
+  separate poll lets the UI throttle the mmWave refresh
+  independently (saves bandwidth if many clients connect).
+* Keeps the SensingUpdate schema stable — older WS consumers don't
+  need a migration.
+
+UI polls the endpoint once per visible WS tick. ~5-10 Hz refresh.
+
+### D5 — UI badge in `SensingTab`, hidden when unavailable
+
+New card "mmWave Radar (24 GHz)" with a blue badge showing
+`<distance> cm` and an age bar (100 % at 0 ms → 0 % at 2 s). The
+whole card hides via `display: none` when the endpoint reports
+`available: false`, so deployments without the radar see no
+clutter.
+
+### D6 — Parse only the `distance:<n>` Normal Mode format
+
+HLK-LD2402 also has an "Engineering Mode" emitting binary frames
+with per-range-gate energy. Out of scope for v1 — Normal Mode
+covers the live-readout use case. Engineering Mode parsing is a
+separate ADR if/when we need per-gate data for vitals fusion.
+
+## Files Touched
+
+```
+v2/crates/wifi-densepose-sensing-server/Cargo.toml
+  + serialport.workspace = true
+v2/crates/wifi-densepose-sensing-server/src/mmwave.rs   (new, ~130 LoC)
+  + pub struct MmwaveReading { distance_cm: u32, at: Instant }
+  + static LATEST: OnceLock<Mutex<Option<MmwaveReading>>>
+  + pub fn current(staleness) -> Option<MmwaveReading>
+  + pub fn spawn_reader(port, baud)
+  + fn parse_distance(line: &str) -> Option<u32>
+  + 1 unit test
+v2/crates/wifi-densepose-sensing-server/src/lib.rs
+  + pub mod mmwave;
+v2/crates/wifi-densepose-sensing-server/src/main.rs
+  + Args { mmwave_port, mmwave_baud }
+  + spawn_reader call in main()
+  + async fn mmwave_latest
+  + route /api/v1/mmwave/latest
+ui/components/SensingTab.js
+  + #mmwaveCard hidden-by-default card with #mmwaveLabel + age bar
+  + fetch /api/v1/mmwave/latest each visible tick, show/hide card
+docs/adr/ADR-121-mmwave-hlk-ld2402.md  (this)
+```
+
+## Verified Acceptance
+
+Live with the module attached:
+
+```
+$ ./target/release/sensing-server --mmwave-port /dev/cu.usbserial-1140 …
+  ADR-121 mmWave reader: opened /dev/cu.usbserial-1140 @ 115200
+
+$ curl :8080/api/v1/mmwave/latest
+  {"age_ms":55,"available":true,"distance_cm":149}
+  {"age_ms":90,"available":true,"distance_cm":152}
+  {"age_ms":127,"available":true,"distance_cm":153}
+```
+
+Live without module attached (port arg omitted): server starts cleanly,
+endpoint returns `{"available": false}`, Sensing tab card hidden.
+
+## Out of Scope / Follow-ups
+
+* **Engineering Mode binary parser** — needed if we want per-gate
+  energy for vitals (breathing band) or person-counting from
+  per-gate occupancy.
+* **Vitals fusion (ADR-021 cross-check)** — log mmWave breathing
+  rate side-by-side with WiFi-CSI vitals for 5 min, compute
+  Pearson correlation, decide whether to weight one over the other
+  in the final vitals output.
+* **W-MLP feature input** — once vitals fusion proves out, expose
+  mmWave distance as a 23rd feature in the W-MLP and retrain.
+  Would warrant ADR-122.
+* **Auto-reconnect** — current behaviour: open fails or read errors
+  exit the reader thread. Add a retry loop with 2-second backoff
+  if the operator wants USB hot-plug recovery.
+
+## References
+
+* ADR-021 — WiFi-CSI vitals detector (the candidate cross-check
+  partner for HLK-LD2402 breathing-rate output).
+* `assets/sensors/sensor_03.jpeg` / `_04.jpeg` / `_05.jpeg` —
+  hardware photos and inventory entry for the module + CP2102
+  bridge.

docs/architecture.md

@@ -0,0 +1,34 @@
+# RuView · How It Works
+
+Extracted from the main README to keep the landing page short. See
+[`../README.md`](../README.md) for the high-level overview.
+
+---
+
+
+WiFi routers flood every room with radio waves. When a person moves — or even breathes — those waves scatter differently. WiFi DensePose reads that scattering pattern and reconstructs what happened:
+
+```
+WiFi Router → radio waves pass through room → hit human body → scatter
+    ↓
+ESP32 mesh (4-6 nodes) captures CSI on channels 1/6/11 via TDM protocol
+    ↓
+Multi-Band Fusion: 3 channels × 56 subcarriers = 168 virtual subcarriers per link
+    ↓
+Multistatic Fusion: N×(N-1) links → attention-weighted cross-viewpoint embedding
+    ↓
+Coherence Gate: accept/reject measurements → stable for days without tuning
+    ↓
+Signal Processing: Hampel, SpotFi, Fresnel, BVP, spectrogram → clean features
+    ↓
+AI Backbone (RuVector): attention, graph algorithms, compression, field model
+    ↓
+Signal-Line Protocol (CRV): 6-stage gestalt → sensory → topology → coherence → search → model
+    ↓
+Neural Network: processed signals → 17 body keypoints + vital signs + room model
+    ↓
+Output: real-time pose, breathing, heart rate, room fingerprint, drift alerts
+```
+
+No training cameras required — the [Self-Learning system (ADR-024)](adr/ADR-024-contrastive-csi-embedding-model.md) bootstraps from raw WiFi data alone. [MERIDIAN (ADR-027)](adr/ADR-027-cross-environment-domain-generalization.md) ensures the model works in any room, not just the one it trained in.
+

docs/claude-swarm.md

@@ -0,0 +1,82 @@
+# RuView · Claude-Flow Swarm Handbook
+
+V3 CLI commands, agent types, memory operations and configuration
+extracted from `CLAUDE.md`. Active when working through claude-flow's
+swarm coordination tools.
+
+## V3 CLI Commands
+
+### Core Commands
+
+| Command | Subcommands | Description |
+|---------|-------------|-------------|
+| `init` | 4 | Project initialization |
+| `agent` | 8 | Agent lifecycle management |
+| `swarm` | 6 | Multi-agent swarm coordination |
+| `memory` | 11 | AgentDB memory with HNSW search |
+| `task` | 6 | Task creation and lifecycle |
+| `session` | 7 | Session state management |
+| `hooks` | 17 | Self-learning hooks + 12 workers |
+| `hive-mind` | 6 | Byzantine fault-tolerant consensus |
+
+### Quick CLI Examples
+
+```bash
+npx @claude-flow/cli@latest init --wizard
+npx @claude-flow/cli@latest agent spawn -t coder --name my-coder
+npx @claude-flow/cli@latest swarm init --v3-mode
+npx @claude-flow/cli@latest memory search --query "authentication patterns"
+npx @claude-flow/cli@latest doctor --fix
+```
+
+## Available Agents (60+ Types)
+
+### Core Development
+`coder`, `reviewer`, `tester`, `planner`, `researcher`
+
+### Specialized
+`security-architect`, `security-auditor`, `memory-specialist`, `performance-engineer`
+
+### Swarm Coordination
+`hierarchical-coordinator`, `mesh-coordinator`, `adaptive-coordinator`
+
+### GitHub & Repository
+`pr-manager`, `code-review-swarm`, `issue-tracker`, `release-manager`
+
+### SPARC Methodology
+`sparc-coord`, `sparc-coder`, `specification`, `pseudocode`, `architecture`
+
+## Memory Commands Reference
+
+```bash
+# Store (REQUIRED: --key, --value; OPTIONAL: --namespace, --ttl, --tags)
+npx @claude-flow/cli@latest memory store --key "pattern-auth" --value "JWT with refresh" --namespace patterns
+
+# Search (REQUIRED: --query; OPTIONAL: --namespace, --limit, --threshold)
+npx @claude-flow/cli@latest memory search --query "authentication patterns"
+
+# List (OPTIONAL: --namespace, --limit)
+npx @claude-flow/cli@latest memory list --namespace patterns --limit 10
+
+# Retrieve (REQUIRED: --key; OPTIONAL: --namespace)
+npx @claude-flow/cli@latest memory retrieve --key "pattern-auth" --namespace patterns
+```
+
+## Quick Setup
+
+```bash
+claude mcp add claude-flow -- npx -y @claude-flow/cli@latest
+npx @claude-flow/cli@latest daemon start
+npx @claude-flow/cli@latest doctor --fix
+```
+
+## Claude Code vs CLI Tools
+
+- Claude Code's Task tool handles ALL execution: agents, file ops, code generation, git
+- CLI tools handle coordination via Bash: swarm init, memory, hooks, routing
+- NEVER use CLI tools as a substitute for Task tool agents
+
+## Support
+
+- Documentation: https://github.com/ruvnet/claude-flow
+- Issues: https://github.com/ruvnet/claude-flow/issues

docs/dev-handbook.md

@@ -0,0 +1,212 @@
+# RuView · Developer Handbook
+
+All the detailed crate maps, module tables, build commands, firmware
+flashing recipes, publishing order, witness verification, swarm tooling
+and memory commands extracted from `CLAUDE.md` to keep that file
+under 200 lines. CLAUDE.md retains only the rules-of-engagement
+sections; everything else lives here.
+
+---
+
+## Rust Workspace (v2/)
+
+### Key Rust Crates
+| Crate | Description |
+|-------|-------------|
+| `wifi-densepose-core` | Core types, traits, error types, CSI frame primitives |
+| `wifi-densepose-signal` | SOTA signal processing + RuvSense multistatic sensing (14 modules) |
+| `wifi-densepose-nn` | Neural network inference (ONNX, PyTorch, Candle backends) |
+| `wifi-densepose-train` | Training pipeline with ruvector integration + ruview_metrics |
+| `wifi-densepose-mat` | Mass Casualty Assessment Tool — disaster survivor detection |
+| `wifi-densepose-hardware` | ESP32 aggregator, TDM protocol, channel hopping firmware |
+| `wifi-densepose-ruvector` | RuVector v2.0.4 integration + cross-viewpoint fusion (5 modules) |
+| `wifi-densepose-api` | REST API (Axum) |
+| `wifi-densepose-db` | Database layer (Postgres, SQLite, Redis) |
+| `wifi-densepose-config` | Configuration management |
+| `wifi-densepose-wasm` | WebAssembly bindings for browser deployment |
+| `wifi-densepose-cli` | CLI tool (`wifi-densepose` binary) |
+| `wifi-densepose-sensing-server` | Lightweight Axum server for WiFi sensing UI |
+| `wifi-densepose-wifiscan` | Multi-BSSID WiFi scanning (ADR-022) |
+| `wifi-densepose-vitals` | ESP32 CSI-grade vital sign extraction (ADR-021) |
+| `nvsim` | Deterministic NV-diamond magnetometer pipeline simulator (ADR-089) — standalone leaf, WASM-ready |
+| `vendor/rvcsi` (submodule) | **rvCSI** — edge RF sensing runtime (ADR-095/096): 9 crates (`rvcsi-core`/`-dsp`/`-events`/`-adapter-file`/`-adapter-nexmon`/`-ruvector`/`-runtime`/`-node`/`-cli`). Lives in its own repo ([github.com/ruvnet/rvcsi](https://github.com/ruvnet/rvcsi)), vendored here under `vendor/rvcsi`, published to crates.io as `rvcsi-* 0.3.x` and to npm as `@ruv/rvcsi`. Not a `v2/` workspace member — depend on the published crates (or the submodule's `crates/rvcsi-*` paths). Normalized `CsiFrame`/`CsiWindow`/`CsiEvent` schema, validate-before-FFI, reusable DSP, typed confidence-scored events, the napi-c Nexmon shim (real nexmon_csi `.pcap` from a Raspberry Pi 5 / 4 / 3B+ — BCM43455c0), the napi-rs SDK, the `rvcsi` CLI, a Claude Code plugin. |
+
+### RuvSense Modules (`signal/src/ruvsense/`)
+| Module | Purpose |
+|--------|---------|
+| `multiband.rs` | Multi-band CSI frame fusion, cross-channel coherence |
+| `phase_align.rs` | Iterative LO phase offset estimation, circular mean |
+| `multistatic.rs` | Attention-weighted fusion, geometric diversity |
+| `coherence.rs` | Z-score coherence scoring, DriftProfile |
+| `coherence_gate.rs` | Accept/PredictOnly/Reject/Recalibrate gate decisions |
+| `pose_tracker.rs` | 17-keypoint Kalman tracker with AETHER re-ID embeddings |
+| `field_model.rs` | SVD room eigenstructure, perturbation extraction |
+| `tomography.rs` | RF tomography, ISTA L1 solver, voxel grid |
+| `longitudinal.rs` | Welford stats, biomechanics drift detection |
+| `intention.rs` | Pre-movement lead signals (200-500ms) |
+| `cross_room.rs` | Environment fingerprinting, transition graph |
+| `gesture.rs` | DTW template matching gesture classifier |
+| `adversarial.rs` | Physically impossible signal detection, multi-link consistency |
+
+### Cross-Viewpoint Fusion (`ruvector/src/viewpoint/`)
+| Module | Purpose |
+|--------|---------|
+| `attention.rs` | CrossViewpointAttention, GeometricBias, softmax with G_bias |
+| `geometry.rs` | GeometricDiversityIndex, Cramer-Rao bounds, Fisher Information |
+| `coherence.rs` | Phase phasor coherence, hysteresis gate |
+| `fusion.rs` | MultistaticArray aggregate root, domain events |
+
+### RuVector v2.0.4 Integration (ADR-016 complete, ADR-017 proposed)
+All 5 ruvector crates integrated in workspace:
+- `ruvector-mincut` → `metrics.rs` (DynamicPersonMatcher) + `subcarrier_selection.rs`
+- `ruvector-attn-mincut` → `model.rs` (apply_antenna_attention) + `spectrogram.rs`
+- `ruvector-temporal-tensor` → `dataset.rs` (CompressedCsiBuffer) + `breathing.rs`
+- `ruvector-solver` → `subcarrier.rs` (sparse interpolation 114→56) + `triangulation.rs`
+- `ruvector-attention` → `model.rs` (apply_spatial_attention) + `bvp.rs`
+
+
+## Architecture Decisions, Hardware, Build Commands
+
+### Architecture Decisions
+43 ADRs in `docs/adr/` (ADR-001 through ADR-043). Key ones:
+- ADR-014: SOTA signal processing (Accepted)
+- ADR-015: MM-Fi + Wi-Pose training datasets (Accepted)
+- ADR-016: RuVector training pipeline integration (Accepted — complete)
+- ADR-017: RuVector signal + MAT integration (Proposed — next target)
+- ADR-024: Contrastive CSI embedding / AETHER (Accepted)
+- ADR-027: Cross-environment domain generalization / MERIDIAN (Accepted)
+- ADR-028: ESP32 capability audit + witness verification (Accepted)
+- ADR-029: RuvSense multistatic sensing mode (Proposed)
+- ADR-030: RuvSense persistent field model (Proposed)
+- ADR-031: RuView sensing-first RF mode (Proposed)
+- ADR-032: Multistatic mesh security hardening (Proposed)
+
+### Supported Hardware
+
+| Device | Port | Chip | Role | Cost |
+|--------|------|------|------|------|
+| ESP32-S3 (8MB flash) | COM7 | Xtensa dual-core | WiFi CSI sensing node | ~$9 |
+| ESP32-S3 SuperMini (4MB) | — | Xtensa dual-core | WiFi CSI (compact) | ~$6 |
+| ESP32-C6 + Seeed MR60BHA2 | COM4 | RISC-V + 60 GHz FMCW | mmWave HR/BR/presence | ~$15 |
+| HLK-LD2410 | — | 24 GHz FMCW | Presence + distance | ~$3 |
+
+**Not supported:** ESP32 (original), ESP32-C3 — single-core, can't run CSI DSP pipeline.
+
+### Build & Test Commands (this repo)
+```bash
+# Rust — full workspace tests (1,031+ tests, ~2 min)
+cd v2
+cargo test --workspace --no-default-features
+
+# Rust — single crate check (no GPU needed)
+cargo check -p wifi-densepose-train --no-default-features
+
+# Python — deterministic proof verification (SHA-256)
+python archive/v1/data/proof/verify.py
+
+# Python — test suite
+cd archive/v1 && python -m pytest tests/ -x -q
+```
+
+
+## Firmware Build + Flash + Provision + Release + Publish
+
+### ESP32 Firmware Build (Windows — Python subprocess required)
+```bash
+# Build 8MB firmware (real WiFi CSI mode, no mocks)
+# See CLAUDE.local.md for the full Python subprocess command
+# Key: must strip MSYSTEM env vars for ESP-IDF v5.4 on Git Bash
+
+# Build 4MB firmware
+cp sdkconfig.defaults.4mb sdkconfig.defaults
+# then same build process
+
+# Flash to COM7
+# [python, idf_py, '-p', 'COM7', 'flash']
+
+# Provision WiFi
+python firmware/esp32-csi-node/provision.py --port COM7 \
+  --ssid "YourWiFi" --password "secret" --target-ip 192.168.1.20
+
+# Monitor serial
+python -m serial.tools.miniterm COM7 115200
+```
+
+### Firmware Release Process
+1. Build 8MB from `sdkconfig.defaults.template` (no mock)
+2. Build 4MB from `sdkconfig.defaults.4mb` (no mock)
+3. Save 6 binaries: `esp32-csi-node.bin`, `bootloader.bin`, `partition-table.bin`, `ota_data_initial.bin`, `esp32-csi-node-4mb.bin`, `partition-table-4mb.bin`
+4. Tag: `git tag v0.X.Y-esp32 && git push origin v0.X.Y-esp32`
+5. Release: `gh release create v0.X.Y-esp32 <binaries> --title "..." --notes-file ...`
+6. Verify on real hardware (COM7) before publishing
+7. **CRITICAL:** Always test with real WiFi CSI, not mock mode — mock missed the Kconfig threshold bug
+
+### Crate Publishing Order
+Crates must be published in dependency order:
+1. `wifi-densepose-core` (no internal deps)
+2. `wifi-densepose-vitals` (no internal deps)
+3. `wifi-densepose-wifiscan` (no internal deps)
+4. `wifi-densepose-hardware` (no internal deps)
+5. `wifi-densepose-config` (no internal deps)
+6. `wifi-densepose-db` (no internal deps)
+7. `wifi-densepose-signal` (depends on core)
+8. `wifi-densepose-nn` (no internal deps, workspace only)
+9. `wifi-densepose-ruvector` (no internal deps, workspace only)
+10. `wifi-densepose-train` (depends on signal, nn)
+11. `wifi-densepose-mat` (depends on core, signal, nn)
+12. `wifi-densepose-api` (no internal deps)
+13. `wifi-densepose-wasm` (depends on mat)
+14. `wifi-densepose-sensing-server` (depends on wifiscan)
+15. `wifi-densepose-cli` (depends on mat)
+
+
+## Validation & Witness Verification (ADR-028)
+
+### Validation & Witness Verification (ADR-028)
+
+**After any significant code change, run the full validation:**
+
+```bash
+# 1. Rust tests — must be 1,031+ passed, 0 failed
+cd v2
+cargo test --workspace --no-default-features
+
+# 2. Python proof — must print VERDICT: PASS
+cd ..
+python archive/v1/data/proof/verify.py
+
+# 3. Generate witness bundle (includes both above + firmware hashes)
+bash scripts/generate-witness-bundle.sh
+
+# 4. Self-verify the bundle — must be 7/7 PASS
+cd dist/witness-bundle-ADR028-*/
+bash VERIFY.sh
+```
+
+**If the Python proof hash changes** (e.g., numpy/scipy version update):
+```bash
+# Regenerate the expected hash, then verify it passes
+python archive/v1/data/proof/verify.py --generate-hash
+python archive/v1/data/proof/verify.py
+```
+
+**Witness bundle contents** (`dist/witness-bundle-ADR028-<sha>.tar.gz`):
+- `WITNESS-LOG-028.md` — 33-row attestation matrix with evidence per capability
+- `ADR-028-esp32-capability-audit.md` — Full audit findings
+- `proof/verify.py` + `expected_features.sha256` — Deterministic pipeline proof
+- `test-results/rust-workspace-tests.log` — Full cargo test output
+- `firmware-manifest/source-hashes.txt` — SHA-256 of all 7 ESP32 firmware files
+- `crate-manifest/versions.txt` — All 15 crates with versions
+- `VERIFY.sh` — One-command self-verification for recipients
+
+**Key proof artifacts:**
+- `archive/v1/data/proof/verify.py` — Trust Kill Switch: feeds reference signal through production pipeline, hashes output
+- `archive/v1/data/proof/expected_features.sha256` — Published expected hash
+- `archive/v1/data/proof/sample_csi_data.json` — 1,000 synthetic CSI frames (seed=42)
+- `docs/WITNESS-LOG-028.md` — 11-step reproducible verification procedure
+- `docs/adr/ADR-028-esp32-capability-audit.md` — Complete audit record
+
+### Branch
+Default branch: `main`
+Active feature branch: `ruvsense-full-implementation` (PR #77)
+

docs/use-cases.md

@@ -0,0 +1,347 @@
+# RuView · Use Cases & Applications
+
+This file lists every concrete deployment scenario RuView is designed to
+serve, plus the 60-module ADR-041 Edge Intelligence catalogue and the
+ADR-024 self-learning system. Pulled out of the main README to keep the
+landing page short — see [`../README.md`](../README.md) for the
+high-level overview and quick start.
+
+---
+
+
+WiFi sensing works anywhere WiFi exists. No new hardware in most cases — just software on existing access points or a $8 ESP32 add-on. Because there are no cameras, deployments avoid privacy regulations (GDPR video, HIPAA imaging) by design.
+
+**Scaling:** Each AP distinguishes ~3-5 people (56 subcarriers). Multi-AP multiplies linearly — a 4-AP retail mesh covers ~15-20 occupants. No hard software limit; the practical ceiling is signal physics.
+
+| | Why WiFi sensing wins | Traditional alternative |
+|---|----------------------|----------------------|
+| 🔒 | **No video, no GDPR/HIPAA imaging rules** | Cameras require consent, signage, data retention policies |
+| 🧱 | **Works through walls, shelving, debris** | Cameras need line-of-sight per room |
+| 🌙 | **Works in total darkness** | Cameras need IR or visible light |
+| 💰 | **$0-$8 per zone** (existing WiFi or ESP32) | Camera systems: $200-$2,000 per zone |
+| 🔌 | **WiFi already deployed everywhere** | PIR/radar sensors require new wiring per room |
+
+<details>
+<summary><strong>🏥 Everyday</strong> — Healthcare, retail, office, hospitality (commodity WiFi)</summary>
+
+| Use Case | What It Does | Hardware | Key Metric | Edge Module |
+|----------|-------------|----------|------------|-------------|
+| **Elderly care / assisted living** | Fall detection, nighttime activity monitoring, breathing rate during sleep — no wearable compliance needed | 1 ESP32-S3 per room ($8) | Fall alert <2s | [Sleep Apnea](edge-modules/medical.md), [Gait Analysis](edge-modules/medical.md) |
+| **Hospital patient monitoring** | Continuous breathing + heart rate for non-critical beds without wired sensors; nurse alert on anomaly | 1-2 APs per ward | Breathing: 6-30 BPM | [Respiratory Distress](edge-modules/medical.md), [Cardiac Arrhythmia](edge-modules/medical.md) |
+| **Emergency room triage** | Automated occupancy count + wait-time estimation; detect patient distress (abnormal breathing) in waiting areas | Existing hospital WiFi | Occupancy accuracy >95% | [Queue Length](edge-modules/retail.md), [Panic Motion](edge-modules/security.md) |
+| **Retail occupancy & flow** | Real-time foot traffic, dwell time by zone, queue length — no cameras, no opt-in, GDPR-friendly | Existing store WiFi + 1 ESP32 | Dwell resolution ~1m | [Customer Flow](edge-modules/retail.md), [Dwell Heatmap](edge-modules/retail.md) |
+| **Office space utilization** | Which desks/rooms are actually occupied, meeting room no-shows, HVAC optimization based on real presence | Existing enterprise WiFi | Presence latency <1s | [Meeting Room](edge-modules/building.md), [HVAC Presence](edge-modules/building.md) |
+| **Hotel & hospitality** | Room occupancy without door sensors, minibar/bathroom usage patterns, energy savings on empty rooms | Existing hotel WiFi | 15-30% HVAC savings | [Energy Audit](edge-modules/building.md), [Lighting Zones](edge-modules/building.md) |
+| **Restaurants & food service** | Table turnover tracking, kitchen staff presence, restroom occupancy displays — no cameras in dining areas | Existing WiFi | Queue wait ±30s | [Table Turnover](edge-modules/retail.md), [Queue Length](edge-modules/retail.md) |
+| **Parking garages** | Pedestrian presence in stairwells and elevators where cameras have blind spots; security alert if someone lingers | Existing WiFi | Through-concrete walls | [Loitering](edge-modules/security.md), [Elevator Count](edge-modules/building.md) |
+
+</details>
+
+<details>
+<summary><strong>🏟️ Specialized</strong> — Events, fitness, education, civic (CSI-capable hardware)</summary>
+
+| Use Case | What It Does | Hardware | Key Metric | Edge Module |
+|----------|-------------|----------|------------|-------------|
+| **Smart home automation** | Room-level presence triggers (lights, HVAC, music) that work through walls — no dead zones, no motion-sensor timeouts | 2-3 ESP32-S3 nodes ($24) | Through-wall range ~5m | [HVAC Presence](edge-modules/building.md), [Lighting Zones](edge-modules/building.md) |
+| **Fitness & sports** | Rep counting, posture correction, breathing cadence during exercise — no wearable, no camera in locker rooms | 3+ ESP32-S3 mesh | Pose: 17 keypoints | [Breathing Sync](edge-modules/exotic.md), [Gait Analysis](edge-modules/medical.md) |
+| **Childcare & schools** | Naptime breathing monitoring, playground headcount, restricted-area alerts — privacy-safe for minors | 2-4 ESP32-S3 per zone | Breathing: ±1 BPM | [Sleep Apnea](edge-modules/medical.md), [Perimeter Breach](edge-modules/security.md) |
+| **Event venues & concerts** | Crowd density mapping, crush-risk detection via breathing compression, emergency evacuation flow tracking | Multi-AP mesh (4-8 APs) | Density per m² | [Customer Flow](edge-modules/retail.md), [Panic Motion](edge-modules/security.md) |
+| **Stadiums & arenas** | Section-level occupancy for dynamic pricing, concession staffing, emergency egress flow modeling | Enterprise AP grid | 15-20 per AP mesh | [Dwell Heatmap](edge-modules/retail.md), [Queue Length](edge-modules/retail.md) |
+| **Houses of worship** | Attendance counting without facial recognition — privacy-sensitive congregations, multi-room campus tracking | Existing WiFi | Zone-level accuracy | [Elevator Count](edge-modules/building.md), [Energy Audit](edge-modules/building.md) |
+| **Warehouse & logistics** | Worker safety zones, forklift proximity alerts, occupancy in hazardous areas — works through shelving and pallets | Industrial AP mesh | Alert latency <500ms | [Forklift Proximity](edge-modules/industrial.md), [Confined Space](edge-modules/industrial.md) |
+| **Civic infrastructure** | Public restroom occupancy (no cameras possible), subway platform crowding, shelter headcount during emergencies | Municipal WiFi + ESP32 | Real-time headcount | [Customer Flow](edge-modules/retail.md), [Loitering](edge-modules/security.md) |
+| **Museums & galleries** | Visitor flow heatmaps, exhibit dwell time, crowd bottleneck alerts — no cameras near artwork (flash/theft risk) | Existing WiFi | Zone dwell ±5s | [Dwell Heatmap](edge-modules/retail.md), [Shelf Engagement](edge-modules/retail.md) |
+
+</details>
+
+<details>
+<summary><strong>🤖 Robotics & Industrial</strong> — Autonomous systems, manufacturing, android spatial awareness</summary>
+
+WiFi sensing gives robots and autonomous systems a spatial awareness layer that works where LIDAR and cameras fail — through dust, smoke, fog, and around corners. The CSI signal field acts as a "sixth sense" for detecting humans in the environment without requiring line-of-sight.
+
+| Use Case | What It Does | Hardware | Key Metric | Edge Module |
+|----------|-------------|----------|------------|-------------|
+| **Cobot safety zones** | Detect human presence near collaborative robots — auto-slow or stop before contact, even behind obstructions | 2-3 ESP32-S3 per cell | Presence latency <100ms | [Forklift Proximity](edge-modules/industrial.md), [Perimeter Breach](edge-modules/security.md) |
+| **Warehouse AMR navigation** | Autonomous mobile robots sense humans around blind corners, through shelving racks — no LIDAR occlusion | ESP32 mesh along aisles | Through-shelf detection | [Forklift Proximity](edge-modules/industrial.md), [Loitering](edge-modules/security.md) |
+| **Android / humanoid spatial awareness** | Ambient human pose sensing for social robots — detect gestures, approach direction, and personal space without cameras always on | Onboard ESP32-S3 module | 17-keypoint pose | [Gesture Language](edge-modules/exotic.md), [Emotion Detection](edge-modules/exotic.md) |
+| **Manufacturing line monitoring** | Worker presence at each station, ergonomic posture alerts, headcount for shift compliance — works through equipment | Industrial AP per zone | Pose + breathing | [Confined Space](edge-modules/industrial.md), [Gait Analysis](edge-modules/medical.md) |
+| **Construction site safety** | Exclusion zone enforcement around heavy machinery, fall detection from scaffolding, personnel headcount | Ruggedized ESP32 mesh | Alert <2s, through-dust | [Panic Motion](edge-modules/security.md), [Structural Vibration](edge-modules/industrial.md) |
+| **Agricultural robotics** | Detect farm workers near autonomous harvesters in dusty/foggy field conditions where cameras are unreliable | Weatherproof ESP32 nodes | Range ~10m open field | [Forklift Proximity](edge-modules/industrial.md), [Rain Detection](edge-modules/exotic.md) |
+| **Drone landing zones** | Verify landing area is clear of humans — WiFi sensing works in rain, dust, and low light where downward cameras fail | Ground ESP32 nodes | Presence: >95% accuracy | [Perimeter Breach](edge-modules/security.md), [Tailgating](edge-modules/security.md) |
+| **Clean room monitoring** | Personnel tracking without cameras (particle contamination risk from camera fans) — gown compliance via pose | Existing cleanroom WiFi | No particulate emission | [Clean Room](edge-modules/industrial.md), [Livestock Monitor](edge-modules/industrial.md) |
+
+</details>
+
+<details>
+<summary><strong>🔥 Extreme</strong> — Through-wall, disaster, defense, underground</summary>
+
+These scenarios exploit WiFi's ability to penetrate solid materials — concrete, rubble, earth — where no optical or infrared sensor can reach. The WiFi-Mat disaster module (ADR-001) is specifically designed for this tier.
+
+| Use Case | What It Does | Hardware | Key Metric | Edge Module |
+|----------|-------------|----------|------------|-------------|
+| **Search & rescue (WiFi-Mat)** | Detect survivors through rubble/debris via breathing signature, START triage color classification, 3D localization | Portable ESP32 mesh + laptop | Through 30cm concrete | [Respiratory Distress](edge-modules/medical.md), [Seizure Detection](edge-modules/medical.md) |
+| **Firefighting** | Locate occupants through smoke and walls before entry; breathing detection confirms life signs remotely | Portable mesh on truck | Works in zero visibility | [Sleep Apnea](edge-modules/medical.md), [Panic Motion](edge-modules/security.md) |
+| **Prison & secure facilities** | Cell occupancy verification, distress detection (abnormal vitals), perimeter sensing — no camera blind spots | Dedicated AP infrastructure | 24/7 vital signs | [Cardiac Arrhythmia](edge-modules/medical.md), [Loitering](edge-modules/security.md) |
+| **Military / tactical** | Through-wall personnel detection, room clearing confirmation, hostage vital signs at standoff distance | Directional WiFi + custom FW | Range: 5m through wall | [Perimeter Breach](edge-modules/security.md), [Weapon Detection](edge-modules/security.md) |
+| **Border & perimeter security** | Detect human presence in tunnels, behind fences, in vehicles — passive sensing, no active illumination to reveal position | Concealed ESP32 mesh | Passive / covert | [Perimeter Breach](edge-modules/security.md), [Tailgating](edge-modules/security.md) |
+| **Mining & underground** | Worker presence in tunnels where GPS/cameras fail, breathing detection after collapse, headcount at safety points | Ruggedized ESP32 mesh | Through rock/earth | [Confined Space](edge-modules/industrial.md), [Respiratory Distress](edge-modules/medical.md) |
+| **Maritime & naval** | Below-deck personnel tracking through steel bulkheads (limited range, requires tuning), man-overboard detection | Ship WiFi + ESP32 | Through 1-2 bulkheads | [Structural Vibration](edge-modules/industrial.md), [Panic Motion](edge-modules/security.md) |
+| **Wildlife research** | Non-invasive animal activity monitoring in enclosures or dens — no light pollution, no visual disturbance | Weatherproof ESP32 nodes | Zero light emission | [Livestock Monitor](edge-modules/industrial.md), [Dream Stage](edge-modules/exotic.md) |
+
+</details>
+
+<details>
+<summary><strong>🧩 Edge Intelligence (<a href="docs/adr/ADR-041-wasm-module-collection.md">ADR-041</a>)</strong> — 60 WASM modules across 13 categories, all implemented (609 tests)</summary>
+
+Small programs that run directly on the ESP32 sensor — no internet needed, no cloud fees, instant response. Each module is a tiny WASM file (5-30 KB) that you upload to the device over-the-air. It reads WiFi signal data and makes decisions locally in under 10 ms. [ADR-041](adr/ADR-041-wasm-module-collection.md) defines 60 modules across 13 categories — all 60 are implemented with 609 tests passing.
+
+| | Category | Examples |
+|---|----------|---------|
+| 🏥 | [**Medical & Health**](edge-modules/medical.md) | Sleep apnea detection, cardiac arrhythmia, gait analysis, seizure detection |
+| 🔐 | [**Security & Safety**](edge-modules/security.md) | Intrusion detection, perimeter breach, loitering, panic motion |
+| 🏢 | [**Smart Building**](edge-modules/building.md) | Zone occupancy, HVAC control, elevator counting, meeting room tracking |
+| 🛒 | [**Retail & Hospitality**](edge-modules/retail.md) | Queue length, dwell heatmaps, customer flow, table turnover |
+| 🏭 | [**Industrial**](edge-modules/industrial.md) | Forklift proximity, confined space monitoring, structural vibration |
+| 🔮 | [**Exotic & Research**](edge-modules/exotic.md) | Sleep staging, emotion detection, sign language, breathing sync |
+| 📡 | [**Signal Intelligence**](edge-modules/signal-intelligence.md) | Cleans and sharpens raw WiFi signals — focuses on important regions, filters noise, fills in missing data, and tracks which person is which |
+| 🧠 | [**Adaptive Learning**](edge-modules/adaptive-learning.md) | The sensor learns new gestures and patterns on its own over time — no cloud needed, remembers what it learned even after updates |
+| 🗺️ | [**Spatial Reasoning**](edge-modules/spatial-temporal.md) | Figures out where people are in a room, which zones matter most, and tracks movement across areas using graph-based spatial logic |
+| ⏱️ | [**Temporal Analysis**](edge-modules/spatial-temporal.md) | Learns daily routines, detects when patterns break (someone didn't get up), and verifies safety rules are being followed over time |
+| 🛡️ | [**AI Security**](edge-modules/ai-security.md) | Detects signal replay attacks, WiFi jamming, injection attempts, and flags abnormal behavior that could indicate tampering |
+| ⚛️ | [**Quantum-Inspired**](edge-modules/autonomous.md) | Uses quantum-inspired math to map room-wide signal coherence and search for optimal sensor configurations |
+| 🤖 | [**Autonomous & Exotic**](edge-modules/autonomous.md) | Self-managing sensor mesh — auto-heals dropped nodes, plans its own actions, and explores experimental signal representations |
+
+All implemented modules are `no_std` Rust, share a [common utility library](../v2/crates/wifi-densepose-wasm-edge/src/vendor_common.rs), and talk to the host through a 12-function API. Full documentation: [**Edge Modules Guide**](edge-modules/README.md). See the [complete implemented module list](#edge-module-list) below.
+
+</details>
+
+<details id="edge-module-list">
+<summary><strong>🧩 Edge Intelligence — <a href="docs/edge-modules/README.md">All 65 Modules Implemented</a></strong> (ADR-041 complete)</summary>
+
+All 60 modules are implemented, tested (609 tests passing), and ready to deploy. They compile to `wasm32-unknown-unknown`, run on ESP32-S3 via WASM3, and share a [common utility library](../v2/crates/wifi-densepose-wasm-edge/src/vendor_common.rs). Source: [`crates/wifi-densepose-wasm-edge/src/`](../v2/crates/wifi-densepose-wasm-edge/src/)
+
+**Core modules** (ADR-040 flagship + early implementations):
+
+| Module | File | What It Does |
+|--------|------|-------------|
+| Gesture Classifier | [`gesture.rs`](../v2/crates/wifi-densepose-wasm-edge/src/gesture.rs) | DTW template matching for hand gestures |
+| Coherence Filter | [`coherence.rs`](../v2/crates/wifi-densepose-wasm-edge/src/coherence.rs) | Phase coherence gating for signal quality |
+| Adversarial Detector | [`adversarial.rs`](../v2/crates/wifi-densepose-wasm-edge/src/adversarial.rs) | Detects physically impossible signal patterns |
+| Intrusion Detector | [`intrusion.rs`](../v2/crates/wifi-densepose-wasm-edge/src/intrusion.rs) | Human vs non-human motion classification |
+| Occupancy Counter | [`occupancy.rs`](../v2/crates/wifi-densepose-wasm-edge/src/occupancy.rs) | Zone-level person counting |
+| Vital Trend | [`vital_trend.rs`](../v2/crates/wifi-densepose-wasm-edge/src/vital_trend.rs) | Long-term breathing and heart rate trending |
+| RVF Parser | [`rvf.rs`](../v2/crates/wifi-densepose-wasm-edge/src/rvf.rs) | RVF container format parsing |
+
+**Vendor-integrated modules** (24 modules, ADR-041 Category 7):
+
+**📡 Signal Intelligence** — Real-time CSI analysis and feature extraction
+
+| Module | File | What It Does | Budget |
+|--------|------|-------------|--------|
+| Flash Attention | [`sig_flash_attention.rs`](../v2/crates/wifi-densepose-wasm-edge/src/sig_flash_attention.rs) | Tiled attention over 8 subcarrier groups — finds spatial focus regions and entropy | S (<5ms) |
+| Coherence Gate | [`sig_coherence_gate.rs`](../v2/crates/wifi-densepose-wasm-edge/src/sig_coherence_gate.rs) | Z-score phasor gating with hysteresis: Accept / PredictOnly / Reject / Recalibrate | L (<2ms) |
+| Temporal Compress | [`sig_temporal_compress.rs`](../v2/crates/wifi-densepose-wasm-edge/src/sig_temporal_compress.rs) | 3-tier adaptive quantization (8-bit hot / 5-bit warm / 3-bit cold) | L (<2ms) |
+| Sparse Recovery | [`sig_sparse_recovery.rs`](../v2/crates/wifi-densepose-wasm-edge/src/sig_sparse_recovery.rs) | ISTA L1 reconstruction for dropped subcarriers | H (<10ms) |
+| Person Match | [`sig_mincut_person_match.rs`](../v2/crates/wifi-densepose-wasm-edge/src/sig_mincut_person_match.rs) | Hungarian-lite bipartite assignment for multi-person tracking | S (<5ms) |
+| Optimal Transport | [`sig_optimal_transport.rs`](../v2/crates/wifi-densepose-wasm-edge/src/sig_optimal_transport.rs) | Sliced Wasserstein-1 distance with 4 projections | L (<2ms) |
+
+**🧠 Adaptive Learning** — On-device learning without cloud connectivity
+
+| Module | File | What It Does | Budget |
+|--------|------|-------------|--------|
+| DTW Gesture Learn | [`lrn_dtw_gesture_learn.rs`](../v2/crates/wifi-densepose-wasm-edge/src/lrn_dtw_gesture_learn.rs) | User-teachable gesture recognition — 3-rehearsal protocol, 16 templates | S (<5ms) |
+| Anomaly Attractor | [`lrn_anomaly_attractor.rs`](../v2/crates/wifi-densepose-wasm-edge/src/lrn_anomaly_attractor.rs) | 4D dynamical system attractor classification with Lyapunov exponents | H (<10ms) |
+| Meta Adapt | [`lrn_meta_adapt.rs`](../v2/crates/wifi-densepose-wasm-edge/src/lrn_meta_adapt.rs) | Hill-climbing self-optimization with safety rollback | L (<2ms) |
+| EWC Lifelong | [`lrn_ewc_lifelong.rs`](../v2/crates/wifi-densepose-wasm-edge/src/lrn_ewc_lifelong.rs) | Elastic Weight Consolidation — remembers past tasks while learning new ones | S (<5ms) |
+
+**🗺️ Spatial Reasoning** — Location, proximity, and influence mapping
+
+| Module | File | What It Does | Budget |
+|--------|------|-------------|--------|
+| PageRank Influence | [`spt_pagerank_influence.rs`](../v2/crates/wifi-densepose-wasm-edge/src/spt_pagerank_influence.rs) | 4x4 cross-correlation graph with power iteration PageRank | L (<2ms) |
+| Micro HNSW | [`spt_micro_hnsw.rs`](../v2/crates/wifi-densepose-wasm-edge/src/spt_micro_hnsw.rs) | 64-vector navigable small-world graph for nearest-neighbor search | S (<5ms) |
+| Spiking Tracker | [`spt_spiking_tracker.rs`](../v2/crates/wifi-densepose-wasm-edge/src/spt_spiking_tracker.rs) | 32 LIF neurons + 4 output zone neurons with STDP learning | S (<5ms) |
+
+**⏱️ Temporal Analysis** — Activity patterns, logic verification, autonomous planning
+
+| Module | File | What It Does | Budget |
+|--------|------|-------------|--------|
+| Pattern Sequence | [`tmp_pattern_sequence.rs`](../v2/crates/wifi-densepose-wasm-edge/src/tmp_pattern_sequence.rs) | Activity routine detection and deviation alerts | S (<5ms) |
+| Temporal Logic Guard | [`tmp_temporal_logic_guard.rs`](../v2/crates/wifi-densepose-wasm-edge/src/tmp_temporal_logic_guard.rs) | LTL formula verification on CSI event streams | S (<5ms) |
+| GOAP Autonomy | [`tmp_goap_autonomy.rs`](../v2/crates/wifi-densepose-wasm-edge/src/tmp_goap_autonomy.rs) | Goal-Oriented Action Planning for autonomous module management | S (<5ms) |
+
+**🛡️ AI Security** — Tamper detection and behavioral anomaly profiling
+
+| Module | File | What It Does | Budget |
+|--------|------|-------------|--------|
+| Prompt Shield | [`ais_prompt_shield.rs`](../v2/crates/wifi-densepose-wasm-edge/src/ais_prompt_shield.rs) | FNV-1a replay detection, injection detection (10x amplitude), jamming (SNR) | L (<2ms) |
+| Behavioral Profiler | [`ais_behavioral_profiler.rs`](../v2/crates/wifi-densepose-wasm-edge/src/ais_behavioral_profiler.rs) | 6D behavioral profile with Mahalanobis anomaly scoring | S (<5ms) |
+
+**⚛️ Quantum-Inspired** — Quantum computing metaphors applied to CSI analysis
+
+| Module | File | What It Does | Budget |
+|--------|------|-------------|--------|
+| Quantum Coherence | [`qnt_quantum_coherence.rs`](../v2/crates/wifi-densepose-wasm-edge/src/qnt_quantum_coherence.rs) | Bloch sphere mapping, Von Neumann entropy, decoherence detection | S (<5ms) |
+| Interference Search | [`qnt_interference_search.rs`](../v2/crates/wifi-densepose-wasm-edge/src/qnt_interference_search.rs) | 16 room-state hypotheses with Grover-inspired oracle + diffusion | S (<5ms) |
+
+**🤖 Autonomous Systems** — Self-governing and self-healing behaviors
+
+| Module | File | What It Does | Budget |
+|--------|------|-------------|--------|
+| Psycho-Symbolic | [`aut_psycho_symbolic.rs`](../v2/crates/wifi-densepose-wasm-edge/src/aut_psycho_symbolic.rs) | 16-rule forward-chaining knowledge base with contradiction detection | S (<5ms) |
+| Self-Healing Mesh | [`aut_self_healing_mesh.rs`](../v2/crates/wifi-densepose-wasm-edge/src/aut_self_healing_mesh.rs) | 8-node mesh with health tracking, degradation/recovery, coverage healing | S (<5ms) |
+
+**🔮 Exotic (Vendor)** — Novel mathematical models for CSI interpretation
+
+| Module | File | What It Does | Budget |
+|--------|------|-------------|--------|
+| Time Crystal | [`exo_time_crystal.rs`](../v2/crates/wifi-densepose-wasm-edge/src/exo_time_crystal.rs) | Autocorrelation subharmonic detection in 256-frame history | S (<5ms) |
+| Hyperbolic Space | [`exo_hyperbolic_space.rs`](../v2/crates/wifi-densepose-wasm-edge/src/exo_hyperbolic_space.rs) | Poincare ball embedding with 32 reference locations, hyperbolic distance | S (<5ms) |
+
+**🏥 Medical & Health** (Category 1) — Contactless health monitoring
+
+| Module | File | What It Does | Budget |
+|--------|------|-------------|--------|
+| Sleep Apnea | [`med_sleep_apnea.rs`](../v2/crates/wifi-densepose-wasm-edge/src/med_sleep_apnea.rs) | Detects breathing pauses during sleep | S (<5ms) |
+| Cardiac Arrhythmia | [`med_cardiac_arrhythmia.rs`](../v2/crates/wifi-densepose-wasm-edge/src/med_cardiac_arrhythmia.rs) | Monitors heart rate for irregular rhythms | S (<5ms) |
+| Respiratory Distress | [`med_respiratory_distress.rs`](../v2/crates/wifi-densepose-wasm-edge/src/med_respiratory_distress.rs) | Alerts on abnormal breathing patterns | S (<5ms) |
+| Gait Analysis | [`med_gait_analysis.rs`](../v2/crates/wifi-densepose-wasm-edge/src/med_gait_analysis.rs) | Tracks walking patterns and detects changes | S (<5ms) |
+| Seizure Detection | [`med_seizure_detect.rs`](../v2/crates/wifi-densepose-wasm-edge/src/med_seizure_detect.rs) | 6-state machine for tonic-clonic seizure recognition | S (<5ms) |
+
+**🔐 Security & Safety** (Category 2) — Perimeter and threat detection
+
+| Module | File | What It Does | Budget |
+|--------|------|-------------|--------|
+| Perimeter Breach | [`sec_perimeter_breach.rs`](../v2/crates/wifi-densepose-wasm-edge/src/sec_perimeter_breach.rs) | Detects boundary crossings with approach/departure | S (<5ms) |
+| Weapon Detection | [`sec_weapon_detect.rs`](../v2/crates/wifi-densepose-wasm-edge/src/sec_weapon_detect.rs) | Metal anomaly detection via CSI amplitude shifts | S (<5ms) |
+| Tailgating | [`sec_tailgating.rs`](../v2/crates/wifi-densepose-wasm-edge/src/sec_tailgating.rs) | Detects unauthorized follow-through at access points | S (<5ms) |
+| Loitering | [`sec_loitering.rs`](../v2/crates/wifi-densepose-wasm-edge/src/sec_loitering.rs) | Alerts when someone lingers too long in a zone | S (<5ms) |
+| Panic Motion | [`sec_panic_motion.rs`](../v2/crates/wifi-densepose-wasm-edge/src/sec_panic_motion.rs) | Detects fleeing, struggling, or panic movement | S (<5ms) |
+
+**🏢 Smart Building** (Category 3) — Automation and energy efficiency
+
+| Module | File | What It Does | Budget |
+|--------|------|-------------|--------|
+| HVAC Presence | [`bld_hvac_presence.rs`](../v2/crates/wifi-densepose-wasm-edge/src/bld_hvac_presence.rs) | Occupancy-driven HVAC control with departure countdown | S (<5ms) |
+| Lighting Zones | [`bld_lighting_zones.rs`](../v2/crates/wifi-densepose-wasm-edge/src/bld_lighting_zones.rs) | Auto-dim/off lighting based on zone activity | S (<5ms) |
+| Elevator Count | [`bld_elevator_count.rs`](../v2/crates/wifi-densepose-wasm-edge/src/bld_elevator_count.rs) | Counts people entering/leaving with overload warning | S (<5ms) |
+| Meeting Room | [`bld_meeting_room.rs`](../v2/crates/wifi-densepose-wasm-edge/src/bld_meeting_room.rs) | Tracks meeting lifecycle: start, headcount, end, availability | S (<5ms) |
+| Energy Audit | [`bld_energy_audit.rs`](../v2/crates/wifi-densepose-wasm-edge/src/bld_energy_audit.rs) | Tracks after-hours usage and room utilization rates | S (<5ms) |
+
+**🛒 Retail & Hospitality** (Category 4) — Customer insights without cameras
+
+| Module | File | What It Does | Budget |
+|--------|------|-------------|--------|
+| Queue Length | [`ret_queue_length.rs`](../v2/crates/wifi-densepose-wasm-edge/src/ret_queue_length.rs) | Estimates queue size and wait times | S (<5ms) |
+| Dwell Heatmap | [`ret_dwell_heatmap.rs`](../v2/crates/wifi-densepose-wasm-edge/src/ret_dwell_heatmap.rs) | Shows where people spend time (hot/cold zones) | S (<5ms) |
+| Customer Flow | [`ret_customer_flow.rs`](../v2/crates/wifi-densepose-wasm-edge/src/ret_customer_flow.rs) | Counts ins/outs and tracks net occupancy | S (<5ms) |
+| Table Turnover | [`ret_table_turnover.rs`](../v2/crates/wifi-densepose-wasm-edge/src/ret_table_turnover.rs) | Restaurant table lifecycle: seated, dining, vacated | S (<5ms) |
+| Shelf Engagement | [`ret_shelf_engagement.rs`](../v2/crates/wifi-densepose-wasm-edge/src/ret_shelf_engagement.rs) | Detects browsing, considering, and reaching for products | S (<5ms) |
+
+**🏭 Industrial & Specialized** (Category 5) — Safety and compliance
+
+| Module | File | What It Does | Budget |
+|--------|------|-------------|--------|
+| Forklift Proximity | [`ind_forklift_proximity.rs`](../v2/crates/wifi-densepose-wasm-edge/src/ind_forklift_proximity.rs) | Warns when people get too close to vehicles | S (<5ms) |
+| Confined Space | [`ind_confined_space.rs`](../v2/crates/wifi-densepose-wasm-edge/src/ind_confined_space.rs) | OSHA-compliant worker monitoring with extraction alerts | S (<5ms) |
+| Clean Room | [`ind_clean_room.rs`](../v2/crates/wifi-densepose-wasm-edge/src/ind_clean_room.rs) | Occupancy limits and turbulent motion detection | S (<5ms) |
+| Livestock Monitor | [`ind_livestock_monitor.rs`](../v2/crates/wifi-densepose-wasm-edge/src/ind_livestock_monitor.rs) | Animal presence, stillness, and escape alerts | S (<5ms) |
+| Structural Vibration | [`ind_structural_vibration.rs`](../v2/crates/wifi-densepose-wasm-edge/src/ind_structural_vibration.rs) | Seismic events, mechanical resonance, structural drift | S (<5ms) |
+
+**🔮 Exotic & Research** (Category 6) — Experimental sensing applications
+
+| Module | File | What It Does | Budget |
+|--------|------|-------------|--------|
+| Dream Stage | [`exo_dream_stage.rs`](../v2/crates/wifi-densepose-wasm-edge/src/exo_dream_stage.rs) | Contactless sleep stage classification (wake/light/deep/REM) | S (<5ms) |
+| Emotion Detection | [`exo_emotion_detect.rs`](../v2/crates/wifi-densepose-wasm-edge/src/exo_emotion_detect.rs) | Arousal, stress, and calm detection from micro-movements | S (<5ms) |
+| Gesture Language | [`exo_gesture_language.rs`](../v2/crates/wifi-densepose-wasm-edge/src/exo_gesture_language.rs) | Sign language letter recognition via WiFi | S (<5ms) |
+| Music Conductor | [`exo_music_conductor.rs`](../v2/crates/wifi-densepose-wasm-edge/src/exo_music_conductor.rs) | Tempo and dynamic tracking from conducting gestures | S (<5ms) |
+| Plant Growth | [`exo_plant_growth.rs`](../v2/crates/wifi-densepose-wasm-edge/src/exo_plant_growth.rs) | Monitors plant growth, circadian rhythms, wilt detection | S (<5ms) |
+| Ghost Hunter | [`exo_ghost_hunter.rs`](../v2/crates/wifi-densepose-wasm-edge/src/exo_ghost_hunter.rs) | Environmental anomaly classification (draft/insect/wind/unknown) | S (<5ms) |
+| Rain Detection | [`exo_rain_detect.rs`](../v2/crates/wifi-densepose-wasm-edge/src/exo_rain_detect.rs) | Detects rain onset, intensity, and cessation via signal scatter | S (<5ms) |
+| Breathing Sync | [`exo_breathing_sync.rs`](../v2/crates/wifi-densepose-wasm-edge/src/exo_breathing_sync.rs) | Detects synchronized breathing between multiple people | S (<5ms) |
+
+</details>
+
+---
+
+<details>
+<summary><strong>🧠 Self-Learning WiFi AI (ADR-024)</strong> — Adaptive recognition, self-optimization, and intelligent anomaly detection</summary>
+
+Every WiFi signal that passes through a room creates a unique fingerprint of that space. WiFi-DensePose already reads these fingerprints to track people, but until now it threw away the internal "understanding" after each reading. The Self-Learning WiFi AI captures and preserves that understanding as compact, reusable vectors — and continuously optimizes itself for each new environment.
+
+**What it does in plain terms:**
+- Turns any WiFi signal into a 128-number "fingerprint" that uniquely describes what's happening in a room
+- Learns entirely on its own from raw WiFi data — no cameras, no labeling, no human supervision needed
+- Recognizes rooms, detects intruders, identifies people, and classifies activities using only WiFi
+- Runs on an $8 ESP32 chip (the entire model fits in 55 KB of memory)
+- Produces both body pose tracking AND environment fingerprints in a single computation
+
+**Key Capabilities**
+
+| What | How it works | Why it matters |
+|------|-------------|----------------|
+| **Self-supervised learning** | The model watches WiFi signals and teaches itself what "similar" and "different" look like, without any human-labeled data | Deploy anywhere — just plug in a WiFi sensor and wait 10 minutes |
+| **Room identification** | Each room produces a distinct WiFi fingerprint pattern | Know which room someone is in without GPS or beacons |
+| **Anomaly detection** | An unexpected person or event creates a fingerprint that doesn't match anything seen before | Automatic intrusion and fall detection as a free byproduct |
+| **Person re-identification** | Each person disturbs WiFi in a slightly different way, creating a personal signature | Track individuals across sessions without cameras |
+| **Environment adaptation** | MicroLoRA adapters (1,792 parameters per room) fine-tune the model for each new space | Adapts to a new room with minimal data — 93% less than retraining from scratch |
+| **Memory preservation** | EWC++ regularization remembers what was learned during pretraining | Switching to a new task doesn't erase prior knowledge |
+| **Hard-negative mining** | Training focuses on the most confusing examples to learn faster | Better accuracy with the same amount of training data |
+
+**Architecture**
+
+```
+WiFi Signal [56 channels] → Transformer + Graph Neural Network
+                                  ├→ 128-dim environment fingerprint (for search + identification)
+                                  └→ 17-joint body pose (for human tracking)
+```
+
+**Quick Start**
+
+```bash
+# Step 1: Learn from raw WiFi data (no labels needed)
+cargo run -p wifi-densepose-sensing-server -- --pretrain --dataset data/csi/ --pretrain-epochs 50
+
+# Step 2: Fine-tune with pose labels for full capability
+cargo run -p wifi-densepose-sensing-server -- --train --dataset data/mmfi/ --epochs 100 --save-rvf model.rvf
+
+# Step 3: Use the model — extract fingerprints from live WiFi
+cargo run -p wifi-densepose-sensing-server -- --model model.rvf --embed
+
+# Step 4: Search — find similar environments or detect anomalies
+cargo run -p wifi-densepose-sensing-server -- --model model.rvf --build-index env
+```
+
+**Training Modes**
+
+| Mode | What you need | What you get |
+|------|--------------|-------------|
+| Self-Supervised | Just raw WiFi data | A model that understands WiFi signal structure |
+| Supervised | WiFi data + body pose labels | Full pose tracking + environment fingerprints |
+| Cross-Modal | WiFi data + camera footage | Fingerprints aligned with visual understanding |
+
+**Fingerprint Index Types**
+
+| Index | What it stores | Real-world use |
+|-------|---------------|----------------|
+| `env_fingerprint` | Average room fingerprint | "Is this the kitchen or the bedroom?" |
+| `activity_pattern` | Activity boundaries | "Is someone cooking, sleeping, or exercising?" |
+| `temporal_baseline` | Normal conditions | "Something unusual just happened in this room" |
+| `person_track` | Individual movement signatures | "Person A just entered the living room" |
+
+**Model Size**
+
+| Component | Parameters | Memory (on ESP32) |
+|-----------|-----------|-------------------|
+| Transformer backbone | ~28,000 | 28 KB |
+| Embedding projection head | ~25,000 | 25 KB |
+| Per-room MicroLoRA adapter | ~1,800 | 2 KB |
+| **Total** | **~55,000** | **55 KB** (of 520 KB available) |
+
+The self-learning system builds on the AI Backbone (RuVector) signal-processing layer — attention, graph algorithms, and compression — adding contrastive learning on top. See [`architecture.md`](architecture.md) for the full pipeline.
+
+See [`docs/adr/ADR-024-contrastive-csi-embedding-model.md`](adr/ADR-024-contrastive-csi-embedding-model.md) for full architectural details.
+
+</details>
+

ui/components/SensingTab.js

@@ -105,6 +105,19 @@ export class SensingTab {
             </div>
           </div>
 
+          <!-- ADR-121: mmWave radar (HLK-LD2402) — auxiliary range modality -->
+          <div class="sensing-card" id="mmwaveCard" style="display:none;">
+            <div class="sensing-card-title">mmWave Radar (24 GHz)</div>
+            <div class="sensing-classification">
+              <div class="sensing-class-label" id="mmwaveLabel" style="background:rgba(33,150,243,0.15);color:rgb(33,150,243);">— cm</div>
+              <div class="sensing-confidence">
+                <label>Age</label>
+                <div class="sensing-bar"><div class="sensing-bar-fill confidence" id="mmwaveAgeBar" style="background:rgb(33,150,243);"></div></div>
+                <span class="sensing-meter-val" id="mmwaveAge">— ms</span>
+              </div>
+            </div>
+          </div>
+
           <!-- Setup info -->
           <div class="sensing-card">
             <div class="sensing-card-title">About This Data</div>
@@ -268,6 +281,35 @@ export class SensingTab {
     const confPct = ((c.confidence || 0) * 100).toFixed(0);
     this._setBar('barConfidence', c.confidence, 1.0, 'valConfidence', confPct + '%');
 
+    // ADR-121: poll mmWave radar in parallel with the WS-driven update.
+    // Kick once per visible update; skip if already in flight.
+    if (!this._mmwaveBusy) {
+      this._mmwaveBusy = true;
+      fetch('/api/v1/mmwave/latest')
+        .then(r => r.json())
+        .then(j => {
+          const card = this.container.querySelector('#mmwaveCard');
+          if (!card) { this._mmwaveBusy = false; return; }
+          if (j && j.available) {
+            card.style.display = '';
+            const lbl = this.container.querySelector('#mmwaveLabel');
+            if (lbl) lbl.textContent = j.distance_cm + ' cm';
+            const age = this.container.querySelector('#mmwaveAge');
+            if (age) age.textContent = (j.age_ms || 0) + ' ms';
+            const bar = this.container.querySelector('#mmwaveAgeBar');
+            if (bar) {
+              // Age 0..2000 ms → 100..0% width (fresher = fuller bar).
+              const pct = Math.max(0, 100 - (j.age_ms || 0) / 20);
+              bar.style.width = pct + '%';
+            }
+          } else {
+            card.style.display = 'none';
+          }
+        })
+        .catch(() => { /* server down or no port — silently hide */ })
+        .finally(() => { this._mmwaveBusy = false; });
+    }
+
     // Details
     this._setText('valDomFreq', (f.dominant_freq_hz || 0).toFixed(3) + ' Hz');
     this._setText('valChangePoints', String(f.change_points || 0));

ui/raw.html

@@ -0,0 +1,583 @@
+<!doctype html>
+<html lang="en"><head>
+<meta charset="utf-8"/>
+<meta name="viewport" content="width=device-width, initial-scale=1"/>
+<title>RuView — Raw Signals</title>
+<style>
+  :root { color-scheme: dark; }
+  body { margin:0; padding:14px; font-family:-apple-system,Inter,system-ui,sans-serif;
+         background:#0a0e13; color:#e6edf3; font-size:12px; }
+  h1 { font-size:15px; font-weight:600; margin:0 0 2px; }
+  .sub { font-size:11px; color:#888; margin:0 0 12px; }
+  .topbar { display:flex; gap:14px; align-items:center; margin-bottom:10px; flex-wrap:wrap; }
+  .pill { padding:4px 10px; border-radius:4px; font-family:JetBrains Mono,monospace; font-size:11px;
+          background:#1c2128; }
+  .pill.dis { background:#3a1418; color:#ff6a6a; }
+  .pill.ok  { background:#0e2a1a; color:#7ce38b; }
+  button { background:#21262d; color:#e6edf3; border:1px solid #30363d; border-radius:4px;
+           padding:4px 10px; font-size:11px; cursor:pointer; }
+  .node { background:#161b22; border:1px solid #30363d; border-radius:6px;
+          padding:10px 12px; margin-bottom:10px; }
+  .node h2 { margin:0 0 6px; font-size:12px; font-weight:600; color:#7cb6ff;
+             font-family:JetBrains Mono,monospace; display:flex; gap:14px; align-items:baseline; }
+  .node h2 .stat { color:#888; font-weight:normal; font-size:11px; }
+  .node h2 .stat b { color:#e6edf3; font-weight:600; }
+  .badge { font-family:JetBrains Mono,monospace; font-size:11px; padding:2px 8px; border-radius:3px; }
+  .badge.absent        { background:#21262d; color:#888; }
+  .badge.present_still { background:#1c3a55; color:#7cb6ff; }
+  .badge.present_moving{ background:#3a5520; color:#90d36b; }
+  .badge.active        { background:#552020; color:#ff7a7a; }
+  .row { display:grid; grid-template-columns: 1fr 360px; gap:10px; }
+  @media (max-width: 900px) { .row { grid-template-columns: 1fr; } }
+  canvas { display:block; width:100%; background:#0a0e13; border-radius:3px; }
+  canvas.bars   { height: 130px; }
+  canvas.trace  { height: 130px; }
+  canvas.spark  { height: 48px; margin-top: 6px; }
+  .lbl { color:#666; font-size:10px; font-family:JetBrains Mono,monospace; margin:2px 0 0; }
+  .controls { display:flex; gap:8px; margin-left:auto; }
+  .controls label { font-size:11px; color:#aaa; }
+</style>
+</head>
+<body>
+<h1>RuView — Raw CSI signals</h1>
+<p class="sub">Per-node subcarrier amplitudes + RSSI/broadband traces. No DSP, no classification. Stream straight from the sensor.</p>
+
+<div class="topbar">
+  <span id="status" class="pill dis">disconnected</span>
+  <span class="pill" id="rate">0 fps</span>
+  <span class="pill" id="lastTs">last: --</span>
+  <span class="badge absent" id="globalBadge" style="font-size:13px;padding:4px 12px;">absent</span>
+  <span class="pill" id="globalCV">CV 0%</span>
+  <!-- ADR-021: WiFi-CSI vital signs — breathing + heart rate (computed server-side). -->
+  <span class="pill" id="brPill"
+        style="background:rgba(63,185,80,0.18); color:rgb(63,185,80); border:1px solid rgb(63,185,80);"
+        title="WiFi-CSI breathing rate from bandpass 0.1-0.5 Hz on broadband amplitude (ADR-021)">
+    🫁 <b id="brBpm">— BPM</b> <span id="brConf" style="opacity:0.7;font-size:11px">·</span>
+  </span>
+  <span class="pill" id="hrPill"
+        style="background:rgba(248,81,73,0.18); color:rgb(248,81,73); border:1px solid rgb(248,81,73);"
+        title="WiFi-CSI heart rate from bandpass 0.8-2.0 Hz on broadband amplitude (ADR-021)">
+    💓 <b id="hrBpm">— BPM</b> <span id="hrConf" style="opacity:0.7;font-size:11px">·</span>
+  </span>
+  <!-- ADR-121: HLK-LD2402 24 GHz mmWave radar pill — hidden until first reading. -->
+  <span class="pill" id="mmwavePill" style="display:none; background:rgba(33,150,243,0.18);
+                                              color:rgb(33,150,243); border:1px solid rgb(33,150,243);"
+        title="HLK-LD2402 24 GHz radar — distance to closest target">
+    📡 mmWave <b id="mmwaveDist">— cm</b> <span id="mmwaveAge" style="opacity:0.7;font-size:11px">·</span>
+  </span>
+  <div class="controls">
+    <label>peak-hold <input type="checkbox" id="peakHold" checked></label>
+    <label>log-y <input type="checkbox" id="logY"></label>
+    <button onclick="resetState()">reset</button>
+    <button id="calibrateBtn" onclick="startCalibrate()" title="Step out of the room, click, wait 90 s">calibrate empty</button>
+    <span class="pill" id="calibStatus" style="display:none"></span>
+    <!-- ADR-107: visible progress bar shown while baseline capture runs. -->
+    <div id="calibProgress" style="display:none; position:relative; width:140px; height:14px;
+                                   border:1px solid #30363d; border-radius:7px; overflow:hidden;
+                                   background:#0a0e13;">
+      <div id="calibProgressFill" style="position:absolute; left:0; top:0; bottom:0; width:0%;
+                                          background:linear-gradient(90deg,#1f6feb,#3fb950);
+                                          transition: width 0.4s linear;"></div>
+      <span id="calibProgressLabel" style="position:absolute; inset:0; display:flex;
+                                            align-items:center; justify-content:center;
+                                            font-size:10px; font-family:JetBrains Mono,monospace;
+                                            color:#e6edf3; text-shadow:0 0 2px #000;"></span>
+    </div>
+  </div>
+</div>
+
+<div id="nodes"></div>
+
+<script>
+// ── State ──────────────────────────────────────────────────────────
+const TRACE_SEC = 30;             // seconds of history per node
+const TRACE_MAX_PTS = 1200;       // safety cap
+const state = new Map();          // node_id -> { amp, peak, rssiHist[], meanAmpHist[], lastTs, frames }
+let frameCount = 0;
+let lastRateTs = performance.now();
+let rateFps = 0;
+let logY = false;
+let peakHold = true;
+
+function resetState() {
+  state.clear();
+  document.getElementById('nodes').innerHTML = '';
+  frameCount = 0;
+}
+
+document.getElementById('peakHold').addEventListener('change', e => { peakHold = e.target.checked; });
+document.getElementById('logY').addEventListener('change', e => { logY = e.target.checked; });
+
+// ── Per-node block factory ─────────────────────────────────────────
+function ensureNodeBlock(nodeId) {
+  if (state.has(nodeId)) return state.get(nodeId);
+  const ent = {
+    amp: [],
+    peak: [],
+    rssiHist: [],          // { t, v }
+    meanAmpHist: [],
+    driftHist: [],         // { t, v } — ADR-104 per-sub drift score
+    lastTs: 0,
+    frames: 0,
+    lastFrameWall: performance.now(),
+    fps: 0,
+  };
+  state.set(nodeId, ent);
+
+  const wrap = document.createElement('div');
+  wrap.className = 'node';
+  wrap.id = 'node-' + nodeId;
+  wrap.innerHTML = `
+    <h2>
+      Node ${nodeId}
+      <span class="badge absent" id="n${nodeId}-badge">absent</span>
+      <span class="stat">CV <b id="n${nodeId}-cv">0%</b></span>
+      <span class="stat">subc <b id="n${nodeId}-sub">0</b></span>
+      <span class="stat">rssi <b id="n${nodeId}-rssi">--</b> dBm</span>
+      <span class="stat">mean A <b id="n${nodeId}-meanA">0</b></span>
+      <span class="stat">peak A <b id="n${nodeId}-peakA">0</b></span>
+      <span class="stat">drift <b id="n${nodeId}-drift">--</b></span>
+      <span class="stat">node fps <b id="n${nodeId}-fps">0</b></span>
+    </h2>
+    <div class="row">
+      <div>
+        <canvas class="bars" id="n${nodeId}-bars"></canvas>
+        <p class="lbl">subcarrier amplitude bars (left → low freq, right → high freq)</p>
+      </div>
+      <div>
+        <canvas class="trace" id="n${nodeId}-trace"></canvas>
+        <p class="lbl"><span style="color:#8b949e">RSSI</span> &nbsp; <span style="color:#3fb950">broadband mean amplitude</span> &nbsp; (last ${TRACE_SEC}s)</p>
+        <canvas class="spark" id="n${nodeId}-driftSpark"></canvas>
+        <p class="lbl"><span style="color:#d29922">per-sub drift</span> — off-axis presence channel (ADR-104); dashed line = presence threshold 0.10</p>
+      </div>
+    </div>`;
+  document.getElementById('nodes').appendChild(wrap);
+  return ent;
+}
+
+// ── Drawing ────────────────────────────────────────────────────────
+function drawBars(canvas, amps, peaks) {
+  const w = canvas.clientWidth, h = canvas.clientHeight;
+  if (canvas.width !== w || canvas.height !== h) { canvas.width = w; canvas.height = h; }
+  const ctx = canvas.getContext('2d');
+  ctx.fillStyle = '#0a0e13'; ctx.fillRect(0, 0, w, h);
+  if (!amps.length) return;
+
+  // Determine scale
+  let maxV = peakHold && peaks.length
+    ? Math.max(...peaks)
+    : Math.max(...amps);
+  if (!isFinite(maxV) || maxV <= 0) maxV = 1;
+
+  const n = amps.length;
+  const bw = w / n;
+  const margin = 4;
+
+  // Bars
+  for (let i = 0; i < n; i++) {
+    let v = amps[i];
+    let pv = peaks[i] || 0;
+    if (logY) {
+      v = v > 0 ? Math.log10(v + 1) : 0;
+      pv = pv > 0 ? Math.log10(pv + 1) : 0;
+    }
+    const scaleMax = logY ? Math.log10(maxV + 1) : maxV;
+    const bh = Math.max(1, (v / scaleMax) * (h - margin));
+    const ph = Math.max(1, (pv / scaleMax) * (h - margin));
+    const x = i * bw;
+    // peak (faint)
+    if (peakHold && pv > 0) {
+      ctx.fillStyle = '#1f3a5a';
+      ctx.fillRect(x, h - ph, Math.max(1, bw - 1), 1.5);
+    }
+    // bar (active)
+    const hue = 200 + (i / n) * 100;
+    ctx.fillStyle = `hsl(${hue}, 70%, 55%)`;
+    ctx.fillRect(x, h - bh, Math.max(1, bw - 1), bh);
+  }
+
+  // Y-axis label
+  ctx.fillStyle = '#555'; ctx.font = '9px monospace';
+  ctx.fillText('max=' + maxV.toFixed(0), 4, 10);
+  ctx.fillText('n=' + n, w - 40, 10);
+}
+
+function drawTrace(canvas, rssiHist, meanAmpHist) {
+  const w = canvas.clientWidth, h = canvas.clientHeight;
+  if (canvas.width !== w || canvas.height !== h) { canvas.width = w; canvas.height = h; }
+  const ctx = canvas.getContext('2d');
+  ctx.fillStyle = '#0a0e13'; ctx.fillRect(0, 0, w, h);
+
+  const now = performance.now() / 1000;
+  const t0 = now - TRACE_SEC;
+
+  const drawSeries = (arr, color, getRange) => {
+    if (arr.length < 2) return;
+    const visible = arr.filter(p => p.t >= t0);
+    if (visible.length < 2) return;
+    const { min, max } = getRange(visible);
+    const span = (max - min) || 1;
+    ctx.strokeStyle = color; ctx.lineWidth = 1.5; ctx.beginPath();
+    for (let i = 0; i < visible.length; i++) {
+      const p = visible[i];
+      const x = ((p.t - t0) / TRACE_SEC) * w;
+      const y = h - ((p.v - min) / span) * (h - 8) - 4;
+      if (i === 0) ctx.moveTo(x, y); else ctx.lineTo(x, y);
+    }
+    ctx.stroke();
+    // y-range text
+    ctx.fillStyle = color; ctx.font = '9px monospace';
+    return { min, max };
+  };
+
+  const rssiR = drawSeries(rssiHist, '#8b949e', arr => {
+    const vals = arr.map(p => p.v);
+    return { min: Math.min(...vals), max: Math.max(...vals) };
+  });
+  const ampR = drawSeries(meanAmpHist, '#3fb950', arr => {
+    const vals = arr.map(p => p.v);
+    return { min: 0, max: Math.max(...vals) };
+  });
+
+  // labels
+  ctx.font = '9px monospace';
+  if (rssiR) { ctx.fillStyle = '#8b949e'; ctx.fillText(`rssi ${rssiR.min.toFixed(0)}…${rssiR.max.toFixed(0)} dBm`, 4, 10); }
+  if (ampR)  { ctx.fillStyle = '#3fb950'; ctx.fillText(`A ${ampR.min.toFixed(0)}…${ampR.max.toFixed(0)}`, 4, 22); }
+
+  // grid line at now
+  ctx.strokeStyle = '#1c2128'; ctx.beginPath();
+  ctx.moveTo(w - 1, 0); ctx.lineTo(w - 1, h); ctx.stroke();
+}
+
+// ADR-104: per-sub drift sparkline. Fixed Y range [0, 0.30] so the
+// presence threshold (0.10, dashed) and warning threshold (0.15) are
+// directly readable across nodes — re-scaling per node would make it
+// impossible to tell "Node 0 fired" from "Node 1 fired" at a glance.
+const DRIFT_PRESENCE_THRESH = 0.10;
+const DRIFT_WARN_THRESH = 0.15;
+const DRIFT_MAX = 0.30;
+
+function drawDriftSpark(canvas, hist) {
+  const w = canvas.clientWidth, h = canvas.clientHeight;
+  if (canvas.width !== w || canvas.height !== h) { canvas.width = w; canvas.height = h; }
+  const ctx = canvas.getContext('2d');
+  ctx.fillStyle = '#0a0e13'; ctx.fillRect(0, 0, w, h);
+
+  const now = performance.now() / 1000;
+  const t0 = now - TRACE_SEC;
+  const yOf = v => h - (Math.min(v, DRIFT_MAX) / DRIFT_MAX) * (h - 4) - 2;
+
+  // Threshold lines.
+  ctx.setLineDash([3, 3]);
+  ctx.strokeStyle = '#5a4a1a'; ctx.lineWidth = 1; ctx.beginPath();
+  ctx.moveTo(0, yOf(DRIFT_PRESENCE_THRESH)); ctx.lineTo(w, yOf(DRIFT_PRESENCE_THRESH));
+  ctx.stroke();
+  ctx.strokeStyle = '#7a3030'; ctx.beginPath();
+  ctx.moveTo(0, yOf(DRIFT_WARN_THRESH)); ctx.lineTo(w, yOf(DRIFT_WARN_THRESH));
+  ctx.stroke();
+  ctx.setLineDash([]);
+
+  const visible = hist.filter(p => p.t >= t0);
+  if (visible.length >= 2) {
+    ctx.strokeStyle = '#d29922'; ctx.lineWidth = 1.5; ctx.beginPath();
+    for (let i = 0; i < visible.length; i++) {
+      const p = visible[i];
+      const x = ((p.t - t0) / TRACE_SEC) * w;
+      const y = yOf(p.v);
+      if (i === 0) ctx.moveTo(x, y); else ctx.lineTo(x, y);
+    }
+    ctx.stroke();
+  }
+
+  // Axis text.
+  ctx.fillStyle = '#666'; ctx.font = '9px monospace';
+  ctx.fillText('0', 2, h - 2);
+  ctx.fillText(DRIFT_MAX.toFixed(2), 2, 10);
+}
+
+// ── Frame ingestion ────────────────────────────────────────────────
+function handleSensingUpdate(d) {
+  const nodes = d.nodes || [];
+  const ts = d.timestamp || (Date.now() / 1000);
+  const now = performance.now() / 1000;
+  for (const n of nodes) {
+    const id = n.node_id;
+    const amps = n.amplitude || [];
+    // Skip empty-amp ticks (feature_state path doesn't carry raw CSI).
+    // Bars/traces only refresh on real raw-CSI frames so what you see
+    // is always a live snapshot, not a repeated stale vector.
+    if (!amps.length) continue;
+    const ent = ensureNodeBlock(id);
+    ent.amp = amps;
+    // peak-hold update
+    if (ent.peak.length !== amps.length) ent.peak = amps.slice();
+    else for (let i = 0; i < amps.length; i++) if (amps[i] > ent.peak[i]) ent.peak[i] = amps[i];
+
+    const meanA = amps.reduce((s, x) => s + x, 0) / amps.length;
+    // Only push valid (non-zero) RSSI samples so the trace doesn't
+    // jump between real dBm values and the "0 = no data" sentinel.
+    if (n.rssi_dbm && n.rssi_dbm !== 0) {
+      ent.rssiHist.push({ t: now, v: n.rssi_dbm });
+    }
+    ent.meanAmpHist.push({ t: now, v: meanA });
+    const cutoff = now - TRACE_SEC;
+    while (ent.rssiHist.length && ent.rssiHist[0].t < cutoff) ent.rssiHist.shift();
+    while (ent.meanAmpHist.length && ent.meanAmpHist[0].t < cutoff) ent.meanAmpHist.shift();
+    if (ent.rssiHist.length > TRACE_MAX_PTS) ent.rssiHist.splice(0, ent.rssiHist.length - TRACE_MAX_PTS);
+    if (ent.meanAmpHist.length > TRACE_MAX_PTS) ent.meanAmpHist.splice(0, ent.meanAmpHist.length - TRACE_MAX_PTS);
+
+    // per-node fps: count frames in the last second, refresh once a sec
+    // (instantaneous 1/dt was wildly noisy because multiple WS paths
+    // emit duplicate per-node updates back-to-back).
+    ent.fpsCounter = (ent.fpsCounter || 0) + 1;
+    const nowMs = performance.now();
+    if (!ent.fpsWindowStart) ent.fpsWindowStart = nowMs;
+    if (nowMs - ent.fpsWindowStart >= 1000) {
+      ent.fps = ent.fpsCounter * 1000 / (nowMs - ent.fpsWindowStart);
+      ent.fpsCounter = 0;
+      ent.fpsWindowStart = nowMs;
+    }
+    ent.lastFrameWall = nowMs;
+    ent.frames++;
+    ent.lastTs = ts;
+
+    document.getElementById(`n${id}-sub`).textContent = amps.length;
+    // n.rssi_dbm comes from sensing_update.nodes[]; it can be 0 on
+    // early ticks (history not yet populated). Coerce to "--" so the
+    // operator doesn't think the AP is dead.
+    const rssiVal = (n.rssi_dbm && Number.isFinite(n.rssi_dbm) && n.rssi_dbm !== 0)
+      ? n.rssi_dbm.toFixed(1)
+      : '--';
+    document.getElementById(`n${id}-rssi`).textContent = rssiVal;
+    // Push to RSSI trace history if non-zero (so the chart shows the
+    // real ladder of dBm steps, not a fake "0 → -54" jump on boot).
+    if (n.rssi_dbm && n.rssi_dbm !== 0) {
+      // (handled by ent.rssiHist push below)
+    }
+    document.getElementById(`n${id}-meanA`).textContent = meanA.toFixed(1);
+    document.getElementById(`n${id}-peakA`).textContent = Math.max(...ent.peak).toFixed(1);
+    document.getElementById(`n${id}-fps`).textContent = ent.fps.toFixed(1);
+  }
+
+  document.getElementById('lastTs').textContent = 'last: ' + new Date(ts * 1000).toLocaleTimeString();
+
+  // Global classification badge (ADR-101 fused).
+  const gcl = d.classification || {};
+  const glvl = gcl.motion_level || 'absent';
+  const gb = document.getElementById('globalBadge');
+  if (gb) { gb.textContent = glvl; gb.className = 'badge ' + glvl; gb.style.fontSize = '13px'; gb.style.padding = '4px 12px'; }
+  const gcv = document.getElementById('globalCV');
+  if (gcv) gcv.textContent = 'CV ' + ((gcl.confidence || 0) * 100).toFixed(1) + '%';
+
+  // ADR-021 — WiFi-CSI vital signs (breathing + heart rate).
+  // `vital_signs` is embedded in SensingUpdate; values may be null
+  // when the detector hasn't accumulated enough history yet (~10s).
+  const vs = d.vital_signs || {};
+  const brBpm  = document.getElementById('brBpm');
+  const brConf = document.getElementById('brConf');
+  const hrBpm  = document.getElementById('hrBpm');
+  const hrConf = document.getElementById('hrConf');
+  const brPill = document.getElementById('brPill');
+  const hrPill = document.getElementById('hrPill');
+  if (vs && typeof vs.breathing_rate_bpm === 'number' && Number.isFinite(vs.breathing_rate_bpm) && vs.breathing_rate_bpm > 0) {
+    if (brBpm)  brBpm.textContent  = vs.breathing_rate_bpm.toFixed(1) + ' BPM';
+    if (brConf) brConf.textContent = '· ' + ((vs.breathing_confidence || 0) * 100).toFixed(0) + '%';
+    if (brPill) brPill.style.opacity = (vs.breathing_confidence || 0) < 0.2 ? '0.5' : '1.0';
+  } else {
+    if (brBpm)  brBpm.textContent  = '— BPM';
+    if (brConf) brConf.textContent = '·';
+    if (brPill) brPill.style.opacity = '0.5';
+  }
+  if (vs && typeof vs.heart_rate_bpm === 'number' && Number.isFinite(vs.heart_rate_bpm) && vs.heart_rate_bpm > 0) {
+    if (hrBpm)  hrBpm.textContent  = vs.heart_rate_bpm.toFixed(0) + ' BPM';
+    if (hrConf) hrConf.textContent = '· ' + ((vs.heartbeat_confidence || 0) * 100).toFixed(0) + '%';
+    if (hrPill) hrPill.style.opacity = (vs.heartbeat_confidence || 0) < 0.2 ? '0.5' : '1.0';
+  } else {
+    if (hrBpm)  hrBpm.textContent  = '— BPM';
+    if (hrConf) hrConf.textContent = '·';
+    if (hrPill) hrPill.style.opacity = '0.5';
+  }
+
+  // Per-node level badge from node_features[i].classification (ADR-101).
+  const nfNow = performance.now() / 1000;
+  const nf = d.node_features || [];
+  for (const f of nf) {
+    const id = f.node_id;
+    const cls = f.classification || {};
+    const lvl = cls.motion_level || 'absent';
+    const badge = document.getElementById(`n${id}-badge`);
+    if (badge) {
+      badge.textContent = lvl;
+      badge.className = 'badge ' + lvl;
+    }
+    const cvEl = document.getElementById(`n${id}-cv`);
+    if (cvEl) cvEl.textContent = ((cls.confidence || 0) * 100).toFixed(1) + '%';
+
+    // ADR-104 per-sub drift score (off-axis presence). May be absent
+    // when no per-sub baseline is loaded for this node — show '--'
+    // instead of '0.000' so the operator can tell the channel is
+    // unknown vs. known and stable.
+    const driftEl = document.getElementById(`n${id}-drift`);
+    const driftLive = state.get(id);
+    if (typeof f.drift_score === 'number' && Number.isFinite(f.drift_score)) {
+      if (driftEl) driftEl.textContent = f.drift_score.toFixed(3);
+      if (driftLive) {
+        driftLive.driftHist.push({ t: nfNow, v: f.drift_score });
+        const cutoff = nfNow - TRACE_SEC;
+        while (driftLive.driftHist.length && driftLive.driftHist[0].t < cutoff) {
+          driftLive.driftHist.shift();
+        }
+        if (driftLive.driftHist.length > TRACE_MAX_PTS) {
+          driftLive.driftHist.splice(0, driftLive.driftHist.length - TRACE_MAX_PTS);
+        }
+      }
+    } else if (driftEl) {
+      driftEl.textContent = '--';
+    }
+  }
+  frameCount++;
+}
+
+function renderTick() {
+  for (const [id, ent] of state) {
+    const bars  = document.getElementById('n' + id + '-bars');
+    const trace = document.getElementById('n' + id + '-trace');
+    const spark = document.getElementById('n' + id + '-driftSpark');
+    if (bars)  drawBars(bars,  ent.amp, ent.peak);
+    if (trace) drawTrace(trace, ent.rssiHist, ent.meanAmpHist);
+    if (spark) drawDriftSpark(spark, ent.driftHist);
+  }
+  // fps pill
+  const now = performance.now();
+  if (now - lastRateTs > 500) {
+    rateFps = (frameCount * 1000) / (now - lastRateTs);
+    document.getElementById('rate').textContent = rateFps.toFixed(1) + ' fps total';
+    frameCount = 0;
+    lastRateTs = now;
+  }
+  requestAnimationFrame(renderTick);
+}
+requestAnimationFrame(renderTick);
+
+// ── ADR-107: baseline calibrate button + progress bar ─────────────
+let calibPollTimer = null;
+const CALIB_DURATION_SEC = 90;
+
+function setCalibProgress(pct, label) {
+  const bar = document.getElementById('calibProgress');
+  const fill = document.getElementById('calibProgressFill');
+  const txt = document.getElementById('calibProgressLabel');
+  if (!bar || !fill || !txt) return;
+  bar.style.display = pct < 0 ? 'none' : 'inline-block';
+  fill.style.width = Math.max(0, Math.min(100, pct)) + '%';
+  txt.textContent = label || '';
+}
+
+async function startCalibrate() {
+  if (!confirm(`Step OUT of the room now. Calibration will record for ${CALIB_DURATION_SEC} s.\nClick OK when you are out.`)) return;
+  const btn = document.getElementById('calibrateBtn');
+  const stat = document.getElementById('calibStatus');
+  btn.disabled = true; btn.textContent = 'recording…';
+  // Hide the text-pill while the progress bar is the primary indicator;
+  // it reappears only on terminal status messages (error / complete).
+  stat.style.display = 'none';
+  setCalibProgress(0, 'starting…');
+  try {
+    const res = await fetch('/api/v1/baseline/calibrate', {
+      method: 'POST',
+      headers: {'Content-Type': 'application/json'},
+      body: JSON.stringify({ duration_sec: CALIB_DURATION_SEC, trim_sec: 15, clean_window_sec: 30 }),
+    });
+    const j = await res.json();
+    if (!j.started) {
+      setCalibProgress(-1, '');
+      stat.style.display = 'inline-block';
+      stat.textContent = j.reason || 'failed to start';
+      btn.disabled = false; btn.textContent = 'calibrate empty';
+      return;
+    }
+  } catch (e) {
+    setCalibProgress(-1, '');
+    stat.style.display = 'inline-block';
+    stat.textContent = 'network error';
+    btn.disabled = false; btn.textContent = 'calibrate empty';
+    return;
+  }
+  if (calibPollTimer) clearInterval(calibPollTimer);
+  let elapsed = 0;
+  calibPollTimer = setInterval(async () => {
+    elapsed += 2;
+    try {
+      const r = await fetch('/api/v1/baseline'); const j = await r.json();
+      const s = j.calibration_status || 'idle';
+      if (s.startsWith('running')) {
+        const pct = Math.min(99, (elapsed / CALIB_DURATION_SEC) * 100);
+        setCalibProgress(pct, `${elapsed}/${CALIB_DURATION_SEC} s`);
+      } else {
+        clearInterval(calibPollTimer); calibPollTimer = null;
+        btn.disabled = false; btn.textContent = 'calibrate empty';
+        if (s === 'complete') {
+          setCalibProgress(100, 'done');
+          stat.style.display = 'inline-block';
+          stat.textContent = 'baseline updated ✓';
+          setTimeout(() => setCalibProgress(-1, ''), 3000);
+        } else {
+          setCalibProgress(-1, '');
+          stat.style.display = 'inline-block';
+          stat.textContent = s;
+        }
+      }
+    } catch (e) {}
+  }, 2000);
+}
+
+// ── WS ─────────────────────────────────────────────────────────────
+function connect() {
+  const ws = new WebSocket('ws://' + location.hostname + ':8765/ws/sensing');
+  ws.onopen = () => {
+    const p = document.getElementById('status');
+    p.textContent = 'connected'; p.className = 'pill ok';
+  };
+  ws.onclose = () => {
+    const p = document.getElementById('status');
+    p.textContent = 'disconnected — reconnecting'; p.className = 'pill dis';
+    setTimeout(connect, 1500);
+  };
+  ws.onmessage = (e) => {
+    try {
+      const d = JSON.parse(e.data);
+      if (d.type === 'sensing_update') handleSensingUpdate(d);
+    } catch (_) {}
+  };
+}
+connect();
+
+// ── ADR-121: poll HLK-LD2402 mmWave radar @ 5 Hz ─────────────────────
+const mmwavePill = document.getElementById('mmwavePill');
+const mmwaveDist = document.getElementById('mmwaveDist');
+const mmwaveAge  = document.getElementById('mmwaveAge');
+let mmwaveBusy = false;
+async function pollMmwave() {
+  if (mmwaveBusy) return; mmwaveBusy = true;
+  try {
+    const r = await fetch('/api/v1/mmwave/latest', { cache: 'no-store' });
+    if (!r.ok) throw new Error('http ' + r.status);
+    const j = await r.json();
+    if (j && j.available) {
+      mmwavePill.style.display = '';
+      mmwaveDist.textContent = j.distance_cm + ' cm';
+      const age = Math.round(j.age_ms || 0);
+      mmwaveAge.textContent = '· ' + age + ' ms';
+      // Fade pill if stale (>1.5 s) before server hides at 2 s.
+      mmwavePill.style.opacity = age > 1500 ? '0.5' : '1.0';
+    } else {
+      mmwavePill.style.display = 'none';
+    }
+  } catch (_) {
+    mmwavePill.style.display = 'none';
+  } finally { mmwaveBusy = false; }
+}
+pollMmwave();
+setInterval(pollMmwave, 200);
+</script>
+</body></html>

ui/style.css

@@ -2033,6 +2033,22 @@ canvas {
   color: var(--color-error);
 }
 
+/* ADR-117 follow-up: extended class vocabulary from adaptive_classifier. */
+.sensing-class-label.present_moving {
+  background: rgba(var(--color-warning-rgb), 0.15);
+  color: var(--color-warning);
+}
+
+.sensing-class-label.waving {
+  background: rgba(155, 89, 182, 0.15);   /* purple — gestures, body still */
+  color: rgb(155, 89, 182);
+}
+
+.sensing-class-label.transition {
+  background: rgba(230, 126, 34, 0.18);   /* orange — discrete event */
+  color: rgb(230, 126, 34);
+}
+
 .sensing-confidence {
   display: grid;
   grid-template-columns: 70px 1fr 40px;

v2/Cargo.lock

@@ -8550,6 +8550,7 @@ dependencies = [
  "ruvector-mincut",
  "serde",
  "serde_json",
+ "serialport",
  "tempfile",
  "tokio",
  "tower 0.4.13",

v2/crates/wifi-densepose-sensing-server/Cargo.toml

@@ -56,6 +56,10 @@ wifi-densepose-signal = { version = "0.3.0", path = "../wifi-densepose-signal",
 midstreamer-temporal-compare = "0.2"   # DTW / LCS / Edit-Distance pattern matching
 midstreamer-attractor        = "0.2"   # Lyapunov + regime classification
 
+# ADR-121: HLK-LD2402 24 GHz mmWave radar over UART (auxiliary vitals/
+# range modality). Optional — server runs fine without the module attached.
+serialport.workspace = true
+
 [dev-dependencies]
 tempfile = "3.10"
 # `tower::ServiceExt::oneshot` for in-process Router tests (bearer_auth).

v2/crates/wifi-densepose-sensing-server/src/lib.rs

@@ -21,3 +21,5 @@ pub mod sparse_inference;
 pub mod embedding;
 /// ADR-116: WiFlow-v1 supervised pose model loader + Rust forward pass.
 pub mod wiflow_v1;
+/// ADR-121: HLK-LD2402 24 GHz mmWave radar reader (auxiliary modality).
+pub mod mmwave;

v2/crates/wifi-densepose-sensing-server/src/main.rs

@@ -1174,6 +1174,20 @@ struct Args {
     /// Independent from `--model` (RVF container) and `--load-rvf`.
     #[arg(long, value_name = "PATH")]
     wiflow_model: Option<PathBuf>,
+
+    /// ADR-121: Path to HLK-LD2402 24 GHz mmWave radar UART (via
+    /// CP2102 USB bridge). Example: `/dev/cu.usbserial-1140` (macOS)
+    /// or `/dev/ttyUSB0` (Linux). When set, the server reads
+    /// `distance:<cm>` lines at 6 Hz and surfaces them on the
+    /// `mmwave` field of every SensingUpdate plus the
+    /// `/api/v1/mmwave/latest` REST endpoint. Missing port = no
+    /// mmWave, server still runs.
+    #[arg(long, value_name = "PATH")]
+    mmwave_port: Option<String>,
+
+    /// ADR-121: HLK-LD2402 baud rate. Factory default is 115200 8N1.
+    #[arg(long, default_value = "115200")]
+    mmwave_baud: u32,
 }
 
 /// ADR-116: globally-shared WiFlow-v1 model. Loaded once at startup if
@@ -5069,6 +5083,23 @@ async fn adaptive_debug() -> Json<serde_json::Value> {
     }))
 }
 
+/// ADR-121: GET /api/v1/mmwave/latest — latest HLK-LD2402 reading or
+/// `{ available: false }` when the reader thread isn't running OR the
+/// most recent reading is stale (>2 seconds old).
+async fn mmwave_latest() -> Json<serde_json::Value> {
+    use wifi_densepose_sensing_server::mmwave;
+    match mmwave::current(std::time::Duration::from_secs(2)) {
+        Some(r) => Json(serde_json::json!({
+            "available": true,
+            "distance_cm": r.distance_cm,
+            "age_ms": r.at.elapsed().as_millis() as u64,
+        })),
+        None => Json(serde_json::json!({
+            "available": false,
+        })),
+    }
+}
+
 /// POST /api/v1/adaptive/unload — unload the adaptive model (revert to thresholds).
 async fn adaptive_unload(State(state): State<SharedState>) -> Json<serde_json::Value> {
     let mut s = state.write().await;
@@ -7364,6 +7395,11 @@ async fn main() {
         }
     };
 
+    // ADR-121: spawn the HLK-LD2402 mmWave reader thread when --mmwave-port is set.
+    if let Some(port) = args.mmwave_port.clone() {
+        wifi_densepose_sensing_server::mmwave::spawn_reader(port, args.mmwave_baud);
+    }
+
     // Load trained model via --model (uses progressive loading if --progressive set)
     let model_path = args.model.as_ref().or(args.load_rvf.as_ref());
     let mut progressive_loader: Option<ProgressiveLoader> = None;
@@ -7648,6 +7684,7 @@ async fn main() {
         .route("/api/v1/adaptive/train", post(adaptive_train))
         .route("/api/v1/adaptive/status", get(adaptive_status))
         .route("/api/v1/adaptive/debug", get(adaptive_debug))
+        .route("/api/v1/mmwave/latest", get(mmwave_latest))
         .route("/api/v1/adaptive/unload", post(adaptive_unload))
         // Field model calibration (eigenvalue-based person counting)
         .route("/api/v1/calibration/start", post(calibration_start))

v2/crates/wifi-densepose-sensing-server/src/mmwave.rs

@@ -0,0 +1,116 @@
+//! ADR-121: HLK-LD2402 24 GHz mmWave radar reader.
+//!
+//! Auxiliary range/vitals modality, attached over a CP2102 USB-UART
+//! bridge. The module ships factory firmware that emits ASCII
+//! `distance:<cm>\r\n` lines @ 115200 baud, ~6 Hz, in Normal Mode.
+//!
+//! This reader runs in a dedicated thread (blocking serial I/O is
+//! awkward inside tokio) and pushes the latest reading + monotonic
+//! timestamp into a global `OnceLock<Mutex<…>>` that the broadcast
+//! tick task reads.
+//!
+//! Cold-start tolerance: if the port cannot be opened, the thread
+//! logs once and exits cleanly — the server keeps running with WiFi
+//! sensing only. No panics, no retries (operator can hot-plug; if
+//! they want auto-reconnect we can add it later).
+
+use std::io::Read;
+use std::sync::{Mutex, OnceLock};
+use std::time::{Duration, Instant};
+
+/// Latest mmWave reading + when it landed.
+#[derive(Debug, Clone, Copy)]
+pub struct MmwaveReading {
+    pub distance_cm: u32,
+    pub at: Instant,
+}
+
+static LATEST: OnceLock<Mutex<Option<MmwaveReading>>> = OnceLock::new();
+
+fn latest() -> &'static Mutex<Option<MmwaveReading>> {
+    LATEST.get_or_init(|| Mutex::new(None))
+}
+
+/// Returns the most recent reading if it landed within `staleness`.
+pub fn current(staleness: Duration) -> Option<MmwaveReading> {
+    let g = latest().lock().unwrap();
+    let r = (*g)?;
+    if r.at.elapsed() <= staleness { Some(r) } else { None }
+}
+
+/// Spawn the blocking serial reader thread. Returns immediately.
+/// `port` example: `/dev/cu.usbserial-1140` (macOS) or `/dev/ttyUSB0`
+/// (Linux). `baud` should be 115200 for HLK-LD2402 default firmware.
+pub fn spawn_reader(port: String, baud: u32) {
+    std::thread::Builder::new()
+        .name("mmwave-reader".into())
+        .spawn(move || run(port, baud))
+        .expect("failed to spawn mmwave-reader thread");
+}
+
+fn run(port: String, baud: u32) {
+    let mut serial = match serialport::new(&port, baud)
+        .timeout(Duration::from_millis(500))
+        .open()
+    {
+        Ok(s) => {
+            tracing::info!("ADR-121 mmWave reader: opened {port} @ {baud}");
+            s
+        }
+        Err(e) => {
+            tracing::warn!("ADR-121 mmWave reader: cannot open {port} @ {baud}: {e}");
+            return;
+        }
+    };
+
+    let mut buf = Vec::with_capacity(256);
+    let mut tmp = [0u8; 128];
+    loop {
+        match serial.read(&mut tmp) {
+            Ok(0) => continue,
+            Ok(n) => buf.extend_from_slice(&tmp[..n]),
+            Err(e) => {
+                if e.kind() == std::io::ErrorKind::TimedOut { continue; }
+                tracing::warn!("ADR-121 mmWave reader: read error: {e}");
+                return;
+            }
+        }
+        // Drain complete lines.
+        while let Some(pos) = buf.iter().position(|&b| b == b'\n') {
+            let raw_line: Vec<u8> = buf.drain(..=pos).collect();
+            let line = String::from_utf8_lossy(&raw_line).trim().to_string();
+            if let Some(cm) = parse_distance(&line) {
+                *latest().lock().unwrap() = Some(MmwaveReading {
+                    distance_cm: cm,
+                    at: Instant::now(),
+                });
+            } else if !line.is_empty() {
+                tracing::trace!("mmwave non-distance line: {line:?}");
+            }
+        }
+        // Guard against runaway buffer if module emits non-newline garbage.
+        if buf.len() > 1024 { buf.clear(); }
+    }
+}
+
+/// Parse `distance:<digits>` (HLK-LD2402 Normal Mode line format).
+pub fn parse_distance(line: &str) -> Option<u32> {
+    let lower = line.trim().to_ascii_lowercase();
+    let rest = lower.strip_prefix("distance:")?;
+    rest.parse::<u32>().ok()
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn parses_distance_lines() {
+        assert_eq!(parse_distance("distance:228"), Some(228));
+        assert_eq!(parse_distance("Distance:0"), Some(0));
+        assert_eq!(parse_distance(" distance:42 "), Some(42));
+        assert_eq!(parse_distance("OFF"), None);
+        assert_eq!(parse_distance(""), None);
+        assert_eq!(parse_distance("distance:abc"), None);
+    }
+}

v2/crates/wifi-densepose-sensing-server/static/raw.html

@@ -48,6 +48,23 @@
   <span class="pill" id="lastTs">last: --</span>
   <span class="badge absent" id="globalBadge" style="font-size:13px;padding:4px 12px;">absent</span>
   <span class="pill" id="globalCV">CV 0%</span>
+  <!-- ADR-021: WiFi-CSI vital signs — breathing + heart rate (computed server-side). -->
+  <span class="pill" id="brPill"
+        style="background:rgba(63,185,80,0.18); color:rgb(63,185,80); border:1px solid rgb(63,185,80);"
+        title="WiFi-CSI breathing rate from bandpass 0.1-0.5 Hz on broadband amplitude (ADR-021)">
+    🫁 <b id="brBpm">— BPM</b> <span id="brConf" style="opacity:0.7;font-size:11px">·</span>
+  </span>
+  <span class="pill" id="hrPill"
+        style="background:rgba(248,81,73,0.18); color:rgb(248,81,73); border:1px solid rgb(248,81,73);"
+        title="WiFi-CSI heart rate from bandpass 0.8-2.0 Hz on broadband amplitude (ADR-021)">
+    💓 <b id="hrBpm">— BPM</b> <span id="hrConf" style="opacity:0.7;font-size:11px">·</span>
+  </span>
+  <!-- ADR-121: HLK-LD2402 24 GHz mmWave radar pill — hidden until first reading. -->
+  <span class="pill" id="mmwavePill" style="display:none; background:rgba(33,150,243,0.18);
+                                              color:rgb(33,150,243); border:1px solid rgb(33,150,243);"
+        title="HLK-LD2402 24 GHz radar — distance to closest target">
+    📡 mmWave <b id="mmwaveDist">— cm</b> <span id="mmwaveAge" style="opacity:0.7;font-size:11px">·</span>
+  </span>
   <div class="controls">
     <label>peak-hold <input type="checkbox" id="peakHold" checked></label>
     <label>log-y <input type="checkbox" id="logY"></label>
@@ -352,6 +369,35 @@ function handleSensingUpdate(d) {
   const gcv = document.getElementById('globalCV');
   if (gcv) gcv.textContent = 'CV ' + ((gcl.confidence || 0) * 100).toFixed(1) + '%';
 
+  // ADR-021 — WiFi-CSI vital signs (breathing + heart rate).
+  // `vital_signs` is embedded in SensingUpdate; values may be null
+  // when the detector hasn't accumulated enough history yet (~10s).
+  const vs = d.vital_signs || {};
+  const brBpm  = document.getElementById('brBpm');
+  const brConf = document.getElementById('brConf');
+  const hrBpm  = document.getElementById('hrBpm');
+  const hrConf = document.getElementById('hrConf');
+  const brPill = document.getElementById('brPill');
+  const hrPill = document.getElementById('hrPill');
+  if (vs && typeof vs.breathing_rate_bpm === 'number' && Number.isFinite(vs.breathing_rate_bpm) && vs.breathing_rate_bpm > 0) {
+    if (brBpm)  brBpm.textContent  = vs.breathing_rate_bpm.toFixed(1) + ' BPM';
+    if (brConf) brConf.textContent = '· ' + ((vs.breathing_confidence || 0) * 100).toFixed(0) + '%';
+    if (brPill) brPill.style.opacity = (vs.breathing_confidence || 0) < 0.2 ? '0.5' : '1.0';
+  } else {
+    if (brBpm)  brBpm.textContent  = '— BPM';
+    if (brConf) brConf.textContent = '·';
+    if (brPill) brPill.style.opacity = '0.5';
+  }
+  if (vs && typeof vs.heart_rate_bpm === 'number' && Number.isFinite(vs.heart_rate_bpm) && vs.heart_rate_bpm > 0) {
+    if (hrBpm)  hrBpm.textContent  = vs.heart_rate_bpm.toFixed(0) + ' BPM';
+    if (hrConf) hrConf.textContent = '· ' + ((vs.heartbeat_confidence || 0) * 100).toFixed(0) + '%';
+    if (hrPill) hrPill.style.opacity = (vs.heartbeat_confidence || 0) < 0.2 ? '0.5' : '1.0';
+  } else {
+    if (hrBpm)  hrBpm.textContent  = '— BPM';
+    if (hrConf) hrConf.textContent = '·';
+    if (hrPill) hrPill.style.opacity = '0.5';
+  }
+
   // Per-node level badge from node_features[i].classification (ADR-101).
   const nfNow = performance.now() / 1000;
   const nf = d.node_features || [];
@@ -505,5 +551,33 @@ function connect() {
   };
 }
 connect();
+
+// ── ADR-121: poll HLK-LD2402 mmWave radar @ 5 Hz ─────────────────────
+const mmwavePill = document.getElementById('mmwavePill');
+const mmwaveDist = document.getElementById('mmwaveDist');
+const mmwaveAge  = document.getElementById('mmwaveAge');
+let mmwaveBusy = false;
+async function pollMmwave() {
+  if (mmwaveBusy) return; mmwaveBusy = true;
+  try {
+    const r = await fetch('/api/v1/mmwave/latest', { cache: 'no-store' });
+    if (!r.ok) throw new Error('http ' + r.status);
+    const j = await r.json();
+    if (j && j.available) {
+      mmwavePill.style.display = '';
+      mmwaveDist.textContent = j.distance_cm + ' cm';
+      const age = Math.round(j.age_ms || 0);
+      mmwaveAge.textContent = '· ' + age + ' ms';
+      // Fade pill if stale (>1.5 s) before server hides at 2 s.
+      mmwavePill.style.opacity = age > 1500 ? '0.5' : '1.0';
+    } else {
+      mmwavePill.style.display = 'none';
+    }
+  } catch (_) {
+    mmwavePill.style.display = 'none';
+  } finally { mmwaveBusy = false; }
+}
+pollMmwave();
+setInterval(pollMmwave, 200);
 </script>
 </body></html>

v2/data/adaptive_model.json

@@ -1,267 +0,0 @@
-{
-  "class_stats": [
-    {
-      "label": "absent",
-      "count": 862,
-      "mean": [
-        66.68196972264862,
-        67.23973219951662,
-        65.0340640002779,
-        205.65861248066514,
-        1.2587006960556917,
-        8.192575406032482,
-        0.0,
-        9.823395623712905,
-        6.970045450727901,
-        -0.04488812678641681,
-        -0.9594767860850162,
-        10.78889030301701,
-        0.8330000846014487,
-        22.47189099978742,
-        22.47189099978742
-      ],
-      "stddev": [
-        64.0493846652119,
-        90.27545165651007,
-        40.157907144682206,
-        161.60550836256004,
-        1.3807130815029451,
-        3.2814660018571113,
-        0.0,
-        2.219723108446689,
-        1.6521309619598676,
-        0.342852106459665,
-        0.30620004291079783,
-        3.529722483499124,
-        0.17574148506941875,
-        5.519861526721805,
-        5.519861526721805
-      ]
-    },
-    {
-      "label": "present_still",
-      "count": 852,
-      "mean": [
-        66.39259262094396,
-        64.42298266818027,
-        68.34546366405283,
-        203.34049479166666,
-        1.1900821596244182,
-        8.200704225352112,
-        0.0,
-        10.032339700775715,
-        7.234479413048846,
-        0.027056637948278107,
-        -0.9161490234231624,
-        10.991429347401095,
-        0.8298622589530178,
-        23.588978503428145,
-        23.588978503428145
-      ],
-      "stddev": [
-        59.144593976065984,
-        82.61098004853669,
-        40.08306971525127,
-        152.89405234329087,
-        1.2031203046363153,
-        3.0571012493320526,
-        0.0,
-        2.22294769203091,
-        1.6508044238677446,
-        0.3315329147240876,
-        0.29437997092330526,
-        3.3214071045026303,
-        0.17096813624285292,
-        5.622953396738593,
-        5.622953396738593
-      ]
-    },
-    {
-      "label": "present_moving",
-      "count": 808,
-      "mean": [
-        65.17005228763453,
-        66.55424930761484,
-        63.785855267654334,
-        208.73719832920793,
-        1.3400990099009942,
-        7.570544554455446,
-        0.0,
-        10.069915394050431,
-        6.923405617584522,
-        -0.1440461642917184,
-        -1.0022460352626226,
-        10.664608744841848,
-        0.8384559212414682,
-        21.798331033369895,
-        21.798331033369895
-      ],
-      "stddev": [
-        66.1800697503931,
-        93.22042148141067,
-        42.07226450730718,
-        164.93282045618218,
-        1.3706144246607475,
-        3.1453995481213224,
-        0.0,
-        2.431170975696439,
-        1.672707406405861,
-        0.35643090355922863,
-        0.30897080072710387,
-        3.325911716352165,
-        0.1806597020966414,
-        5.418714527442832,
-        5.418714527442832
-      ]
-    },
-    {
-      "label": "active",
-      "count": 794,
-      "mean": [
-        61.85289600233076,
-        61.12723986655727,
-        62.468831971775344,
-        193.2018524349286,
-        1.2329974811083138,
-        8.083123425692696,
-        0.0,
-        9.747035051350043,
-        7.009904234422278,
-        0.007176072447431498,
-        -0.9950501087764124,
-        11.015545839210892,
-        0.8278984910895401,
-        22.445656559614797,
-        22.445656559614797
-      ],
-      "stddev": [
-        50.44687370766278,
-        74.07914900524236,
-        31.558067649516538,
-        121.0762294406304,
-        1.2507304998955402,
-        3.4503520526220344,
-        0.0,
-        2.2730029390882156,
-        1.6768264387667406,
-        0.3214256392367928,
-        0.31003127617615406,
-        3.1187829194728285,
-        0.1772099351197549,
-        5.595050695741912,
-        5.595050695741912
-      ]
-    }
-  ],
-  "weights": [
-    [
-      0.9923736589617821,
-      -0.4600422332552322,
-      -0.3922101552522972,
-      -0.1686954616947851,
-      -0.08471937018349271,
-      0.033940973559074515,
-      0.0,
-      -1.116294981490482,
-      -0.213861080404439,
-      -0.41727297566573723,
-      0.08025552056009382,
-      0.20864577739519874,
-      0.36814779033318357,
-      0.46242679535538855,
-      0.46242679535538855,
-      0.09475205040199337
-    ],
-    [
-      0.04661470129518883,
-      0.7974124099989739,
-      0.3953040913806362,
-      -1.2708868935843511,
-      0.10073070355913086,
-      0.0735810797517633,
-      0.0,
-      -0.3957608057630568,
-      0.22091779039114648,
-      -0.43105406953304665,
-      0.24907697332262252,
-      -0.17604200203759515,
-      -0.5059663705836186,
-      0.5740861193153091,
-      0.5740861193153091,
-      0.020569218347928304
-    ],
-    [
-      -0.5295363836864718,
-      0.14729609046092632,
-      0.16131671233151712,
-      0.15039859740752318,
-      0.08189110214725194,
-      -0.1429062024394049,
-      0.0,
-      2.459247211223509,
-      -0.162133339181718,
-      0.6345474095048843,
-      0.16626892477248892,
-      0.2710091094981082,
-      -0.08197569509399917,
-      -1.2007197895193034,
-      -1.2007197895193034,
-      -0.10027402587742726
-    ],
-    [
-      -0.5094519765704947,
-      -0.48466626720467487,
-      -0.1644106484598614,
-      1.2891837578716183,
-      -0.0979024355228887,
-      0.0353841491285671,
-      0.0,
-      -0.9471914239699604,
-      0.15507662919500606,
-      0.2137796356938993,
-      -0.49560141865520463,
-      -0.30361288485571664,
-      0.21979427534444013,
-      0.16420687484859928,
-      0.16420687484859928,
-      -0.015047242872495047
-    ]
-  ],
-  "global_mean": [
-    65.08291570815048,
-    64.88537161757283,
-    64.96650236787292,
-    202.8304440905207,
-    1.25474969843183,
-    8.016887816646562,
-    0.0,
-    9.918865477040464,
-    7.036167472733628,
-    -0.038097952045357715,
-    -0.9672836370393502,
-    10.86491812646321,
-    0.8323017200972911,
-    22.58850497890069,
-    22.58850497890069
-  ],
-  "global_std": [
-    60.376895354908775,
-    85.49291935872783,
-    38.814475392686795,
-    151.54766198012683,
-    1.3049002582695195,
-    3.2446975526483737,
-    1e-9,
-    2.2904371592847603,
-    1.667114434239705,
-    0.34470363318292857,
-    0.3067332188136679,
-    3.334427501751985,
-    0.17614366955910027,
-    5.577838072123601,
-    5.577838072123601
-  ],
-  "trained_frames": 3316,
-  "training_accuracy": 0.4149577804583836,
-  "version": 1
-}