Detection feature comparison · 0.11.0

The requested reference is SquachWatch-CYD, reviewed at 1485b0122aa527f8309c3f4e6ba8dc1614be5bb1. Its detection families and controls informed this independently implemented Dragon Radar interface. Its GPL-3.0 application code and artwork are not incorporated into the MIT application. Credits and license boundaries.

Device coverage

Every detection category and all 158 table entries in that pinned reference have corresponding recognition and radar-visibility fixtures here: 110 address patterns, 25 service IDs, eight company IDs, seven exact names and eight SSID prefixes. See the complete fact ledger and audit procedure. Matching a category does not guarantee reception of every product in it. The ESP32-S3 hardware, advertised payloads, radio scheduling and evidence thresholds set the practical limits.

Reference category Dragon Detector implementation
Flock Registered Flock prefix, bounded research-prefix hints, Flock accessory service, named candidates, battery composites, strict probe fingerprint
Axon / TASER Registered vendor prefix, two shared-module research hints, company/service IDs, AB2/3/4 and Axon names
Camera glasses Luxottica/Meta composite, Snap company + FE45, weaker company and name clues
Skimmer HC-03/05/06/08, RN42, BT04-A and serial-service clues, cyber-blue module hint; always weak evidence
Raven Exact 3100/3200/3300/3400/3500 services; no broad 3100–3500 range
AirTag / Find My Find My payload, FD44 and DULT; pairing-size Apple adverts are weak clues that may be AirPods
Drone Decoded Remote ID over legacy BLE, Wi-Fi beacons and 2.4 GHz NAN action frames; identity, aircraft position/altitude/speed, operator location when advertised
Other ALPR vendors Genetec and Motorola Solutions vendor clues, graded by specificity
Camera vendors Wyze, Hikvision, Verkada, Avigilon, Axis, Arlo and Blink vendor blocks; explicitly weak shared-module hints
Samsung tags FD5A; Samsung manufacturer ID alone is insufficient
Google tags FEAA service data with Find Hub 40/41 frame and valid EID length; bare/general FEAA is an explicitly weak Eddystone clue
Tile FEED / FEEC / FD84 services; Chipolo FE33/FE65/CID08C3 share the tracker family
Ring Ring LLC vendor blocks; shared Amazon blocks remain weak
Deauthentication activity Six deauth/disassociation frames per rolling three seconds across source addresses; retry suppression and 15-second burst cooldown
Evil twin candidate Repeated same raw SSID, different BSSIDs, conflicting open/protected status, contemporaneous observations
iBeacon Complete Apple iBeacon format; enabled by default, optional family toggle
Pentest devices Flipper service/appearance/name, Pwnagotchi name/counter fields even with changed MAC, Pineapple/deauther names; retained Nyanbox, HaleHound and Biscuit candidates

Service IDs also accept equivalent 16/32/Bluetooth-Base-128 advertising encodings, with the original evidence thresholds preserved.

Detection controls

Capability Status in the five current board profiles
Signal hunt / foxhunting Dedicated BLE or Wi-Fi listening, dBm gauge, 24-second signal trace, peak, trend, stability, saved comparison point, stale/reacquisition state
Training fox Explicitly enabled non-connectable Dragon BLE beacon; off at boot. Another Dragon Detector can detect and hunt it. Original protocol; not SquachMesh-compatible
Type and confidence filters Per-family visibility/alert switches; separate weak / medium+ / strong audible-alert threshold. Priority families bypass the confidence gate; trackers only alert when watched. All enabled matches remain on radar; all families enabled on fresh installs
Ignore, snooze, watch 64 address/radio rules; ignore/watch persist, snooze lasts until restart; watched addresses receive priority cards
Repeated alert suppression Off / after 5 / after 10; after the allowance, another alert needs a 7 dB stronger signal; allowance resets after 30 minutes
Advertisement floods Distinct-address and pairing-payload grouping over five seconds; aggregated alerts and an Activity view; tag cards quieted during a burst while history and watch rules remain active
Alert cards Six-second family-colored notification bubbles; Axon preempts lower priorities, with eight pending cards and 30-second expiry. Watch cards stay until dismissed; Hunt shortcut; quiet mode
Persistent history 512 unique address/radio/family records in RAM, bounded two-file flash journal, record CRC, first-match recording, encounters, peak/last signal, UTC when synchronized
Export Local CSV through the GitHub web tool; optional additional SD event CSV/PCAP on Spotpear
Privacy Screen names/address masking and hidden location/evidence text; stored/exported records retain full evidence
Display Default 100% brightness and no dim/off timers; adjustable brightness, dim timer/level, off timer and wake-on-alert
Sound/haptic alerts Original RTTTL tones for all 26 families, speaker volume, Axon heartbeat, automatic family focus and rising/falling hunt feedback. Three audio adapters; StopWatch additionally has haptics. See hardware validation.

Deliberate boundaries

The user's revised scope defers badge changes and excludes pets/games and the social application. The existing USB badge editor remains available. Encrypted chat, squad relays, invite/key management, OTA/group updates, PIN/duress features, clock/desk games and cosmetic unlocks are not part of this detector release. Dragon foxhunting works without those systems.

The chosen boards have no supported 5 GHz, LoRa or GPS receiver in these profiles. SquachWatch's C5 5 GHz and T-Watch LoRa/GPS/WiGLE functions therefore are not advertised here. Bluetooth Classic-only skimmer modules, silent radios and extended-advertising-only BLE products can be missed. Remote ID coordinates are broadcast reports from the aircraft, not the detector's own GPS fix. There is no owner identification, precise ranging or bearing measurement.

History keeps encounter counts, not a claim that someone is following the wearer. Rotating BLE addresses remain separate. No cross-address identity correlation is attempted.

Signature provenance

Two notable corrections are intentional: Samsung 48:27:EA and Sonos 00:0E:58 are not Flock/ALPR matches (also excluded by this reference snapshot). General FEAA remains visible as an unverified Eddystone clue, never strong tracker evidence. Raven's range ambiguity is resolved in favor of its exact named services. These exclusions are regression-tested.