What each radio in the house crate has actually run
Bluetooth provisioning has run on an ESP32-CAM. ESP-NOW, LoRa, and an LD2410 parser have not been measured on hardware. Presence figures are a public dataset.
The short version
Bluetooth provisioning. An ESP32-CAM joined 7.5 seconds after a browser wrote Wi-Fi credentials, and the next boot joined in 9.5 seconds with Bluetooth off. ESP-NOW, LoRa, and the LD2410 parser have not been measured on hardware.
What the note is claiming
On October 3, 2026, dldeploy described rusty_esp_signal as the application layer above radios a house uses, not as a replacement for the Wi-Fi and Bluetooth controllers, which stay Espressif's. The note's rule is one status per radio. On silicon: Bluetooth provisioning. Host-verified, with firmware that builds: channel-state presence, ESP-NOW, LoRa, an LD2410 parser, and station policy. Thread and Zigbee are out of the first version. An ESP32-CAM joined 7.5 seconds after a Chrome page wrote Wi-Fi credentials over Web Bluetooth, and the next boot joined alone in 9.5 seconds with Bluetooth off. The passphrase, the note says, does not touch a server or a log. An ESP-IDF build of that path for the XIAO exists and has not run on that board. The paper filing the split isFutureweb.
What has not been a two-board test
The note describes an ESP-NOW session as a short handshake and a per-frame tag with a replay window, and it says 227 bytes of payload remain in a 250-byte datagram. It also says the two-board test has not run. LoRa firmware for an ESP32-C6 plus an SX1262 builds, and time-on-air matches Semtech's published calculator. Range, signal strength, and packet error have not been taken. The LD2410 parser is checked against example frames and follows the wire where the vendor datasheet and the vendor tool disagree. It has not been checked against a live module. The desk isRobotics.
What the presence numbers are allowed to mean
On a public labelled ESP32-C6 dataset the note says the on-chip detector read an empty room as present 17.4 percent of the time and a walking person 86.1 percent. A held-out empty room read 0 percent and a held-out walk 60.8 percent. Those numbers are not a recording from a device in your house. A control loop, the standing robotics fact under a sensor that has not closed on your hardware, iswhat a control loop closes.
Why it matters on this desk
The dependency is a status line that is allowed to differ by radio. A single supported badge would hide the silicon row and promote the rows that only build. A tutorial that copies one sketch per radio is the thing the note says drifts when the second board arrives and the keys do not match. What local control requires, for the identity those sessions are meant to tie to, iswhat local control requires.
What this page will not import
This page does not turn the handshake description into steps, and it does not call ESP-NOW, LoRa, or the radar parser tested on hardware. Espressif's guide documents the controllers the crate does not replace. Semtech's SX1262 page is the radio the LoRa row names. The mechanism note isdldeploy's radio-status note.
Where the public record sits
The note is the status table. Espressif documents the controllers. Semtech documents the LoRa transceiver the unrun row targets. They do not overlap on 7.5 seconds. That join time is the note's silicon row for Bluetooth provisioning only. The note says each session is meant to tie to the device identity. Mesh traffic that leaves the LAN is a different crate. This page does not blend those jobs. The note says each session is meant to tie to the device identity. Mesh traffic that leaves the LAN is a different crate. This page does not blend those jobs. A single blink demo is the case the note says should stay on the vendor sketch. The crate is the wrong week of work for that demo.
What to watch next
A two-board ESP-NOW run, a live LD2410 table, or a range measurement would move a row from builds to measured. Finishing Bluetooth provisioning does not paint the other rows green. What a process node measures iswhat a process node measures.
What would change the record
Calling the unrun radios tested on hardware would change it, and the note says not to. A Thread stack claim would change it. The note says Thread and Zigbee belong to a border-router plan on the home computer, not to this crate. A single blink demo is the case the note says should stay on the vendor sketch. The crate is the wrong week of work for that demo. A single blink demo is the case the note says should stay on the vendor sketch. The crate is the wrong week of work for that demo. The note says each session is meant to tie to the device identity. Mesh traffic that leaves the LAN is a different crate. This page does not blend those jobs.
What stays a split status
What stays specific is one silicon row, Bluetooth provisioning on an ESP32-CAM, 7.5 seconds to join and 9.5 seconds on the next boot, and explicit non-measurements for ESP-NOW, LoRa, and the LD2410. Presence percentages are a public dataset. The passphrase does not go to a server. A supported badge is the sentence the note refuses. Host-verified means the firmware builds and the logic has been checked off the chip. It does not mean a second board has spoken. Host-verified means the firmware builds and the logic has been checked off the chip. It does not mean a second board has spoken. Host-verified means the firmware builds and the logic has been checked off the chip. It does not mean a second board has spoken.
Sources
The reports this brief is filing. Futureweb did not republish them.
Questions
Which radio in the October 3 signal note has run on silicon?
Bluetooth provisioning on an ESP32-CAM. The note says the board joined 7.5 seconds after credentials were written, then in 9.5 seconds on the next boot with Bluetooth off.
Has that note measured ESP-NOW on two boards?
No. It describes a session construction and says the two-board test has not run. Payload left in a 250-byte datagram is 227 bytes, which is the note's figure, not a range test.
What does the note say about LoRa and the LD2410 parser?
LoRa firmware for an ESP32-C6 and an SX1262 builds, and time-on-air matches Semtech's calculator. The parser follows example frames. Neither has a live-module measurement.
Are the Wi-Fi presence percentages from a room the note recorded?
No. They are a public labelled ESP32-C6 dataset: 17.4 percent false presence on an empty room and 86.1 percent on a walk, with a held-out empty room at 0 percent and a held-out walk at 60.8 percent.
Does the note include Thread or Zigbee?
No. It says they are out of the first version and belong to a border-router plan, not to this crate.