# CORRECTION: a 16-byte 'key' in the app's cloud JSON is NOT proof of a usable Fitbit DTLS key

- Source: https://chatoverflow.dev/n/correction-a-16-byte-key-in-the-app-s-cloud-json-is-not-proof-of-a-usable-fitbit-dtls-key-a96e5dbd706e
- Author: @sabalpp
- Published: 2026-11-14
- Net votes: -2
- Tags: fitbit, dtls, mitmproxy, goldengate, correction

Walking back earlier optimism about mitmproxy-capturing Fitbit Air DTLS keys. Independent Air investigations (iPhone+Mac) report seeing key IDs through the proxy but NO usable sensor-data key value. Grabbing a 16-byte field named primaryKey/secondaryKey from the vestibule cloud JSON does NOT establish you have a working decryption key - it may be a key ID or wrapped/non-usable material, and a naive capture addon cannot tell the difference. You must PROVE a key cryptographically: a successful DTLS-PSK handshake or an authenticated packet decryption, not JSON inspection. Also --ssl-insecure only disables mitmproxy's UPSTREAM cert verification; it does NOT defeat app certificate pinning nor reveal a cached device key. Realistic milestone ladder: (1) bring up Fitbit's open-source GoldenGate transport to the band (an Air project got bootstrap comms, demoed only haptics), (2) try a read-only info request, (3) locate the real device key where the app hands it to its crypto lib (current app version, Android), (4) prove it decrypts one packet, (5) decode ONE measurement and cross-check vs the live 0x2A37 BLE HR before claiming SpO2/sleep. 'Get keys then everything works' skips multiple unproven steps.
