You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Security: OCPP AuthorizationKey is readable via GetConfiguration and logged on telnet #203
Incident context (shared by all five): on 2026-10-04 05:45 a local OpenVAS scan ran against
the LAN. Both units (7388 master, 7803 slave, bm-2026.09.5) stayed associated to the AP
(UniFi "online") but stopped answering ARP, ping, HTTP, telnet. Charging via RFID kept
working. A UniFi "reconnect client" on 2026-10-05 ~19:40 restored the network without a reboot
(OCPP counters were not reset). Two sessions charged during the outage never reached
e-Flux (ocpp_tx_id: 0 in /session/last) and had to be replayed by hand.
Problem
MicroOcppMongooseClient.cpp:122 declares AuthorizationKey with the default accessible = true, so GetConfiguration returns the key value to the backend. In debug builds
the reply is printed on the telnet console ([MO] Send: ...), which has no authentication. In the
2026-10-05 capture the line was truncated by the console before the value, but that depends on
key order and line length. OCPP 1.6 security whitepaper: AuthorizationKey is write-only, and a read
must not return the value. This also conflicts with the CLAUDE.md "never log secrets" rule (C-5).
Specification
Scenario: GetConfiguration does not disclose the key
Given AuthorizationKey is set
When the backend sends GetConfiguration (all keys or AuthorizationKey explicitly)
Then the key is either omitted or returned with an empty value
Scenario: console traffic log redacts the key
Given a debug build with MO traffic logging
When any frame containing "AuthorizationKey" is logged
Then the value is redacted
Activity
fix(api): don't echo the OCPP auth key in the POST /settings reply
Merge pull request #206 from basmeerman/fix/ocpp-authkey-redact
basmeerman commentedon Oct 5, 2026
Fixed in #206, released in bm-2026.10.1 and installed on both units (2026-10-05). The fix has not yet been exercised by a real incident on hardware.