"Security audit" might be a little misleading, because all systems are secure until proven otherwise. However, we can make a valiant effort.
Here are my ideas so far of how we can hopefully achieve some reasonable degree of security:
1) glorious amounts of unit tests - unit tests for every failure state, unit tests for handling malicious behavior (partial, Issue #8)
2) logging on the server (missing, Issue #6)
3) mechanism for issuing new certs and automatically distributing them to validated clients (missing, Issue #2)
4) good documentation on proper usage (missing, Issue #7)
5) good documentation on how the device pairing process works and other security related features features (missing, Issue #9)
With all of that in place, encouraging clever and mischievous friends to play around with the system and try to break it.
"Security audit" might be a little misleading, because all systems are secure until proven otherwise. However, we can make a valiant effort.
Here are my ideas so far of how we can hopefully achieve some reasonable degree of security: 1) glorious amounts of unit tests - unit tests for every failure state, unit tests for handling malicious behavior (partial, Issue #8) 2) logging on the server (missing, Issue #6) 3) mechanism for issuing new certs and automatically distributing them to validated clients (missing, Issue #2) 4) good documentation on proper usage (missing, Issue #7) 5) good documentation on how the device pairing process works and other security related features features (missing, Issue #9)
With all of that in place, encouraging clever and mischievous friends to play around with the system and try to break it.