Skip to content

Security: Dendifra/DeskUnlock

Security

SECURITY.md

Security policy

DeskUnlock is authentication software and is currently beta software (v0.1.0-beta.1). Do not treat it as independently certified or as a replacement for evaluating the security of the Linux host and phone.

Reporting a vulnerability

If GitHub Private Vulnerability Reporting is enabled, use it for security-sensitive reports. Otherwise open a minimal non-sensitive issue asking the maintainer for a private contact path.

Do not publish keys, tokens, bond secrets, private device identifiers, signing material, exploit details, or proof-of-concept material in a public issue.

Security scope

Please report concerns involving:

  • authentication bypass or fail-open behavior;
  • replay, relay, signature, challenge, or pairing validation;
  • incorrect PAM return behavior or privilege boundaries;
  • leakage of key material or unsafe state permissions;
  • authentication of the wrong user or device.

DeskUnlock uses local BLE/GATT communication and Android biometric/Keystore APIs. No DeskUnlock cloud account or developer-operated backend is required. The Android app does not request INTERNET in the committed manifest.

The release APK is signed with a dedicated DeskUnlock release certificate. The private release key is outside this repository and must never be committed or published. Users should verify the SHA-256 checksum published with a beta APK.

Limitations

No independent professional security audit has been completed. DeskUnlock cannot guarantee safety against:

  • a compromised Linux host, PAM stack, or root account;
  • a compromised or physically coerced phone;
  • an attacker who knows the phone's device credential;
  • malicious or broken Android, Bluetooth, DMS, or other operating-system components;
  • physical attacks outside the normal host and phone threat model.

The configured password or other PAM fallback should remain available when DeskUnlock is unavailable. See docs/security-model.md for the operator-facing model.

There aren't any published security advisories