This page lists categories of reports we have already evaluated and found not to qualify for a reward under the Bug Bounty Program. It exists so researchers — and any tooling they use to generate reports — can check beforehand whether a finding is likely to fall into an already-addressed category, saving everyone’s time.
This list does not replace the program terms. If anything here appears to conflict with the Eligibility, Bounty Focus, or Out-of-Scope sections of Bug Bounty Program, those sections are authoritative.
1. Deterministic key derivation across tools
The device derives its identity key as a function of the device secret
(UDS), the loaded app’s digest, and the User Supplied Secret (USS):
CDI = BLAKE2s(UDS, app_digest, USS). On a specific device, the same
app, loaded with the same USS, will always derive the same key — in
any client tool, on any operating system. This is the intended mechanism
by which a user proves continuity of identity across sessions.
Demonstrating that two tools agree on this key is not a vulnerability.
2. Signing or authorization issues that do not extract a secret
A report may correctly show that a device signs data under an attacker’s control — for example, a local process using an agent socket to request a signature, or a touch-confirmation prompt that does not describe the request accurately. These can be real usability or authorization concerns, but they do not qualify for a reward unless a secret (UDS, CDI, or a key derived from them) is actually extracted from the device. Using a key is not the same as leaking it. We are however still interested to hear about such cases, either per mail or in a GitHub issue.
3. Attacks that require physical possession of the device
Our threat model and the program cover digital and software attacks against the device, such as attacks from a compromised host over USB. An attack chain that depends on an attacker gaining physical access to or possession of the TKey — including scenarios combining a host compromise with later theft, loss, or physical tampering with the device — is out of scope, even if the software portion of the chain would otherwise be in scope.
4. Findings against code outside a released firmware or app
Findings against development branches (such as main), archived or
experimental repositories, hackathon prototypes, or developer-only
test tooling are out of scope, even when the code is public and runs
on a covered device. Coverage is limited to the firmware releases
listed in the Eligibility section and to the app releases listed
via tillitis.se/download.
5. Weak or guessable User Supplied Secret (USS)
Our threat model assumes the user chooses a hard-to-guess USS, and states that, given one, the device should withstand an attack. Reports that recover a deliberately weak or short test USS via brute force, dictionary attack, or similar — without a separate flaw that would also defeat a strong USS — fall within this assumption and do not qualify.
6. Latent code-quality issues with no shipped-build exploit
Reports that point out fragile patterns (for example, a missing defensive qualifier that happens not to be triggered by any compiler flags we currently ship with) do not qualify for a reward unless a working exploit is demonstrated against a currently shipped firmware or app build — a theoretical risk to a future rebuild is not, on its own, a vulnerability in a released product. Please join the community and open an issue on GitHub instead.
A note on report quality
None of the above is about how well a report is written. Reports in these categories can be extremely thorough — detailed root-cause analysis, working proof-of-concept code, and careful verification. Thoroughness does not change whether a finding is in scope or eligible for a reward; please check against this list and the program terms before investing time in a detailed write-up.