Security

Security is the product here, not a feature — so we treat reports as gifts. If you have found a weakness in the YGOOW relay, the Android app, or the cryptography, we want to hear from you.

How to report

Email security@ygoow.com. A machine-readable contact lives at /.well-known/security.txt (RFC 9116). For encrypted reports, use our PGP key — fingerprint 4B01 B8B1 132F 33C5 DC06 7DC2 6893 22E3 4899 BB1B, the same key that signs our warrant canary.

Please tell us what you found, how to reproduce it, and the impact you believe it has. A working proof-of-concept helps us confirm quickly.

What we commit to

• No legal action against good-faith research that respects the boundaries below. Test against your own devices and your own data.
• Acknowledgement within a few days, and honest updates as we work the issue.
• Credit when a fix ships — publicly if you want it, silently if you do not.
• Coordinated disclosure: give us a reasonable window to fix before going public. We will not sit on it.

In scope

• The relay server — API, WebSocket, store-and-forward, one-time token handling.
• The Android app — key handling, on-device storage, the embedded Tor path.
• The cryptographic core — key derivation, AEAD (“Variant C”), the forward-secrecy ratchet, the Double Ratchet and its post-quantum hybrid (X25519 + ML-KEM-768), Shamir quorum, identity and QR exchange.

We are especially interested in the parts we built ourselves rather than inherited: the deterministic bootstrap that replaces a prekey server, the hybrid combiner, and the transactional decryption path. If you can desynchronise a live ratchet, forge a state transition, or show that our ML-KEM implementation deviates from FIPS 203 on any input, we want that report before anyone else gets it.

Out of scope

• Social engineering, physical attacks, or anything targeting our users or staff.
• Denial-of-service or volumetric testing against the live relay.
• Findings that require an already-compromised device — we say plainly that no messenger survives that.
• Automated-scanner output with no demonstrated impact.

What we don’t pretend

We do not run a paid bug bounty yet, and we will not imply otherwise. What we offer is a fast, honest, no-lawyers response and public credit. As the project matures — and ahead of the independent audit we have committed to — this will grow.