Technical
Minyma is built around hybrid post-quantum end-to-end encryption. Every post-quantum mechanism runs alongside a proven classical mechanism, giving message and attachment content layered protection for today's internet and the post-quantum transition.
Initial key agreement is hybrid PQXDH, combining classical X25519 elliptic-curve Diffie-Hellman with ML-KEM-768 (NIST FIPS 203) post-quantum key encapsulation, with downgrade protection that prevents the post-quantum part from being silently stripped. The Double Ratchet then mixes a fresh ML-KEM-768 secret into the key schedule at every ratchet step, and each message is bound with dual Ed25519 and ML-DSA-65 (NIST FIPS 204) signatures.
The relay handles registration, queueing, push delivery, and attachment relay without needing the private keys required to read plaintext content.
Because confidentiality is hybrid and continuous, recorded ciphertext stays protected against a future cryptographically-relevant quantum computer. Minyma is resistant to "harvest now, decrypt later." This page highlights the working security, delivery, voice, translation, and attachment layers behind Minyma so the technical story stays as clear as the product experience.
Each device carries long-lived local identity material and publishes the prekey state needed to bootstrap direct messaging. Every content message is bound with dual signatures - classical Ed25519 and post-quantum ML-DSA-65 (NIST FIPS 204) - so impersonation is resisted even against a quantum adversary.
Sessions begin with hybrid PQXDH (X25519 + ML-KEM-768) key agreement with downgrade protection, so the classical and post-quantum parts are bound together instead of being silently weakened. Sessions then run a Double Ratchet that mixes a fresh ML-KEM-768 secret into the key schedule at every step, delivering post-quantum forward secrecy and post-compromise security on an ongoing basis, not just at session start.
The relay supports registration, queueing, push delivery, and attachment relay. It still sees some operational metadata even though it is not intended to hold plaintext content.
Minyma combines encrypted signaling, sealed-sender delivery for current call traffic, and WebRTC media transport for voice and video calling on iPhone and iPad.
Minyma Voice is designed as an in-app operator. It can understand natural language, but it executes only local tools Minyma exposes. Voice can open chats, New Message, New Group, settings, public read-aloud, WeatherKit weather, language selection, and local pickers, while sensitive actions are staged for review.
Voice notes are recorded locally, previewed before sending, encrypted through the existing attachment flow, and designed to expire after first listen.
Architecture summary
Minyma combines live hybrid post-quantum encryption, network privacy, multi-device session design, relay-backed calling, private call signaling, encrypted voice notes, and voice-agent tool boundaries inside one coherent stack.
The White Paper is the public starting point for Minyma's hybrid post-quantum encryption story. Minyma is built on the platform's vetted cryptographic implementations of NIST FIPS 203 (ML-KEM) and FIPS 204 (ML-DSA); it is under continuous internal security testing and is queued for independent third-party cryptographic review. From there, the documents lane can expand with protocol notes, release security updates, implementation notes, and review summaries when they are ready to publish.
Minyma protects message content end to end and keeps the delivery boundary clear. The relay handles routing, queueing, push delivery, and attachment transfer without holding the private keys required to read plaintext content. Optional voice sessions and translation are separate, user-started product actions with explicit boundaries.
Limited delivery metadata supports routing, queueing, push, and attachment transfer so the product stays fast and reliable.
