Post-quantum readiness

What you retire

• Waiting on a standalone "PQ migration project" while harvest-now risk accumulates on minting identities
• Assuming harder keys alone fix steal-and-replay of lasting request credentials
• Treating KEM choice as a substitute for burn-before-validate and request binding

Recommended stack

Cloud or Sovereign - Foundry mints hybrid agent-mode circuits (X25519 with Kyber768 key agreement); Gate continues to enforce request-bound envelopes with Control's durable nonce ledger. Keep your TLS posture. Scope: PQ key agreement on the prover path under the same envelope contract - not a claim that quantum adversaries are defeated in every model. Computational hardness assumptions versus empty effective channel under the named model are stated on Technology · Math and quantum bounds.

One-minute capability description

Waiting for a future PQ cutover leaves harvest-now-decrypt-later risk unaddressed on the identities that mint authority. ENI6MA hybrid circuits include Kyber768 alongside X25519 for key agreement today, while Gate continues to enforce request-bound envelopes (one-time cryptographic authorizations for a single request) with burn-before-validate. Harder keys can still be stolen; one-shot authority and transcript silence address reuse and channel leakage separately from the key encapsulation mechanism. Absolute replay and binding claims still require the reference architecture (durable ledger, integrity checks, lifecycle), independent of KEM choice. The developer post-quantum page states what is and is not claimed.