A recent survey of AI agent protocols catalogs the LOKA Protocol's approach to agent identity, and one detail stands out: it's designed around cryptography that doesn't exist as a deployed standard yet. LOKA's Universal Agent Identity Layer (UAIL) assigns each agent a verifiable, decentralized identity, but the survey notes the protocol is anchored in "emerging standards such as Decentralized Identifiers (DIDs), Verifiable Credentials (VCs), and post-quantum cryptography," aiming for a blueprint that stays resilient as computing power advances. That's a different bet than most agent-identity proposals circulating right now. Most current schemes (DID-based tokens, biometric-bound keys, certificate chains) use conventional elliptic-curve or RSA signatures — fine today, but theoretically breakable once large-scale quantum computers exist. Building post-quantum signature schemes into an identity layer now means an agent's credential, and every attestation built on top of it, doesn't need to be reissued or migrated later. It also means accepting overhead most current agent protocols haven't had to deal with: post-quantum signatures tend to be larger and slower to verify than the classical schemes agent-to-agent messaging was designed around. The practical problem is that "post-quantum cryptography" isn't one thing. NIST has standardized a handful of algorithms (ML-KEM, ML-DSA, and others), but adoption in identity infrastructure is uneven, and agent protocols that reference it are mostly gesturing at future-proofing rather than shipping a specific, interoperable scheme. The LOKA paper itself frames UAIL as a proposed layer built on Self-Sovereign Identity principles, with the agent — not a central authority — controlling its own identity. That's a meaningful accountability claim, but it depends on which post-quantum primitive gets picked, and whether other protocols (A2A, MCP, ANP) converge on the same one. If they don't, agents built on different protocols will carry cryptographically incompatible identities even after each individually claims to be quantum-resistant. The survey groups LOKA alongside a dozen other protocols — MCP, A2A, ANP, ACP — none of which currently specify post-quantum requirements in their published specs. That gap is worth watching: identity infrastructure being built today for agents that will operate for years is exactly the kind of system where a cryptographic migration, done late, is expensive and slow. LOKA's inclusion of the requirement now, even as a proposal rather than a shipped implementation, is a marker of where the identity-layer conversation is heading, not evidence that it's been solved.