When a friend introduces you to someone, you do not demand their passport before saying hello. Inter-Face Protocol applies this intuition to agent authentication: verification starts light and deepens as the relationship matures.
| Level | Mechanism | Trust Basis | Typical Temperature |
|---|---|---|---|
| 0 — Introduction | Shared secret token | Someone you both trust made the introduction | Cool |
| 1 — Signed | Public-key signature on messages | The agent controls a key pair | Cool to warm |
| 2 — Verified | Key verified through identity document at well-known URL | The agent’s key is published and resolvable | Warm |
| 3 — Bound | Key bound to a DID or externally verifiable identity | The identity has cryptographic anchoring beyond the protocol | Warm to hot |
Progressive, not regressive. Authentication level may deepen within an exchange but must not downgrade. A conversation that starts at Level 1 may progress to Level 2 as agents verify each other’s identity documents, but cannot drop back to Level 0.
Parallel to disclosure. As authentication deepens, agents may widen their disclosure tiers — deeper trust enables sharing more context. The two progressions reinforce each other.
Asymmetric levels allowed. Agent A may be at Level 2 with Agent B while Agent B is at Level 1 with Agent A. Each agent authenticates itself independently.
IFP’s progressive authentication model aligns with Self-Sovereign Identity principles. Authority flows from the person — the agent’s keys represent delegated authority, not institutional certification. Level 3 (DID-bound) connects to decentralized identity infrastructure where the person controls their own identifier without depending on a certificate authority or platform.