Open Standard

SSI Protocol
Sovereign Synthetic Intelligence

An open governance protocol for autonomous AI systems.

SSI v1.0 defines cryptographic audit trails, layered policy enforcement, and deterministic safety boundaries — implementation-agnostic and open to all.

SSI Protocolv1.0|Status:Stable|CC BY 4.0

Protocol architecture

SSI is a layered protocol. Agents MUST NOT execute high-impact actions without a prior Decision from a Kernel. That is the foundational invariant.

AI Agents / Models / Services
Requesting systems
SSI Gateway
Protocol normalization & routing
SSI Kernel
Policy evaluation & decision engine
RPX Audit Chain
Immutable, cryptographically signed records

Implementations MAY colocate components — the logical separation MUST be preserved.

Protocol primitives

Four normative components defined in the SSI v1.0 specification. Any conformant implementation provides all four.

Governance Envelopes
Policy containers

Versioned, signed policy documents that wrap AI systems with constitutional constraints, operational rules, and escalation paths. The primary mechanism for encoding governance into SSI.

Most-restrictive-wins. Deny-by-default.
RPX Record
Request-Permission-Execution

The atomic unit of the audit trail. Every permission evaluation produces exactly one RPX record capturing the request, the decision, the reasoning trace, and the execution result — cryptographically signed.

Ed25519 signed. Append-only. Replayable.
SSI Kernel
Decision engine

The normative evaluation core. Loads applicable Governance Envelopes, evaluates requests through a deterministic pipeline, computes a Decision with a full reasoning trace, and emits a signed RPX record.

4 kernel safety invariants. Fail-closed.
SSI Gateway
Protocol entry point

Normalizes incoming requests from AI agents in any supported protocol format, routes them to Kernel instances, and returns Decisions. Supports HTTP/REST; gRPC and WebSocket optional.

Single-kernel, multi-kernel, federated.

Conformance levels

Any system satisfying the MUST requirements in SSI v1.0 may claim conformance.

Level 1Core Protocol

Development, research, proof-of-concept

  • Gateway HTTP/REST interface
  • Kernel evaluates against Governance Envelopes
  • Signed RPX records persisted to append-only store
  • Deny-by-default enforced
  • All 4 Kernel safety invariants satisfied
Level 2Production Ready

Production deployments, regulated industries

  • Everything in Level 1
  • Multi-layer envelope evaluation
  • Most-restrictive-wins conflict resolution
  • Full reasoning trace in every RPX record
  • Human escalation queue for ESCALATE policies
  • Emergency intervention mechanism
Conformance claim format (SSI v1.0 §8)

[Product Name] implements SSI Protocol v1.0, Level [1|2].

DeAlgo implements SSI Protocol v1.0, Level 2.See product

The standard is open. The implementation is yours to choose.

Built on SSI.
Governed by design.

DeAlgo is the first enterprise platform implementing SSI Protocol v1.0. Start governing your AI agents in under 60 seconds.