Open Protocol · v1.0 · TRL 4

The web needs an identity layer for AI agents.

Sovereignty Verification Protocol introduces the Agent Identity Token — a cryptographic credential that lets autonomous systems declare who they are, what they're authorized to do, and at what assurance level, before touching any resource.

agent-identity-token.json
{
  "alg": "ES256",
  "kid": "sp-key-2026-06",
  "typ": "AIT",

  "iss": "sovereignty-protocol",
  "org": "acme-corp",
  "agent_id": "research-agent-001",
  "agent_type": "research",
  "assurance": "AIT-2",
  "scope": ["read:documents", "query:api"],
  "exp": 1789540000,
  "jti": "550e8400-e29b..."
}

✓ ROUTED → VERIFIED AGENT ZONE
Architecture

Two zones. One standard.

Every inbound request is evaluated at the network edge. The routing decision is deterministic, stateless, and takes under 10ms. No user impact. No false positives for human traffic.

VERIFIED AGENT ZONE

Declared agents get a governed path.

Autonomous systems presenting a valid, non-revoked AIT with authorized scope are routed to the VAZ. Every transaction is audited.

  • JWT signature verification at edge
  • Revocation list check (signed, cached)
  • Scope enforcement per resource
  • Immutable audit log for compliance
  • Rate governance per agent type
GENERAL ACCESS ZONE

Undeclared traffic gets scrutiny, not a block.

All undeclared traffic — humans, legacy software, unknown automation — routes to the GAZ. GAZ does not mean blocked. It means implementation-specific risk assessment applies.

  • Human users served normally
  • Bot management integration (Cloudflare, Akamai)
  • Honeypot trap injection for bad actors
  • Session monitoring and risk scoring
  • RTAC lookback on detonation (Stage 3)
  Traffic Arrives
        ↓
  X-Agent-Token header present?
     /              \
   YES               NO
    ↓                 ↓
  Validate          General Access Zone
  Signature         (risk assessment)
  Revocation
  Scope
    ↓
  All pass?
    ↓
  Verified Agent Zone
  (governed, audited)
Identity Assurance

Five levels. One field in the token.

AIT Assurance Levels map directly to NIST SP 800-63 IAL, DoD Zero Trust Pillar 1, and CMMC 2.0. Declare the right level and get the right access — no separate identity review process.

Level Designation Verification NIST Alignment Use Case
AIT-0 Self-Issued Self-signed. No external verification. IAL1 (self-asserted) Development, testing, open APIs
AIT-1 Registry Verified Enrolled in AIR. Signing key registered. IAL1 (verified) Commercial APIs, developer access
AIT-2 Organization Verified Legal entity verification against official records. IAL2 (remote) Enterprise integrations, B2B agent access
AIT-3 Government Validated Identity verified against PIV, CAC, or Login.gov. IAL2 (in-person eq.) Federal contractors, defense industrial base
AIT-4 High Assurance Multi-factor verification, security review, ongoing compliance attestation. IAL3 DoD systems, IC, critical infrastructure, IL4/IL5
The Safe Internet Blueprint

What the protocol protects.

Every internet surface has a different vulnerability profile. Sovereignty Protocol addresses all three with the same underlying primitive — the Agent Identity Token.

Internet Surface
Vulnerability Without Protocol
Protection With Sovereignty Protocol
Human-Facing Websites
Undeclared AI crawlers scraping content without declaration of intent. Publishers cannot distinguish search indexers from AI training crawlers — same fingerprint, different purpose.
AIT forces crawlers to declare agent_type (search-indexer vs. training-crawler). Publishers route each type independently — allow search indexing, gate AI training behind licensing. No loss of search ranking.
Agentic / API Platforms
Hostile AI agents can chain tool calls to exploit back-end system commands or exfiltrate internal data structures. No audit trail when an agent acts outside its scope.
VAZ scope enforcement rejects any agent request outside its declared scope[] at the edge. Every transaction produces an immutable audit record. Out-of-scope attempts notify the AIR in real time.
Sovereign Infrastructure
Disconnected data systems can be systematically mapped and manipulated by cross-border AI models. No EO 14110-compliant mechanism to verify agent provenance at the network perimeter.
AIT-3 and AIT-4 tokens require government-validated identity. AIR maintains revocation with under-60-second propagation. Full audit log satisfies CMMC 2.0 AU.3.045 and AU.3.046 traceability requirements.
Use Cases

Who's implementing the protocol.

🏛️

Federal Agencies

Satisfy EO 14110 and DoD Zero Trust agent identity requirements. AIT-3/AIT-4 tokens with PIV/CAC validation. Immutable audit trail for compliance reporting.

🏢

Enterprise AI Teams

Give your LangGraph, CrewAI, or AutoGen agents a declared identity. Route them cleanly through APIs without triggering bot detection. Full scope governance and revocation.

📰

Publishers & Content Platforms

Differentiate search indexers from AI training crawlers for the first time. Gate training access behind licensing while maintaining search visibility. Reclaim control over your content's destiny.

🔒

Security Teams

Replace reactive bot detection with declarative identity. Composable with Cloudflare, Akamai, and HUMAN. GAZ risk scoring integrates with your existing SIEM infrastructure.

🛡️

Defense Contractors

CMMC 2.0-ready audit logs. AIT-3 government-validated assurance. Supply chain AI agent governance for prime and sub-contractor environments.

⚙️

Platform Builders

Implement the open AIT schema and join the ecosystem. Python, Rust, and JavaScript SDKs on GitHub. Composable with MCP, OpenAI Agents SDK, and any JWT-compatible infrastructure.

Schedule a Technical Briefing

Open standard. Proprietary infrastructure.

The AIT schema is free and open source (CC0). The validation infrastructure, registry, and enterprise features are proprietary SaaS. Same model as HashiCorp / FIDO Alliance.

OPEN SOURCE
AIT Specification
Free
Forever. CC0 license.
The open standard. Everything you need to implement AIT in your own infrastructure.
  • AIT token schema (JWT profile)
  • AIR interface specification
  • Two-zone routing definition
  • Audit record schema
  • Reference implementation
  • Python, Rust, JS SDKs
  • Community support (GitHub Issues)
View on GitHub →
SOVEREIGN REGISTRY
Enterprise / Federal
Request Early Access
Contact Sales
For federal agencies, defense contractors, and enterprises requiring IL4/IL5 support and CMMC compliance.
  • Dedicated AIR instance
  • AIT-3 and AIT-4 enrollment
  • CMMC 2.0 AU.3.045/AU.3.046 logs
  • IL4/IL5 hosting option
  • RTAC engine access
  • FedRAMP authorization track
  • Dedicated support + SLA
  • Custom scope contract engine
Contact us

Get early access to the hosted gateway.

We're onboarding design partners for the Hosted Gateway and Sovereign Registry. Tell us what you're building and we'll reach out within 48 hours.


Or email directly: contact@bravo-01-labs.com