Protocol Roadmap 2026–2027 Milestones
V1 delivered with delegation, chain linking, and content attestation. Building toward production SDKs and framework integration.
V2 Production
Device key rotation, batch revocation, enterprise SDKs
Framework Tools
LangChain, AutoGen, CrewAI, LlamaIndex integrations
Research
Threshold credentials, conditional disclosure, zero-knowledge
V1 — Complete
The V1 protocol specification is frozen with delegation, chain linking, and content attestation as first-class features.
Delegation Credentials
- ✓ Scoped authority with sub-delegation up to depth 5
- ✓ Time and scope attenuation (monotonic narrowing)
- ✓ Delegation chain verification (9-step pipeline)
- ✓ Cascade revocation via SMT parent checks
Chain Linking & Content
- ✓ Tamper-evident ordered credential logs
- ✓ Content attestation for document provenance
- ✓ Creation method tracking (human, AI-assisted, AI-generated)
- ✓ Hash chain integrity verification
Protocol Foundation
- ✓ 21 domain separators, 10-step verification pipeline
- ✓ 47 error codes across 8 categories
- ✓ 63 conformance test vectors
- ✓ no_std core with WASM targets
Multi-Tier Agent Delegation (V1)
Each level cryptographically attenuates authority. Compromise at any level is contained to that scope. All of this ships in V1.
V2 Production Hardening
Enterprise-ready features for production agent deployments
Core Protocol
- • Device key rotation without credential reissuance
- • Batch revocation for incident response
- • Extended conformance test suite (1000+ vectors)
- • OID4VP + COSE interoperability
Enterprise SDKs
- • TypeScript SDK with full type safety
- • Python SDK with asyncio support
- • Java/Kotlin bindings for Android
- • Swift bindings for iOS/macOS
Goal: Production deployments in Fortune 500 enterprises with 99.99% uptime requirements
Agent Framework Integration
Native support in major AI agent frameworks
LangChain
Credential-aware tools and chains with built-in verification
AutoGen
Multi-agent conversations with credential-based authority
CrewAI
Role-based agent crews with cryptographic delegation
LlamaIndex
RAG pipelines with credential-gated data access
# Two-line integration in any framework
from langchain.tools import ExqubTool
@requires_credential(spend_cap=1000)
async def book_flight(agent, destination):
# Authority automatically verified
return await airline_api.book(destination)Goal: Every major agent framework supports Exqub credentials natively
Research Directions
Threshold Credentials
k-of-n multi-party issuance for distributed authority. Require multiple signers to issue high-value credentials.
Conditional Disclosure
Predicate proofs such as “age > 18” without revealing the actual age. Privacy-preserving attribute verification.
Zero-Knowledge Credentials
Prove properties about credentials without revealing the credential itself. Enable privacy-preserving authority verification.
Cross-Chain Interoperability
Bridge credentials across blockchain networks. Enable multi-chain agent operations with unified authority.
Formal Verification
Machine-checkable proofs of protocol correctness using Lean 4. Prove security properties mathematically.
Blind Signatures
Issuer signs a credential without seeing its content. Enable unlinkable credential issuance for maximum privacy.
Shape the Future of Agent Trust
The roadmap evolves based on community needs and technological advances. Join our design partner program to influence the protocol's direction.