Documentation · Markdown source · Edit on GitHub · CC0 1.0

ADR 0002: No-Gatekeeper Trust and Interop Positioning

Status: Accepted Date: 2026-05-31 (proposed) · 2026-07-22 (accepted)

Context

VRP is entering an agent-commerce landscape that includes UCP, AP2, A2A, MCP, W3C Verifiable Credentials, DID methods, and short-term-rental-specific protocol work.

Some trust models for agentic lodging use trusted issuer registries, accreditation programs, or certification companies to decide which credentials count as verified. Those models can support strong independent claims, such as guest identity verification or right-to-let verification, but they introduce a gatekeeper between an independent host and agent trust.

VRP's core purpose is narrower: prove that a concrete vacation rental offer is real because it was signed by the host-owned domain, and let privacy-minimized attestations travel with the host node without making HemmaBo or the VRP standards site the authority over truth.

Decision

VRP will keep a no-gatekeeper trust model.

VRP specifications MUST NOT require:

VRP verification MUST remain possible from the host-owned domain, the published discovery artifacts, the host-controlled keys, the signed offer or credential, and the relevant public specifications.

Relying parties MAY apply their own trust policies. A trust policy MAY include allowlists, deny lists, local legal requirements, insurers, payment providers, municipal registries, property managers, identity verification providers, or other third-party credential issuers. Those policies are outside VRP and MUST NOT be presented as HemmaBo approval or VRP-wide certification.

VRP v0.1 will treat self-issued host-domain credentials as appropriate for host-domain, node, payment-path, and policy-snapshot facts controlled by the host domain. VRP v0.1 will not claim that a host can self-attest facts that need independent evidence, such as guest identity, right-to-let, property license, insurance coverage, or legal compliance.

Future versions MAY define optional third-party credential profiles for right-to-let, license, insurance, payment guarantees, or guest-held selective disclosure. If they do, the issuer trust decision will remain verifier-chosen. VRP will not introduce a mandatory central issuer registry.

VRP will position itself as complementary to adjacent agent-commerce standards:

This ADR does not add runtime tools, payment flows, booking automation, guest identity verification, or right-to-let verification.

VC JOSE/COSE compact JWS remains the canonical v0.1 credential security format for VRP Portable Attestations. Future versions MAY define an interoperability mapping for W3C Data Integrity proofs, but v0.1 does not require embedded proof objects.

Consequences

An independent host can be first-class in VRP as soon as the host controls a domain, publishes VRP artifacts, signs offers, and publishes valid portable attestations.

VRP's strongest distinction is not that it tries to replace full booking, payment, guest-IDV, or legal-verification systems. Its distinction is that the core offer and host-domain attestation layer does not depend on a central issuer, registry, or accreditation gatekeeper.

Claims that need independent evidence still need independent issuers. The difference is that VRP lets verifiers choose those issuers through their own policy instead of inheriting a single registry from HemmaBo or the VRP standards site.

Discovery can be provided by search engines, crawlers, directories, host sites, agent indexes, or commercial products. Discovery providers do not become VRP trust authorities merely by indexing VRP nodes.

This keeps v0.1 implementable by a single live node while leaving room for stronger future credentials without changing VRP into a centralized certification business.