The Envelope Problem in Device Trust

The trust layer was held together with custom glue. Every time a cloud verifier read a device's integrity claim, it had to negotiate a new pair of glasses for the serialization format and transport on the other side. The interop tax scaled with boundaries, not devices. Call it the envelope problem: when every format and every transport needs its own wrapper, attestation can never become a primitive — it stays a per-pair integration.

Birkholz, Smith, Fossati, and Tschofenig's RFC 9999 names the envelope. The IETF's standards-track wrapper defines one shape for device-attestation messages across four formats — a dedicated CBOR tag, JWT and CWT claims, and an X.509 extension — plus matching media types and CoAP Content-Formats to carry it over HTTP, MIME, and CoAP. The mechanism is portable envelope, not portable payload: a verifier reads the wrapper, then the underlying EAT evidence (RFC 9711) or EAR result, regardless of the serialization or transport underneath.

The two reference implementations — veraison/cmw (Go) and rust-cmw (Rust) — show the approach is feasible. Whether the wrapper becomes a cross-vendor primitive or stalls as a pinned specification with no live traffic is an open question. The lift is real; the layer above it now has somewhere to stand.