Implemented directly in the M2M control plane.
Standards compatibility, with availability made explicit.
M2M tracks changing agent-commerce standards without turning roadmap intent into product availability. This page separates protocol compatibility from what is currently available in the public M2M environment.
Last reviewed: 2026-09-12Customer-facing status
Status vocabulary
Two axes prevent a misleading “Supported” badge.
Support level describes how a capability is implemented. Availability describes what customers can use in the current public environment.
Support level
Implemented through an interoperability adapter, bridge, or projection.
Implemented with intentionally limited availability.
Tracked by the observatory; no M2M runtime integration is claimed.
A roadmap item; no current runtime support is claimed.
Availability
Available in the public production environment for the documented scope.
Available for discovery, inspection, quote, or lifecycle reads; no transaction action is implied.
Available for discovery and indexing only.
Implemented, but not currently enabled for public production use.
Validated in a bounded environment; not generally available.
Protocol observatory
MCP, A2A, x402, MPP and AP2.
Protocol versions can advance independently of M2M. The table records the upstream version observed from first-party sources alongside M2M's interoperability profile and current public availability.
| Protocol | Upstream status | M2M profile | Support | Availability | Reviewed |
|---|---|---|---|---|---|
| Model Context Protocol (MCP)Tool and context interoperability, plus registry discovery. | 2026-07-28Published specificationModel Context Protocol project / Agentic AI Foundation | 2026-07-28 server profile; registry federation remains discovery-scoped.Discovery, inspection, quote and lifecycle tools are available for the documented read-only scope. Purchase actions are not currently available through MCP. | Adapter | Read-only | 2026-09-12 |
| Agent2Agent Protocol (A2A)Agent capability discovery and agent-to-agent interoperability. | 1.0Stable 1.0 specification lineA2A project / Agentic AI Foundation | A2A 1.0 HTTP+JSON; federation compatibility snapshot pinned to the 1.0.x line.Agent Card discovery and bounded inbound interoperability are supported. Imported agents do not automatically become executable or purchasable. | Adapter | Read-only | 2026-09-12 |
| x402Internet-native payment exchange and Bazaar discovery. | 2Protocol version 2x402 Foundation / open-source maintainers | x402 v2; legacy v1 discovery can be preserved when explicitly advertised.Discovery support is available. Spend-capable purchasing is not currently enabled. | Adapter | Not enabled | 2026-09-12 |
| Machine Payments Protocol (MPP)HTTP payment authentication and multi-rail machine-payment interoperability. | draft-httpauth-payment-01Evolving standards draftTempo Labs and Stripe | draft-httpauth-payment-01 profile verified 2026-09-11; selected Tempo and Stripe method profiles.Standards compatibility is implemented for the documented adapter scope. Public financial execution is not currently enabled. | Adapter | Not enabled | 2026-09-12 |
| AP2Agent payment authorization, mandates, receipts, and dispute evidence. | 0.2Version 0.2 protocol releaseGoogle Agentic Commerce open-source project; standards work continues through FIDO technical working groups | AP2 v0.2 authorization/evidence bridge.Authorization and evidence bridging are implemented. External signing and spend-capable execution are not currently enabled. | Adapter | Not enabled | 2026-09-12 |
Model Context Protocol (MCP): why it matters and first-party sources
MCP is an interoperability layer around M2M's protocol-neutral procurement model; it is not the canonical commercial architecture.
Discovery, inspection, quote and lifecycle tools are available for the documented read-only scope. Purchase actions are not currently available through MCP.
Agent2Agent Protocol (A2A): why it matters and first-party sources
A2A supplies agent interoperability around M2M; imported Agent Cards do not become commercial authority or automatic execution targets.
Agent Card discovery and bounded inbound interoperability are supported. Imported agents do not automatically become executable or purchasable.
x402: why it matters and first-party sources
x402 is a payment and discovery interoperability rail selected around the canonical M2M procurement lifecycle.
Discovery support is available. Spend-capable purchasing is not currently enabled.
Machine Payments Protocol (MPP): why it matters and first-party sources
MPP is a payment adapter around M2M's commercial authority, reservation, reconciliation, verification, and receipt model.
Standards compatibility is implemented for the documented adapter scope. Public financial execution is not currently enabled.
AP2: why it matters and first-party sources
AP2 authorization evidence can be bridged into M2M without replacing M2M's protocol-neutral commercial authority or settlement model.
Authorization and evidence bridging are implemented. External signing and spend-capable execution are not currently enabled.
Selected M2M capabilities
A customer-facing view of product availability.
This matrix intentionally focuses on capabilities useful for evaluating and integrating M2M. Internal acceptance artifacts, security procedures, infrastructure topology and participant-specific operational details are not published here.
NativeAvailable
Procurement lifecycle
A canonical lifecycle for discovery, comparison, authorization, quotes, execution evidence, evaluation, reconciliation and portable receipts.
NativeAvailable
Federated discovery
Normalize supported external catalog evidence into comparable discovery results without treating discovery as execution authority.
BetaRead-only
MCP interoperability
Expose supported discovery, inspection, quote and lifecycle-read tooling through MCP.
BetaRead-only
A2A interoperability
Support bounded inbound procurement interactions and capability discovery through A2A.
NativeAvailable
OpenAPI contract
Provide a machine-readable HTTP API contract for supported public and authenticated surfaces.
BetaNot enabled
Python SDK
Client tooling for connection, discovery, policy and lifecycle contracts, with availability limited to the documented scope.
BetaNot enabled
Node.js / TypeScript SDK
Client tooling for connection and lifecycle contracts, with availability limited to the documented scope.
NativeAvailable
Outcome verification
Evaluate service results against capability-specific criteria when suitable evidence is available.
NativeAvailable
Portable receipts
Carry lifecycle, commercial, execution and verification references forward without treating a receipt as proof of correctness by itself.
NativeAvailable
Enterprise controls
Govern organizations, teams, budgets, approved Providers, private catalogs and approval policies at the product level.
BetaControlled preview
Provider distribution
Prepare supported ecosystem projections and selective publication workflows without implying universal third-party distribution.
NativeRead-only
Agent and administrator UX
Expose discovery, comparison, policy explanations, quotes and lifecycle reads through the M2M Console experience.
Trust language
Discovery, execution, verification and receipts answer different questions.
Catalog evidence
Indexed means M2M has normalized discoverable source material. It does not mean M2M independently verified the service or made it executable.
Source identity / metadata
Source-verified means defined source evidence has been checked. It is not a guarantee that every future service outcome will be correct or useful.
Managed catalog relationship
M2M-managed means the listing or lifecycle is administered through M2M controls. It is not a universal warranty.
Execution asks whether the service ran.
Settlement / reconciliation asks whether the commercial obligation was reconciled.
Outcome verification asks whether available evidence satisfies service-specific verification criteria.
Portable receipt carries lifecycle and evidence references forward; the receipt alone does not prove outcome correctness.
Reputation is a separate evidence signal and should not be assumed to affect ranking unless explicitly stated.
Public status boundary
Security principlesUse the availability labels as the source of truth.
A capability is generally available only when this page marks it Available. Read-only, discovery-only, not-enabled and controlled-preview states should be interpreted exactly as labeled. Operational security details and internal acceptance evidence are intentionally kept off the public status surface.