Protocols, Identity and Payments: A Layered Evidence Review of Agentic Commerce

By mid-2026, each of the four core functions of agentic commerce—interaction, authorization, settlement, and identity and reputation—had at least one protocol implementation: A2A (released April 2025; donated to the Linux Foundation in June 2025), AP2 (announced September 2025; contributed to the FIDO Alliance in April 2026), x402 (launched by Coinbase in May 2025; x402 Foundation operational under the Linux Foundation from July 2026) and ERC-8004 (deployed on Ethereum mainnet on 29 January 2026; EIP text still Draft). This note maps the protocols onto four functional planes by primary function, argues that they have no mandatory dependencies and are best treated as composable modules, and proposes a layered authenticity framework: L1 protocol activity, L2 independent economic activity, L3 service-linked economic activity, and a cross-cutting autonomy-attribution dimension, L4. Two evidence streams are read through this framework. Identity: of the first 10,000 ERC-8004 agents, 6.28% received reputation feedback and 0.67% exposed a service record, with highly concentrated ownership and feedback. Payments: cumulative x402 transactions on Base exceeded 100 million, yet volume figures for the same 30-day window differ by an order of magnitude across data sources and filters, and a graph-based study classifies 21.20% of 136.7 million settlements as fictitious and 63.78% as internal to linked clusters. The evidence sits mainly at L1: L2 bounds depend on filtering method, L3 measurement is nascent, and no population-scale L4 method exists. The note synthesizes standards, announcements, published papers and preprints under a disclosed search protocol; it is not an independent on-chain measurement.