AI Wiki · Mediocre Security

Protocols

The connective tissue between models, tools, agents and payments. 9 articles.

ArticleSummaryLinks
Agent Client Protocol (ACP)Protocol, originated at Zed, for how code editors talk to coding agents so one agent can plug into many editors; Android Studio supports 40+ agents through it.2
Agent2Agent (A2A)Google-originated protocol for agents from different vendors to communicate; contributed to the Agentic AI Foundation in August 2026.7
Cross App Access (XAA)Okta-created OAuth extension routing AI agents' access to enterprise apps through the organization's identity policy instead of static API keys; now a formal MCP authorization extension.5
Google Universal Cart (Universal Commerce Protocol)Google's cross-retailer AI shopping cart in Search and the Gemini app, with checkout on its Universal Commerce Protocol (UCP).8
LlamaIndex Document Parsing ProtocolDraft standard (140 rules in v0.1) for how document parsers handle tables, forms and diagrams so agents get consistent structured input.3
Model Context Protocol (MCP)Anthropic-created open standard for connecting AI agents to tools and data; the de facto integration layer of the agent era, now stateless after a breaking 2026 spec revision.42
Personal Agent ProtocolOAuth-based open standard from Sierra and Meta for how personal AI agents authenticate with businesses, with tiered guest, read and write access.8
Trusted Agent Protocol (Visa)Visa's protocol for merchants to verify the identity of AI agents shopping on a customer's behalf; The Next Web puts its reach at about 175 million merchants.4
x402Coinbase-created protocol for AI agents to pay for APIs and services with stablecoin micropayments over HTTP, without accounts or API keys; now governed by the x402 Foundation.6