Every term you need to understand the agent economy — from MCP and A2A to Agent Cards and agent readiness.
Whether you're a developer building agent integrations, a business owner preparing for AI-mediated commerce, or just trying to keep up with the fastest-moving space in tech, this glossary has you covered.
A2A is Google's open protocol for agent-to-agent communication. It defines how AI agents discover each other, negotiate capabilities, and collaborate on tasks across organizational boundaries. A2A uses Agent Cards for discovery and supports both synchronous and streaming interactions.
Why It Matters
As the agent ecosystem grows, agents need to talk to each other, not just to humans. A2A provides a standardized way for your business's agent to receive requests from and collaborate with agents from other platforms like Google, Salesforce, or SAP.
ACP is a standard developed by OpenAI and Stripe for enabling AI agents to make purchases, handle subscriptions, and manage billing on behalf of users. It builds on Stripe's payment infrastructure and defines how agents authenticate, authorize payments, and handle receipts.
Why It Matters
If you want AI agents to be able to buy your products or services, you need a payment protocol they understand. ACP is one of two emerging standards (alongside UCP) that will determine how the agent economy handles money.
The ACP Score is the payment readiness sub-score within the Agent Readiness Score. It measures whether a business's payment infrastructure is accessible to AI agents through protocols like ACP, UCP, x402, or structured pricing data.
Why It Matters
A high ACP score means agents can complete the full transaction loop with your business, from discovery all the way through payment, without requiring human intervention.
An Agent Card is a machine-readable JSON file hosted at /.well-known/agent-card.json that describes an agent's or business's capabilities, supported protocols, authentication methods, and service endpoints. It serves as a digital business card that other agents can automatically read and understand.
Why It Matters
Agent Cards are how AI agents discover what your business can do. Without one, agents have to guess or rely on web scraping. With one, they instantly know your capabilities, how to authenticate, and what protocols you support.
The Agent Economy is the emerging economic system where AI agents autonomously discover, evaluate, use, and pay for services on behalf of humans and organizations. It encompasses the full lifecycle of machine-to-machine commerce, from search and discovery through fulfillment and payment.
Why It Matters
The agent economy is projected to reach $3-5 trillion by 2030. Businesses that are agent-ready now will capture outsized market share as autonomous purchasing becomes mainstream. Those that aren't will become invisible to an entire generation of AI-mediated buyers.
The Agent Readiness Score is AgentHermes's proprietary 0-100 metric that measures how ready a business is for AI agents. It evaluates nine dimensions: Discovery, API Accessibility, Onboarding, Pricing Transparency, Payment Protocols, Data Quality, Security, Reliability, and Agent Experience. Scores map to tiers: Platinum (90+), Gold (75-89), Silver (60-74), Bronze (40-59), and Not Scored (below 40).
Why It Matters
Your Agent Readiness Score is the single number that tells you whether AI agents can find, understand, and do business with you. The average score across 139+ businesses is just 39/100, meaning most businesses are nearly invisible to the agent economy.
agent-hermes.json is AgentHermes's proposed standard file for declaring a business's agent readiness status, supported protocols, available tools, and integration endpoints. Hosted at the root of a website, it provides a comprehensive machine-readable manifest that goes beyond what individual protocol files offer.
Why It Matters
While agent-card.json covers capabilities and llms.txt covers content, agent-hermes.json ties everything together into a single declaration of agent readiness. It tells agents not just what you can do, but how ready you are to work with them.
AGENTS.md is a Linux Foundation standard for declaring agent capabilities within code repositories. It is a markdown file placed in the root of a repository that describes what AI coding agents should know about the project, including build instructions, architecture decisions, and contribution guidelines.
Why It Matters
As AI coding agents like Claude, Copilot, and Cursor become standard development tools, AGENTS.md ensures they understand your codebase. It's the developer-facing equivalent of llms.txt, optimized for repository context rather than website content.
ARL is AgentHermes's 7-level maturity framework that categorizes businesses by how ready they are for AI agent interaction. The levels are: ARL-0 (Dark) for businesses with no web presence, ARL-1 (Static) for basic websites, ARL-2 (Structured) for businesses with structured data, ARL-3 (Accessible) for those with APIs, ARL-4 (Integrated) for full protocol support, ARL-5 (Autonomous) for fully automated agent workflows, and ARL-6 (Interoperable) for businesses that participate in multi-agent ecosystems.
Why It Matters
ARL gives you a clear roadmap. Instead of a single score, it shows you exactly where you are on the maturity curve and what specific steps will move you to the next level. Most businesses today are ARL-1 or ARL-2.
Fulfillment Routing is AgentHermes's system that intelligently routes agent requests to the appropriate business system for completion. It supports four channels: direct API calls for fully integrated businesses, webhooks for event-driven systems, email notifications for businesses without APIs, and lead capture forms for businesses that need human follow-up.
Why It Matters
Not every business has an API. Fulfillment routing means even a local plumber without any technical infrastructure can receive and respond to agent-generated leads. It bridges the gap between the agent economy and businesses at every technology level.
The AgentHermes Gateway is a unified API that provides agents with a single connection point to access multiple business services. Instead of integrating with each business individually, agents connect to the gateway once and gain access to every connected service through standardized tool schemas, authentication, and billing.
Why It Matters
The gateway solves the N-to-N integration problem. Without it, every agent would need to individually integrate with every business. The gateway provides one API key, one authentication flow, and one billing system for the entire network.
Hosted MCP is a service where AgentHermes runs and maintains an MCP server on behalf of a business. Instead of the business deploying their own MCP infrastructure, AgentHermes hosts MCP endpoints that expose the business's tools and data, handling protocol compliance, authentication, and uptime.
Why It Matters
Most businesses don't have the technical resources to deploy and maintain MCP servers. Hosted MCP makes agent readiness accessible to any business, regardless of technical sophistication. You describe what you offer, and we handle the protocol layer.
KYA (Know Your Agent) is an identity verification framework for AI agents. Analogous to KYC (Know Your Customer) in finance, KYA establishes standards for verifying an agent's identity, authorization level, the principal (human or organization) it represents, and the scope of actions it is permitted to take.
Why It Matters
As agents start making purchases and accessing sensitive services, businesses need to know who they're dealing with. KYA prevents unauthorized agent actions, establishes accountability, and enables trust in agent-to-business transactions.
llms.txt is a machine-readable text file placed at the root of a website (e.g., example.com/llms.txt) that helps large language models understand the site's content, structure, and purpose. It provides a concise, AI-optimized summary that is more efficient than having an LLM crawl and parse an entire website.
Why It Matters
When an AI agent needs to understand what your business does, llms.txt gives it the answer in seconds instead of requiring it to crawl your entire site. It's the fastest path from "invisible" to "discoverable" in the agent economy.
MCP (Model Context Protocol) is Anthropic's open protocol for connecting AI models to external tools, data sources, and services. It defines a standardized way for AI agents to discover available tools, understand their parameters, invoke them, and process results. MCP supports resources (data the agent can read), tools (actions the agent can take), and prompts (templates for common interactions).
Why It Matters
MCP is becoming the dominant standard for how AI agents interact with the world. If your business exposes MCP tools, any MCP-compatible agent (Claude, and increasingly others) can immediately use your services. It's the API layer of the agent economy.
An MCP Server is a server that implements the Model Context Protocol to expose tools, resources, and prompts to AI agents. It acts as the bridge between a business's internal systems and the AI agents that want to interact with them. MCP servers can run locally (stdio) or be hosted remotely (HTTP/SSE transport).
Why It Matters
Running an MCP server is how a business makes its capabilities directly callable by AI agents. It is the difference between an agent reading about your services on your website and an agent being able to actually book an appointment, check availability, or place an order.
UCP (Universal Commerce Protocol) is a standard developed by Google and Shopify for enabling AI agents to browse, compare, and purchase products across e-commerce platforms. It defines standardized product schemas, cart operations, and checkout flows that agents can navigate programmatically.
Why It Matters
UCP is designed to make every online store agent-shoppable. If you sell products online, UCP compliance means AI shopping agents can browse your catalog, compare prices, add items to cart, and complete purchases, all without human intervention.
A Vertical Template is a pre-built set of MCP tool schemas, configurations, and best practices designed for a specific business type. AgentHermes provides 15 vertical templates covering industries like restaurants, HVAC, lawn care, dental, legal, SaaS, e-commerce, and more. Each template includes 5-7 tools tailored to the vertical's common operations.
Why It Matters
Vertical templates eliminate the need to design your agent integration from scratch. A restaurant gets book_table, get_menu, and check_availability tools pre-configured. An HVAC company gets schedule_service, get_quote, and check_coverage. You just plug in your data.
In the AgentHermes system, a Wallet is a pre-funded account that agents use to pay for services through the gateway. Businesses and agent operators deposit funds into their wallet, and transactions are deducted automatically as agents consume services. Wallets support real-time balance checks, usage tracking, and automatic top-ups.
Why It Matters
Wallets solve the payment friction problem for agent commerce. Instead of requiring credit card authorization for every transaction, agents can spend from a pre-approved balance, enabling fast, automated, sub-second payments at scale.
x402 is an HTTP-native micropayment protocol that uses the 402 Payment Required HTTP status code. When an agent requests a paid resource, the server responds with a 402 status and payment details (amount, accepted methods, payment address). The agent completes the payment and retries the request with proof of payment in the headers.
Why It Matters
x402 enables true pay-per-use pricing for AI agents. Instead of requiring subscriptions or API keys, businesses can charge per-request at the protocol level. It is especially powerful for micropayments, where an agent might pay fractions of a cent per API call.