PulseAugur
EN
LIVE 09:36:51

AI Agent Definition: Class vs. Instance for MCP Hosts

The article proposes a precise definition for AI agents, distinguishing between an agent's class (definition) and its instance. The agent definition includes components like the host loop, system context, tool policies, LLM, termination limits, task contract, and context-size strategy. An agent instance, on the other hand, comprises the specific goal, working context window, credentials, and consumed budget. This definition is designed to be operational and serializable, particularly for agents operating within an MCP host environment, and addresses limitations in existing definitions by providing a more concrete structure. AI

IMPACT Provides a structured definition for AI agents, potentially aiding in the development of more robust and interoperable agent systems.

RANK_REASON The item is an opinion piece by an individual author defining a technical concept within the AI agent space.

Read on dev.to — MCP tag →

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →

AI Agent Definition: Class vs. Instance for MCP Hosts

COVERAGE [1]

  1. dev.to — MCP tag TIER_1 English(EN) · Jonathan Langens ·

    What is an 'agent'? A class/instance definition, stress-tested against the 2026-07-28 MCP spec

    <h1> Industry standards </h1> <p>In the Gen AI / LLM space, two standards have emerged as the de facto choice in their domains:</p> <ul> <li> <strong>A2A</strong> (v1.0, Linux Foundation) defines how agents integrate and communicate with their peers. It defines the <em>interface<…