---
title: "AI agent — Glossary | LLMind"
description: "An LLM-powered program that can plan, call tools, and act in a loop to accomplish a goal."
url: https://llmind.org/glossary/ai-agent/
source_format: html
---
[← Glossary](https://llmind.org/glossary/)

# AI agent

**An LLM-powered program that can plan, call tools, and act in a loop to accomplish a goal.**

An AI agent is a program that uses an LLM to reason about how to accomplish a goal, then repeatedly takes actions (calling tools) to move toward that goal. Unlike a chatbot that only generates text, an agent reads the output of each action, decides what to do next, and continues until the goal is met or a stopping condition is reached.

## What makes it an agent

The tool-use loop defines an agent: read input, decide what to do, call a tool (search the web, read a file, call an API, run code), interpret the result, decide the next action. This cycle repeats. An agent can search for information, synthesize it, refine its reasoning, and execute complex multi-step plans — all driven by the LLM's reasoning.

## MCP-powered agents

Claude Desktop, Cursor, Windsurf, and custom agents that speak the Model Context Protocol can connect to MCP servers. These servers expose tools as standardized capabilities. MCP is the standard protocol for agents to access external tools and file systems.

## LLMind relationship

Agents that need file context are a primary LLMind use case. Enriched files give agents rich semantic metadata through any MCP filesystem server. Rather than searching a vector DB, agents can read metadata-rich files directly and act on instant, precise context.

## Related terms

-   [MCP](https://llmind.org/glossary/mcp/)
-   [Context window](https://llmind.org/glossary/context-window/)
-   [RAG](https://llmind.org/glossary/rag/)

## See also

-   [AI agent file access](https://llmind.org/use-cases/ai-agents-file-access/)
-   [MCP server](https://llmind.org/mcp/)
-   [Learn](https://llmind.org/learn/)
-   [Spec](https://llmind.org/spec/)