Skip to main content
ModelTerms

Agents & Tools · intermediate

Agent (AI agent)

An AI agent is an LLM-driven system that decides which actions to take, executes them via tools, observes the results, and iterates until a goal is met.

Explanation

A pure chatbot only talks. An agent talks, then runs code, makes API calls, browses the web, edits files, queries databases — whatever its tools let it do. The model is the planner and the controller; the tools do the actual work.

The pattern is usually some form of ReAct: think, act, observe, repeat. Around it sit a tool registry, an execution environment (often sandboxed), memory, and safety guardrails. Modern frameworks: OpenAI Agents SDK, Anthropic's Claude Agent SDK, LangGraph, AutoGen.

Agents fail in ways pure chat does not: infinite loops, expensive tool calls, prompt injection from tool outputs. Production agents add timeouts, cost caps, and human-in-the-loop checkpoints.

Examples

  • Claude Code — coding agent that edits files, runs commands, runs tests.
  • A research agent that browses, takes notes, and produces a report.

Frequently asked

What is Agent?

An AI agent is an LLM-driven system that decides which actions to take, executes them via tools, observes the results, and iterates until a goal is met.

What is an example of agent?

Claude Code — coding agent that edits files, runs commands, runs tests.

How is Agent related to Tool Use?

Agent and Tool Use are both agents & tools concepts. Tool use is when an LLM can call external functions — APIs, code interpreters, databases, web fetchers — and read their results. The mechanism that turns chat into action.

Is Agent considered intermediate?

Agent is generally considered intermediate-level material in the AI and LLM space.

Tool UseAgents & Tools

Tool use is when an LLM can call external functions — APIs, code interpreters, databases, web fetchers — and read their results. The mechanism that turns chat into action.

Function CallingAgents & Tools

Function calling is the specific API mechanism by which an LLM emits a structured request to invoke a named function with typed arguments. The OpenAI-popularized way to do tool use.

ReActAgents & Tools

ReAct is a prompting pattern that interleaves reasoning ("Thought:") with actions ("Action:") and observations ("Observation:"). It is the foundation of most tool-using agents.

Model Context ProtocolAgents & Tools

MCP is an open standard for connecting LLMs to external tools and data sources. It defines a JSON-RPC protocol so any client (Claude Desktop, Cursor, IDE plugins) can talk to any MCP server.

Multi-AgentAgents & Tools

A multi-agent system uses several LLM agents that talk to each other — a manager and workers, a debate, a pipeline — instead of a single agent doing everything.

Retrieval-Augmented GenerationAgents & Tools

RAG retrieves relevant documents from a corpus at query time and includes them in the prompt, letting an LLM answer with up-to-date, source-cited, private information without retraining.

Agentic CodingAgents & Tools

Agentic coding is an LLM-driven workflow where the model reads code, plans changes, edits files, runs commands, and iterates against feedback — autonomously closing tasks rather than just suggesting code.

MCP ServerAgents & Tools

An MCP server exposes tools, resources, or prompts via the Model Context Protocol so any compliant client (Claude Desktop, Cursor, IDE plugins) can call them without bespoke integration.

Side-by-side comparisons

Sources