Core Components of an Agent

Introduction

In this lesson, you will learn each component of an Aura Agent and how they work together.

Components

An Aura Agent consists of four main components:

LLM: Language model that reasons and generates responses
Tools: Cypher Template, Similarity Search, and Text2Cypher (read-only)
Knowledge graph: Your AuraDB graph — structured, connected data for graph-native retrieval
Orchestrator: Coordinates reasoning, tool calls, and response generation

mermaid

Agent components: perceive, reason, act with tools, retrieval, and knowledge graph

%%{init: {
  "theme": "base",
  "securityLevel": "strict",
  "fontFamily": "Public Sans, Arial, Helvetica, sans-serif",
  "themeVariables": {
    "background": "#F5F7FA",
    "primaryTextColor": "#014063",
    "fontSize": "14px",
    "lineColor": "#485662"
  }
}}%%
flowchart TD
    A["AGENTS<br/>Perceive, Reason, Act"]:::agent
    T["TOOLS<br/>Cypher Template, Similarity Search, Text2Cypher"]:::tool
    R["RETRIEVAL<br/>GraphRAG & Vector Search"]:::retrieval
    KB["KNOWLEDGE BASE<br/>Neo4j Graph"]:::memory

    A -->|Invokes capabilities| T
    A -->|Fetches relevant context| R
    A -->|Queries graph| KB
    R -.->|Queries graph and vectors| KB

    classDef agent fill:#014063,stroke:#014063,stroke-width:2px,color:#fff;
    classDef tool fill:#0369a1,stroke:#7dd3fc,stroke-width:2px,color:#fff;
    classDef memory fill:#166534,stroke:#86efac,stroke-width:2px,color:#fff;
    classDef retrieval fill:#b45309,stroke:#fcd34d,stroke-width:2px,color:#fff;

The agent perceives input, reasons about which tools to use, and acts by invoking tools and fetching retrieval context from the knowledge graph. Retrieval queries both the graph structure and vector index in AuraDB.

mermaid

Aura Agent architecture: configuration, invocation, and tooling

%%{init: {
  "theme": "base",
  "securityLevel": "strict",
  "fontFamily": "Public Sans, Arial, Helvetica, sans-serif",
  "themeVariables": {
    "background": "#F5F7FA",
    "primaryTextColor": "#014063",
    "fontSize": "14px",
    "primaryColor": "#eef6f9",
    "primaryBorderColor": "#014063",
    "lineColor": "#485662"
  }
}}%%
flowchart LR
    subgraph IN["Configuration & Invocation"]
        CONSOLE["Aura Console UI<br/>Create, configure, test"]
        API["Customer Application<br/>API HTTP Endpoint"]
        MCP["Customer MCP Client<br/>via Aura Agent MCP Server"]
    end

    AGENT["Aura Agent"]

    LLM["LLM"]
    DB["Aura DB"]
    TOOLBOX["Toolbox<br/>Text-to-Cypher<br/>Cypher Template<br/>Similarity Search"]

    CONSOLE <--> AGENT
    API --> AGENT
    MCP --> AGENT
    AGENT --> LLM
    AGENT <--> DB
    AGENT --> TOOLBOX

    style AGENT fill:#014063,stroke:#014063,color:#fff
    style CONSOLE fill:#eef6f9,stroke:#014063
    style LLM fill:#E7FAFB,stroke:#006E58
    style DB fill:#edf6e8,stroke:#006E58
    style TOOLBOX fill:#edf6e8,stroke:#006E58
    style API fill:#eef6f9,stroke:#014063
    style MCP fill:#eef6f9,stroke:#014063

The Aura Agent sits at the center. Configure it in the console, invoke it via API or MCP, and the agent uses the LLM, AuraDB, and tools to answer queries. For details on each tool type, see Aura Agent documentation.

Role of Each Component

LLM: Interprets user intent, selects tools, and generates natural language answers
Tools: Cypher Template runs predefined queries; Similarity Search uses vectors; Text2Cypher turns natural language into Cypher. All are read-only.
Knowledge graph: The AuraDB graph stores facts. Cypher and GraphRAG retrieve them.
Orchestrator: Manages the workflow from user input through LLM reasoning, tool invocation, result integration, and response

How Components Work Together

Example flow:

User asks a question → Orchestrator passes it to the LLM
LLM reasons that the "Get Customer" tool is needed
Orchestrator invokes the Tool, which runs Cypher against the Knowledge graph
Tool returns rows → Orchestrator adds them to context
LLM generates a response using the tool results

Check Your Understanding

Orchestrator Role

Which component coordinates reasoning, tool calls, and response generation?

❏ The knowledge graph only
❏ The LLM only
✓ The orchestrator
❏ The graph only

Hint

One component coordinates the flow between the LLM, tools, and response generation.

Solution

The orchestrator.

The orchestrator coordinates the agent’s reasoning, invokes tools when needed, and manages the flow of context and response generation.

Summary

In this lesson, you learned the core components: LLM, tools, knowledge graph, and orchestrator.

Building Agents in Neo4j Aura

Architecture and Components of Aura Agents

Design and Implement Simple Aura Agents

Hands-on Challenges

Core Components of an Agent

Introduction

Components

Role of Each Component

How Components Work Together

Check Your Understanding

Orchestrator Role

Summary

Chatbot