Create Your First Agent

Learn how to create an AI agent using both the dashboard and the API. An agent is the core building block of Fluxgate — it connects an AI model with tools, memory, and policies.

What is an Agent?

An agent in Fluxgate is a configuration that defines how an AI model should behave. It includes:

Name — A unique identifier (e.g., "customer-support-bot")
Model — Which AI model to use (e.g., gpt-4o, claude-3, gemini-pro)
System Prompt — Instructions that tell the AI how to behave
Tools — External capabilities the agent can use (APIs, databases, search)
Tags — Labels for organizing and filtering agents

Method 1: Create via Dashboard (Easiest)

The dashboard provides a step-by-step wizard that guides you through creating an agent. No coding needed!

Open the Dashboard

Go to http://localhost:3000/dashboard in your browser. You'll see the Overview tab with platform metrics.

Navigate to the Agents Tab

Click 'Agents' in the left sidebar (look for the robot icon). This shows all your registered agents.

Click 'Create Agent'

Click the 'Create Agent' button in the top right. This opens the creation wizard.

Fill in Basic Details

Enter a name (e.g., 'my-first-agent'), a description (e.g., 'A helpful support agent'), and select a model (e.g., 'gpt-4o'). You can change these later.

Write Your System Prompt

This tells the AI how to behave. For example: 'You are a friendly customer support agent. Answer questions clearly and concisely. If you don't know the answer, say so.' The better your prompt, the better your agent performs.

Attach Tools (Optional)

If you've registered any tools (like web_search or calculator), you can attach them here. Skip this for now — we'll cover tools in the Tools guide.

Save Your Agent

Click 'Create' to save. Your agent will appear in the agents list with a 'draft' status. You can edit it anytime.

Method 2: Create via API

If you prefer working with code or want to automate agent creation, use the REST API:

Create Agent Requestbash

curl -X POST http://localhost:8000/api/v1/agents \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "my-first-agent",
    "description": "A helpful customer support agent",
    "model": "gpt-4o",
    "system_prompt": "You are a friendly customer support agent. Answer questions clearly and concisely. If you don't know the answer, say so.",
    "tools": [],
    "tags": ["support", "v1"]
  }'

Understanding the Response

When you create an agent, the API returns the full agent object:

Agent Responsejson

{
  "id": "agt_abc123def456",
  "name": "my-first-agent",
  "description": "A helpful customer support agent",
  "model": "gpt-4o",
  "system_prompt": "You are a friendly customer support agent...",
  "tools": [],
  "tags": ["support", "v1"],
  "status": "draft",
  "version": 1,
  "created_at": "2024-01-15T10:30:00Z",
  "updated_at": "2024-01-15T10:30:00Z"
}

Save your Agent ID!

Copy the "id" field from the response. You'll need it to run the agent in the next step.

Agent Configuration Fields

Field	Type	Required	Description
name	string	Yes	Unique name for the agent. Use lowercase and hyphens.
description	string	No	Human-readable description of what the agent does.
model	string	Yes	AI model to use (gpt-4o, claude-3-sonnet, gemini-pro, etc.)
system_prompt	string	Yes	Instructions for the AI model. This is the most important field.
tools	string[]	No	List of tool IDs the agent can use. Empty array = no tools.
tags	string[]	No	Labels for organizing agents. Useful for filtering.

Pro tip: Writing good system prompts

Be specific! Instead of "Be helpful", try "You are a customer support agent for an e-commerce store. Answer questions about orders, shipping, and returns. Always ask for the order number first."

Run Your First Agent