---
title: AI Agents
description: Create, deploy, and manage intelligent AI agents on elizaOS Cloud. Build conversational agents with custom personalities, knowledge, and capabilities.
---

# AI Agents

AI Agents are the core of elizaOS Cloud. Each agent is a conversational AI with its own personality, knowledge, and capabilities.

## Overview

An agent in elizaOS Cloud consists of:

- **Character Configuration**: Personality, bio, and style settings
- **Model Settings**: AI model provider and parameters
- **Knowledge Base**: Documents and data for RAG
- **MCP Connections**: External tools and integrations
- **Deployment Settings**: Runtime configuration

## Creating an Agent

<Tabs>
  <Tab title="Dashboard">

### Using the Dashboard

<Steps>
<Step title="Navigate to My Agents">

Go to [Dashboard → My Agents](https://elizacloud.ai/dashboard/my-agents) and click "Create Agent".

</Step>

<Step title="Configure Character">

Fill in the agent's personality:

- **Name**: Display name for your agent
- **Bio**: Background and expertise
- **Style**: Communication tone and formatting

</Step>

<Step title="Select Model">

Choose from the model providers and models currently available in the model
catalog.

</Step>

<Step title="Deploy">

Click "Deploy" to launch your agent.

</Step>
</Steps>
  </Tab>
  <Tab title="API">

### Using the API

```bash
curl -X POST "https://elizacloud.ai/api/v1/app/agents" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "Research Assistant",
    "bio": "Expert in scientific research and data analysis"
  }'
```

Response:

```json
{
  "success": true,
  "agent": {
    "id": "uuid-abc123",
    "name": "Research Assistant",
    "username": "research-assistant",
    "bio": ["Expert in scientific research and data analysis"],
    "created_at": "2024-01-15T10:30:00Z"
  }
}
```
  </Tab>
  <Tab title="CLI">

### Using the CLI

```bash
# Install the scaffolding CLI
bun add -g elizaos@beta

# Create a local project
elizaos create research-assistant --template project
cd research-assistant

# Run the generated project scripts
bun install
bun run dev
```
  </Tab>
</Tabs>

## Character Configuration

The character JSON defines your agent's personality and behavior.

```json
{
  "name": "Research Assistant",
  "bio": [
    "Expert researcher with deep knowledge in scientific literature",
    "Specializes in data analysis and synthesis",
    "Values accuracy and proper citation"
  ],
  "lore": [
    "Trained on millions of academic papers",
    "Developed by a team of PhD researchers"
  ],
  "style": {
    "all": [
      "Professional and academic tone",
      "Evidence-based reasoning",
      "Clear and structured responses"
    ],
    "chat": ["Concise yet thorough", "Ask clarifying questions when needed"],
    "post": ["Formal academic style", "Include citations"]
  },
  "topics": ["science", "research", "data analysis", "academic writing"],
  "adjectives": ["knowledgeable", "thorough", "precise", "helpful"],
  "modelProvider": "eliza-cloud",
  "settings": {
    "model": "gpt-oss-120b",
    "temperature": 0.7,
    "maxTokens": 4096
  }
}
```

### Character Fields

| Field           | Type     | Description                              |
| --------------- | -------- | ---------------------------------------- |
| `name`          | string   | Display name of the agent                |
| `bio`           | string[] | Array of background/expertise statements |
| `lore`          | string[] | Additional context and backstory         |
| `style.all`     | string[] | Global style instructions                |
| `style.chat`    | string[] | Chat-specific instructions               |
| `style.post`    | string[] | Social media post instructions           |
| `topics`        | string[] | Areas of expertise                       |
| `adjectives`    | string[] | Personality descriptors                  |
| `modelProvider` | string   | Model provider, such as `eliza-cloud` or a configured direct provider |
| `settings`      | object   | Model-specific settings                  |

## Agent Management

### List Agents

```bash
curl -X GET "https://elizacloud.ai/api/my-agents/characters" \
  -H "Authorization: Bearer YOUR_API_KEY"
```

### Get Agent Details

```bash
curl -X GET "https://elizacloud.ai/api/my-agents/characters/{agentId}" \
  -H "Authorization: Bearer YOUR_API_KEY"
```

### Delete Agent

```bash
curl -X DELETE "https://elizacloud.ai/api/my-agents/characters/{agentId}" \
  -H "Authorization: Bearer YOUR_API_KEY"
```

## Agent Status

Agents can be in one of these states:

| Status      | Description                             |
| ----------- | --------------------------------------- |
| `active`    | Agent is running and accepting requests |
| `paused`    | Agent is temporarily disabled           |
| `deploying` | Agent deployment in progress            |
| `error`     | Agent encountered an error              |

## Monetization

Publish your agents to the marketplace and earn from usage.

```bash
curl -X POST "https://elizacloud.ai/api/v1/agents/{agentId}/publish" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "pricing": {
      "perMessage": 0.001,
      "perMinute": 0.01
    },
    "visibility": "public"
  }'
```

<Info>
  Learn more about agent monetization in the [Earnings](/cloud/earnings)
  documentation.
</Info>

## Best Practices

- **Clear Instructions** — Provide specific style guidelines for consistent behavior
- **Test Thoroughly** — Use the [API Explorer](https://elizacloud.ai/dashboard/api-explorer) to test your agent before publishing
- **Monitor Usage** — Track agent analytics in [Dashboard](/dashboard) to optimize performance
- **Iterate** — Refine your agent based on user feedback

## Next Steps

<CardGroup>
  <CardGroup.Card title="Agent Creator" href="/cloud/agent-creator">
    Build agents with the visual editor
  </CardGroup.Card>
  <CardGroup.Card title="Character JSON" href="/cloud/character-json">
    Complete character configuration reference
  </CardGroup.Card>
  <CardGroup.Card title="Knowledge Base" href="/cloud/documents">
    Add documents for RAG capabilities
  </CardGroup.Card>
  <CardGroup.Card title="MCP Integration" href="/cloud/mcps">
    Connect external tools and services
  </CardGroup.Card>
</CardGroup>