> ## Documentation Index
> Fetch the complete documentation index at: https://noorle.com/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Creating an Agent

> Step-by-step guide to create your first AI agent in the Console.

Create intelligent agents in minutes. Define behavior with system prompts, choose models, and attach capabilities.

## Step-by-Step Guide

<Steps>
  <Step title="Navigate to Agents">
    1. Go to **Agents** in left sidebar
    2. Click **Create Agent** button
  </Step>

  <Step title="Basic Information">
    Fill in:

    * **Name** - Identifier (e.g., "Research Agent")
    * **Description** - What it does (e.g., "Researches topics and summarizes")
    * Click **Next**
  </Step>

  <Step title="System Prompt">
    Write instructions for agent behavior:

    ```
    You are a helpful research assistant.
    - Search for accurate information
    - Provide sources
    - Ask clarifying questions
    ```

    Or use a template. Click **Next**.
  </Step>

  <Step title="Model Selection">
    Choose routing strategy:

    * **Smart Routing** - Platform picks best model (recommended)
    * **Specific Model** - You choose model

    For Smart: set optional tier (Budget, Standard, Advanced)

    Click **Next**.
  </Step>

  <Step title="Attach Capabilities">
    1. Click **Add Capability**
    2. Choose type (Built-in, Connector, Plugin)
    3. Select specific capabilities
    4. Click **Add**
    5. Repeat for additional capabilities

    Examples:

    * Web Search + Browser for research
    * Code Runner + Files for data analysis
    * Knowledge Retrieval for document Q\&A
  </Step>

  <Step title="Memory Configuration">
    Set memory preferences:

    * **Working Memory** - Tokens for current conversation (default: 8000)
    * **Summary Threshold** - When to create summary (default: 5 messages)

    Or use defaults. Click **Create**.
  </Step>
</Steps>

## System Prompt Tips

Good prompts are:

* **Specific** - Clear role and purpose
* **Constrained** - What agent should/shouldn't do
* **Examples** - Show desired behavior

```
You are a Python expert assistant.
- Write clean, Pythonic code
- Explain reasoning
- Suggest improvements
- Never use Python 2

When user asks to code:
1. Ask clarifying questions
2. Write code
3. Explain the approach
4. Suggest optimizations
```

## Agent Settings

After creation, refine in **Settings**:

| Tab               | Options                           |
| ----------------- | --------------------------------- |
| **General**       | Name, description, enable/disable |
| **System Prompt** | Update behavior instructions      |
| **Model**         | Change routing strategy           |
| **Capabilities**  | Add/remove tools                  |
| **Memory**        | Tune context size                 |
| **Channels**      | Deploy to platforms               |

## First Interaction

Test your agent immediately:

1. Open agent
2. Click **Chat** tab
3. Type a message
4. See agent response
5. Iterate on system prompt if needed

## Configuring Capabilities

### Add Capability

1. **Settings** > **Capabilities**
2. Click **Attach Capability**
3. Choose type and specific tool
4. Configure if needed
5. Click **Save**

### Remove Capability

1. **Settings** > **Capabilities**
2. Find capability
3. Click **Remove**
4. Confirm

## Deployment Options

After testing:

* **Direct Chat** - Use in Console
* **API** - Programmatic access
* **Channels** - Telegram, Slack, SMS
* **Automation** - Scheduled runs

See individual sections for detailed setup.

## Common First Agents

### Research Agent

* Web Search
* Browser
* Code Runner
* System: "Research topics thoroughly..."

### Customer Support

* Knowledge Retrieval
* HTTP Client (customer data)
* Files (ticket storage)
* System: "Help customers with their issues..."

### Coding Assistant

* Code Runner
* Files
* Web Search
* System: "Expert Python/JS programmer..."

## Troubleshooting

### Agent not responding

* Check if **Active** toggle is on
* Verify at least one capability attached
* Test connection by clicking **Test**

### Responses are slow

* Reduce working memory size
* Remove unused capabilities
* Check model selection
* Monitor usage dashboard

### Capability not working

* Verify capability is properly attached
* Check capability requirements
* Test capability individually
* Review error messages

## Next Steps

* [Write System Prompts](/docs/run/agents/system-prompts)
* [Attach Capabilities](/docs/run/agents/attaching-capabilities)
* [Deploy to Slack](/docs/run/channels/slack)
* [Memory Configuration](/docs/run/agents/memory-and-context)