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

# Intents

> Every MCP tool call requires an intent object that describes why the action is being taken. Intents form your agent's audit trail and will eventually give the agent guardrails.

An intent is a structured description of why an action is being taken. Every MCP tool call requires an `intent` object. Agentity records each intent, giving you a complete audit log of everything your agents have done.

## Intent fields

| Field     | Type   | Description                                                    |
| --------- | ------ | -------------------------------------------------------------- |
| `message` | string | A human-readable description of why the action is being taken. |
| `subject` | enum   | The resource the action targets.                               |
| `action`  | enum   | The type of operation being performed.                         |

### subject values

`EMAIL` `PHONE` `VIRTUAL_CARD` `VAULT` `IDENTITY`

### action values

`CREATE` `READ` `UPDATE` `DELETE`

## Example intent object

```json theme={null}
{
  "message": "Creating email inbox for agent to sign up for a service",
  "subject": "EMAIL",
  "action": "CREATE"
}
```

<Tip>
  Use descriptive intent messages — they become your audit log and help you understand what your agents were doing at any point in time.
</Tip>

## Reviewing intents

You can retrieve the full intent history for your account using the dashboard.

## Why intents are required

Intents serve two purposes:

1. **Auditability** — Because agents act autonomously, you need a reliable record of their actions. Requiring an intent on every call ensures no operation goes unlogged.
2. **Clarity** — A descriptive `message` makes it possible to understand agent behavior after the fact, even without replaying execution logs.