# Agents

The **Agents** page — **Agents** in the sidebar — lists every agent in the workspace as a card and provides controls for searching, filtering, creating, and managing agents.

[![Sidebar with Coworkers and Agents under Workplace](../images/sidebar-agents-coworkers.png)](../images/sidebar-agents-coworkers.png)

## Last 24 Hours Usage Summary

At the top right of the Agents page, a usage summary shows aggregate activity across all agents in the workspace over the last 24 hours:

- **Total Cost** — combined spend across all agents, displayed as a compact number with a sparkline chart showing cost trend over the period.
- **Total Tokens** — total token consumption across all agents, with a bar sparkline showing consumption trend.

These figures update in real time as agents handle conversations. Use them to spot unexpected spikes before opening the full Analytics page.

## Agent Cards

Each agent is represented by a card. Cards display at a glance:

- **Agent type** — shown by the icon: Chat Agent, Flow Agent, or Voice Agent.
- **Name and description** — the agent’s display name and brief description.
- **Owner** — *Owned by you*, or the name of the colleague who shared it with you.
- **Status** — shown when the agent is in a non-default state: `Editing by you`, `Editing by <name>`, `Sent for approval`, `Approved and ready to publish`, or `Disabled`.
- **Tags** — any tags associated with the agent.
- **Cost and Tokens** — the agent’s total spend and token usage, followed by any other metrics and the current value of each KPI configured for the agent. Scroll the row sideways to see them all.
- **Version and last updated** — the currently active published version and when the agent was last changed.

Clicking a card opens the agent’s workflow editor (or settings page for Neo Code Agents).

[![Agent card — name, owner, description, tags, cost, tokens, and last updated](../images/cmd-home-card-v2.png)](../images/cmd-home-card-v2.png)

## Searching and Filtering

The search bar and filter controls at the top of the agent grid let you narrow the visible agents:

- **Search** — searches across agent name and description.
- **Type filter** — filter by agent type: Chat Agent, Flow Agent, Voice Agent. Only agent types enabled for the workspace appear as options.
- **Tags filter** — filter by one or more tags.
- **Ownership filter** — show all agents, only agents you own, or only agents shared with you.

All filters can be combined. The grid updates immediately as filters are applied.

## Tags

Tags are free-form labels you can attach to any agent for organisation and filtering. To manage tags on an agent card:

1. Click an existing tag on the card, or click **Add tags**.
2. Select tags from the existing list to add them.
3. Click **Manage tags** in the same popup to create new tags or delete existing ones.

Tags you remove from the agent are not deleted from the workspace — they remain available for other agents.

## Card Actions

Each agent card has a context menu (three-dot button) with actions available based on your role and permissions:

- **Settings** — opens the agent’s Settings tab directly.
- **Share** — opens the sharing dialog to grant other workspace members view or edit access to this agent. Available to the agent owner only.
- **Export** — downloads the agent configuration as a JSON file. You can choose whether to include credentials in the export.
- **Duplicate** — creates a copy of the agent in the same workspace.
- **Enable Agent / Disable Agent** — toggles the agent’s disabled state. Disabled agents appear faded on the card and cannot receive conversations.
- **Delete** — permanently removes the agent. Only available to the owner when the agent is not pending approval, approved, or being edited by another user.

> **Note**
>
> Agents in **Pending Approval** or **Approved** status cannot be deleted. Agents being edited by another user (status: Locked) can only be deleted by the user who has it locked.

## Creating and Importing Agents

Click the **+** button (**Create Agent**) at the top right of the Agents page to open the Create page. The button is visible to users with the `canCreateAgents` permission.

Under **Get started**, each agent type has its own card — **Flow**, **Chat**, **Voice**, and **Remote Agent**. You see only the types your role lets you create. On a card:

- **New** — creates an agent of that type from scratch.
- **Import** — imports an agent from a JSON file previously exported from this or another OneByZero Neo workspace. The import brings in the agent’s full configuration, settings, and associated metadata. Useful for migrating agents between environments, replicating configurations across workspaces, or promoting a tested agent from staging to production. Not available for **Remote Agent**.

New agents open in view-only mode. Click the pencil icon in the agent’s header to start editing.

## Best Practices

- Use tags consistently across your team — agree on a tagging convention (e.g. by department, use case, or environment) before agents proliferate.
- Monitor the Last 24 Hours sparklines regularly. A sudden spike in cost or tokens on a specific card is often the first visible signal of a misconfigured prompt or a loop.
- Disable rather than delete agents you no longer need but may want to restore — deletion is permanent and removes all conversation history.
- Use the Ownership filter to audit which agents are shared with you versus owned by you, especially when managing access for offboarding team members.