---
title: "Dashboard"
description: "Read the account panel, register agents from a harness and understand agent states."
image: "https://docs.agentregistrar.dev/og/docs-default.png"
---

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

# Dashboard

The dashboard is the landing page after onboarding. It shows the human account, the harness setup commands, the registered agents and a summary of confirmed service signups.

![Portal dashboard: a verified human with two agents and one recorded signup - the account panel shows the signed agreement version and each agent card shows its state and kill switch.](/screenshots/portal/portal-dashboard.png)

## Account panel

The top panel shows:

- The human name on the account.
- **ACTIVE** or **SUSPENDED** for the human account.
- **VERIFIED** when the email is verified.
- The signed Master Agreement version and the first characters of its digest.

The human status is separate from an agent status. A suspended human account is blocked from minting assertions and from completing the OAuth token exchange; a revoked agent is a single sub-identity that has been stopped.

## Connect a harness

The **Register an agent from your CLI** panel shows the command for OpenCode V2 and Claude CLI, with tabs for each. The command adds the broker MCP endpoint to the harness; the harness then completes OAuth and can call `identity_register`.

The portal does not have a form that creates an agent directly. Agents are created through the harness after the OAuth consent step. Harness installation guides are being added under [Agents & harnesses](/agents/); until they land, use the command shown in the panel.

## Registered agents

Each agent card shows:

- The agent name and `agt_...` id, with a copy control.
- The harness label.
- The dedicated `<agent>@<domain>` inbox address, with a copy control.
- The granted scopes.
- **ACTIVE** or **REVOKED**.
- For a revoked agent, the reason and timestamp.
- A link to the agent detail view and the kill switch.

Agents are ordered newest first. The grid updates when the database changes, so a newly registered agent can appear without a manual refresh.

### Agent states

| State | Meaning |
| :--- | :--- |
| **ACTIVE** | The agent can mint assertions, use its inbox and accept provider terms within its scopes |
| **REVOKED** | The agent is suspended by the human or by an abuse report; new assertions and inbox tool calls fail |

The current portal has no separate pending-verification state for an agent. A human who has not verified their email cannot reach the dashboard, so an agent is created only under a verified human.

## Signed-up services summary

The lower table is a short version of the service signup ledger: provider, account reference, agent used, the human email shared with the provider, the accepted ToS proof, the signup proof token and the vendor guide pointer. The full review surface is on [Ledger](/portal/ledger/).

The broker records identity and proof. It does not store provider API keys, resource inventory or usage data. Those stay in the provider's system.

## Empty state

A new account has no agents. The grid shows **no agents registered** and points back to the setup panel. Run the harness command, complete OAuth, and call `identity_register`; the first agent appears after registration.

## Next step

Open an agent card to reach [Agent detail](/portal/agent-detail/). Read [Kill switch and revocation](/portal/kill-switch-and-revocation/) before you need it. The delegation scopes and the account states behind these panels are defined under [Protocol](/protocol/); the MCP tools that register and operate an agent are listed under [Reference](/reference/).

Source: https://docs.agentregistrar.dev/portal/dashboard/index.mdx