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

# Discover

> Discover AI agents, models, and services running in your estate.

Diamond evaluates an Agent you have registered. [Dome](/concepts/platform/dome) protects an Agent you have registered. Both start from the same assumption: that you know the Agent exists.

Discover removes that assumption. It finds the AI agents, models, and services running across your estate and classifies each one. You then hand the ones worth governing to the [Agent Registry](/owner-guide/register-agents/registering-agents), where the rest of the platform can act on them.

## How Discover Works

| Part | What it is |
| - | - |
| **[Target](/concepts/discovery/target)** | A place you want looked at: a fleet of laptops, a network, a cluster, a cloud account |
| **[Collector](/concepts/discovery/collector)** | The software that does the looking, installed at or near the Target |
| **[Resource](/concepts/discovery/resource)** | A single thing a Collector found, with the evidence for it |
| **[Classification](/concepts/discovery/classification)** | The verdict on whether a Resource is AI, and why |

## What Discover Finds

Coverage depends on which Collectors you install.

| Collector | Reaches |
| - | - |
| **Endpoint** | AI applications running on macOS, Linux, and Windows laptops and servers |
| **Network scanner** | Private-network services |
| **In-cluster** | Kubernetes services, read from the cluster's own catalogue |
| **Cloud API** | Managed AI such as Bedrock, SageMaker, and AgentCore |

## How Discover Fits In

The platform works as a sequence:

* **Discover** finds what exists and classifies it.
* **[Diamond](/concepts/platform/diamond)** evaluates what you registered and produces a [Trust Score](/concepts/trust-score/introduction).
* **[Dome](/concepts/platform/dome)** protects what you deployed and logs what it sees.
* **[Darwin](/concepts/platform/darwin)** learns from that telemetry and proposes improvements.

## Evidence

Every Resource carries the evidence behind its verdict: where the Collector saw it, what answered on which port, what the Classification rests on, and every scan that has observed it.

Discover also records what each scan could **not** reach.

## Continue With Discover

<CardGroup cols={2}>
  <Card title="Get Started" icon="radar" href="/owner-guide/discover/introduction">
    Create a Target, install your first Collector, and read your first inventory.
  </Card>

  <Card title="Discovery Components" icon="book-open-text" href="/concepts/discovery/introduction">
    The mental model: Target, Collector, Resource, and Classification.
  </Card>
</CardGroup>


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.