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

# Turn knowledge into agent-ready resources

> Publish governed organizational knowledge as MCP resources for AI agents.

Resources give agents the context they need to make informed decisions. Use them to publish approved knowledge from document repositories, internal pages, wikis, and other organizational sources as MCP-readable content.

```mermaid theme={null}
flowchart LR
    A[Knowledge source] --> B[Create MCP resource]
    B --> C[Process and structure content]
    C --> D[Set ownership and visibility]
    D --> E[Attach to virtual server]
    E --> F[Authorized agent discovers and reads resource]
```

## Before you begin

Identify the content owner, intended audience, and virtual server that will expose the resource. Confirm that the source contains only information that is appropriate for the visibility level you plan to assign.

Have the following ready:

* The URL of the approved source, if the content is hosted.
* A clear resource name and optional description.
* The resource type and tags that will help agents and administrators discover it.
* The owning team and the required visibility level.

## Create a resource from a URL

1. In the MCP Foundry dashboard, open **Resources**.
2. Select **Add Resource**.
3. Enter the URL for the document repository, knowledge page, wiki, or other approved source.
4. Provide a descriptive name and, optionally, a description.
5. Select the appropriate resource type and add tags.
6. Assign the resource to the responsible team and configure visibility.
7. Save the resource.

MCP Foundry processes and structures the submitted content so that MCP clients can consume it as agent-ready context.

## Publish inline knowledge

Use inline content when the information is not available at a hosted URL, or when a concise, governed knowledge asset is more appropriate than linking to a broader source.

1. Open **Resources** and select **Add Resource**.
2. Choose the inline-content option.
3. Paste the approved content.
4. Add the name, description, type, tags, ownership, and visibility settings.
5. Save the resource.

Keep inline resources focused on a single task or topic. Include the source owner and a review cadence in the description or team process so the content remains current.

## Choose visibility and ownership

Visibility and ownership help keep knowledge manageable at scale. Choose the narrowest setting that supports the intended use case.

| Setting  | Use when                                                                        |
| -------- | ------------------------------------------------------------------------------- |
| Private  | Only explicitly authorized agents or users should access the resource.          |
| Internal | The resource is appropriate for organization-wide use under your access policy. |
| Public   | The content is deliberately safe to expose to any compatible agent.             |

Assign an owning team so updates, review requests, and lifecycle decisions have a clear accountable group.

> **Security note:** Do not use resource visibility as a substitute for source-system authorization. Remove secrets, sensitive personal data, and restricted content before publishing, and limit virtual-server exposure to the audiences that need it.

## Attach the resource to a virtual server

Creating a resource adds it to the catalog; attaching it to a virtual server makes it available to agents connected to that endpoint.

1. Open **Virtual Servers**.
2. Select an existing virtual server or create a new one for the use case.
3. Attach the approved resource.
4. Save or publish the virtual server.
5. Test discovery and reading with an authorized MCP client or Playground.

```mermaid theme={null}
flowchart TD
    A[Resource catalog item] --> B{Approved for this use case?}
    B -->|No| C[Keep unavailable to this virtual server]
    B -->|Yes| D[Attach to virtual server]
    D --> E[Agent discovers resource]
    E --> F[Agent reads governed context]
```

## Maintain resource quality

* Use descriptive names and consistent tags so agents can identify relevant context.
* Review URL-backed resources when their source content or access policy changes.
* Update inline content when the underlying process, policy, or knowledge changes.
* Test that an agent can access the intended resource and cannot access resources outside the virtual server's composition.
* Monitor logs and feedback for stale, incomplete, or unexpectedly exposed content.

For the broader capability workflow, see [Manage MCP capabilities](./manage-mcp-capabilities) and [MCP Foundry concepts](../mcp-foundry-concepts).