> For the complete documentation index, see [llms.txt](https://support.hivo.co/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://support.hivo.co/integrations/hivo-mcp-connection.md).

# HIVO MCP Connection

This feature is in development and may be subject to change.

### 1. Overview about HIVO MCP Connection

HIVO supports the **Model Context Protocol (MCP)**, allowing compatible AI applications to connect with HIVO to search and query information within your Asset library.

Once connected, you can ask questions directly inside your AI application, such as:

* *"How many asset in summer campaign 2026”*
* *“List assets in the ‘Campaign 2026’ collection.”*
* *“Find images with the keyword ‘summer’.”*

The AI application sends requests to HIVO using the **HIVO\_ASK** tool. The HIVO Agent interprets the query, determines how to process it, and returns only the information that the HIVO user associated with the connection is authorized to access.

👉 This connection retrieves information through the tools provided by HIVO, eliminating the need to open the HIVO user interface or manually perform lookups step by step.&#x20;

* Each MCP key is assigned to a specific HIVO user.
* The AI Agent operates strictly under that user's permissions and cannot bypass their Folder permissions or any other HIVO access rights.
* *Example:* If the assigned user does not have permission to view Folder A, the Agent will also be unable to retrieve information from Folder A.

### 2. Set up a connection between HIVO and a third-party AI Agent

The setup has two parts.&#x20;

* First, a Client Admin or HIVO Admin configures governed access in HIVO.&#x20;
* Second, an Integration Partner or Developer establishes the technical connection from the MCP client.

### Part 1: Configure MCP Access in HIVO

Who can do this? Client Admins and HIVO Admins.

#### 🔹 Step 1: Open MCP/ AI Integration

* Open your HIVO and go to **Client Profile -> Settings.**
* Under System & Integration, select **MCP/ AI Integration.**

<figure><img src="https://2446612760-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F0sjg5GpIPr5HzBTJYqYe%2Fuploads%2FV29eNNBAhs2pbXNdkxzT%2FGroup%20811.png?alt=media&amp;token=1780755b-b1b1-4fea-a07c-072d237dc21a" alt=""><figcaption></figcaption></figure>

#### 🔹 Step 2: Generate and copy the MCP key

* Select the HIVO user as specified in the configuration interface.
  * **Assign to user - Required:** Select the HIVO user whose permissions the connected agent will inherit.
  * **Label:** Enter a recognisable name for the key, for example Agency Agent (Read).
  * **Scope**: Select the MCP access scope appropriate for the integration.
* Click "**Generate**" to create the MCP key.&#x20;

<figure><img src="https://2446612760-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F0sjg5GpIPr5HzBTJYqYe%2Fuploads%2FjN1p4joGIreeIUX8lUpx%2FGroup%20813.png?alt=media&amp;token=505f720a-5d6f-4791-92b5-cd2472587b0b" alt=""><figcaption></figcaption></figure>

* Then copy the full MCP key immediately.&#x20;

<figure><img src="https://2446612760-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F0sjg5GpIPr5HzBTJYqYe%2Fuploads%2FTHE86imJH8a8d7CrAyYA%2FGroup%20810.png?alt=media&amp;token=73bfdb83-d150-4c82-92f9-adb617ab595d" alt=""><figcaption></figcaption></figure>

**⚠️ Important Note:**&#x20;

* HIVO only shows the complete key once. After the screen is dismissed, only a prefix remains in Active Keys and that prefix cannot recover the secret
* Issue an MCP key and assign it to a user.&#x20;
* Tools always run as that user, with their HIVO permissions, folders, and library access. Share the key only with that person.&#x20;
* If it is lost, leaked or exposed in a shared screenshot, revoke it in Active Keys and issue a new key.

#### 🔹 Step 3: Copy the MCP Server URL

On the MCP/ AI Integration page, **copy the MCP Server URL**. This is the HIVO provider endpoint that the third-party MCP client will use.

<figure><img src="https://2446612760-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F0sjg5GpIPr5HzBTJYqYe%2Fuploads%2FyL93kY3VjRY22sCVT3py%2FGroup%20817.png?alt=media&amp;token=0b1142b7-6774-44ef-82a0-19ae1c565648" alt=""><figcaption></figcaption></figure>

#### 🔹 Step 4: Provide connection details to the integrating party

Provide the Integration Partner or Developer with only:

* **MCP Server URL**
* **MCP Key**

A separate HIVO User ID or Client ID is not required. HIVO binds the relevant user and workspace to the key when it is created.

### Part 2: Connect your MCP Client to HIVO

Who is this for? Integration Partners and Developers.

#### 🔹 Step 1: Add HIVO as an MCP server

Configure the third-party MCP client with the MCP Server URL and MCP Key provided by the Super Admin or HIVO admins. The exact configuration location depends on the third-party product.

For Example:&#x20;

<table data-header-hidden><thead><tr><th width="245.4842529296875"></th><th></th></tr></thead><tbody><tr><td><strong>Field</strong></td><td><strong>Value</strong></td></tr><tr><td>Connection Name</td><td>HIVO</td></tr><tr><td>Server URL</td><td><p>MCP_SERVER_URL (this is HIVO provider endpoint)</p><p></p><p>eg: https://mcp.hivo.com/agent/mcp</p></td></tr><tr><td>Authorization</td><td> Bearer &#x3C;hivo_mcp_YOUR_KEY></td></tr></tbody></table>

Replace `<HIVO_TOKEN>` with the full token generated in HIVO.

<figure><img src="https://2446612760-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F0sjg5GpIPr5HzBTJYqYe%2Fuploads%2FaJ0STNax1Fm7XmmWDHbp%2FGroup%20822.png?alt=media&amp;token=8cbc631a-8f8b-435a-ab89-5d7257c7fdfc" alt=""><figcaption></figcaption></figure>

#### 🔹 Step 2: Save and Enable the Connection

* Save the configuration, enable the connection, and check whether the application successfully loads the **HIVO\_ASK** tool.
* If your application does not support these configuration settings, refer to your application's MCP setup guide or contact your administrator.

### 3. How to use HIVO MCP Connection

The connected AI Agent inherits that user's HIVO access and cannot bypass their Folder access, rights or permissions. For example, if the assigned user cannot access Folder A, the connected AI Agent will not be able to retrieve information from Folder A either.

#### 🔹 Step 1: Test the connection

After setting up the MCP connection in your AI application, run a test prompt to verify that the system has successfully connected, authenticated the token, and recognized the authorized assets accessible by the assigned account.&#x20;

* Send a Test Prompt: ***“Can you connect to HIVO”.***
* If the connection is successful, the AI Agent will respond with a confirmation message containing:
  * **Connection & Authentication Status:** Confirms it is connected and authenticated.&#x20;
  * **Library Access Rights:** Confirms the number of assets the corresponding HIVO user is authorized to access.&#x20;
  * **Data State:** Confirms that no changes were made to the system.&#x20;
* Note: If the system returns an authentication error (*Authentication failure*) or fails to locate assets, double-check your Token/ API Key in the Header configuration, or ensure the assigned HIVO user has been granted appropriate access permissions to the relevant Folders/Libraries.

<figure><img src="https://2446612760-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F0sjg5GpIPr5HzBTJYqYe%2Fuploads%2F40G6XrFINxHrQvauNGkI%2Fimage%208.png?alt=media&amp;token=005a544b-2108-44cf-b8fb-9d538d5ab618" alt=""><figcaption></figcaption></figure>

#### 🔹 Step 2: Ask your Agent about your HIVO Library

Once the connection has been successfully tested, you can ask the Agent questions about supported Library information.

#### **1. Search Assets**

Use this when you want to find assets in the Library using filters similar to those available in the HIVO interface.

MCP can utilize existing Library filters, such as:

* Keyword
* File type
* Tags
* Upload date
* Uploader
* Folder
* Metadata
* Other attributes or filters supported by the Library Multiple conditions can be combined within a single query.

For Example:&#x20;

| Use Case                      | Example Prompt                                                           |
| ----------------------------- | ------------------------------------------------------------------------ |
| **Search by Keyword**         | “Find assets with the keyword ‘summer’.”                                 |
| **Search by File Type**       | “Find all PDF files in the library.”                                     |
| **Search by Tag**             | “Find images tagged ‘product’.”                                          |
| **Search by Date**            | “List videos uploaded between 01/10/2026 and 21/10/2026.”                |
| **Search by Uploader**        | “Find assets uploaded by John Smith.”                                    |
| **Combined Filters**          | “Find PNG images tagged ‘product’ uploaded in October 2026.”             |
| **Combined Metadata Filters** | “Find images where Campaign is ‘Summer 2026’ and Market is ‘Australia’.” |

<figure><img src="https://2446612760-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F0sjg5GpIPr5HzBTJYqYe%2Fuploads%2F22T56mMA14zjF7unHkot%2Fimage%2011.png?alt=media&amp;token=3b10d8e5-71f9-4b7c-a814-46e2f5a1ce74" alt=""><figcaption></figcaption></figure>

#### 2. Semantic Search

Used when a user does not know the exact filename, tag, or metadata, but understands the meaning or content of what they are looking for.

Instead of searching strictly by keyword, Semantic Search identifies assets whose content or semantics are relevant to the user's query.

For Example:&#x20;

| Use Case                | Example Prompt                                   |
| ----------------------- | ------------------------------------------------ |
| **Search by Meaning**   | “Find assets related to sustainable packaging.”  |
| **Find Similar Topics** | “Find images about outdoor summer activities.”   |
| **Concept Search**      | “Show me assets related to employee onboarding.” |

<figure><img src="https://2446612760-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F0sjg5GpIPr5HzBTJYqYe%2Fuploads%2FgP7p7ZBFwccpsrDpBBQY%2Fimage%2015.png?alt=media&amp;token=cecde79a-3f05-483a-8b1f-ada20af1d61a" alt=""><figcaption></figcaption></figure>

#### **3. Browse Collections**

Find collections by name and list the assets they contain. Results are limited to content the account is authorized to access.

For Example:&#x20;

| Use Case                     | Example Prompt                                   |
| ---------------------------- | ------------------------------------------------ |
| **Search Collections**       | “Find the collection named ‘Campaign 2026’.”     |
| **View Collection Contents** | “List assets in the ‘Campaign 2026’ collection.” |

<figure><img src="https://2446612760-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F0sjg5GpIPr5HzBTJYqYe%2Fuploads%2FeonRpjJPSq8j0I3aJJ5G%2FGroup%20820%20(1).png?alt=media&amp;token=bb815d47-8c4d-449f-a0c9-2396f786df94" alt=""><figcaption></figcaption></figure>

#### **4. Lookup Metadata**

Find assets using metadata fields or text extracted from documents. The relevant metadata or extracted text must already exist in the system.

For Example:&#x20;

| Use Case                    | Example Prompt                                               |
| --------------------------- | ------------------------------------------------------------ |
| **Lookup Metadata**         | “Find assets where the campaign field equals ‘Summer 2026’.” |
| **Search Within Documents** | “Find PDF documents mentioning the approval process.”        |

<figure><img src="https://2446612760-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F0sjg5GpIPr5HzBTJYqYe%2Fuploads%2FtXO1bx92hWQbNYRk9vAQ%2Fimage%2013.png?alt=media&amp;token=3735fed9-b030-4956-89f0-94519c939d57" alt=""><figcaption></figcaption></figure>

#### **5. Statistics & Analytics**

Count assets, analyze distributions such as file type or tag, and retrieve usage-based results such as most-viewed or most-downloaded assets.

For Example:&#x20;

| Use Case               | Example Prompt                                                 |
| ---------------------- | -------------------------------------------------------------- |
| **Count Assets**       | "How many asset in summer campaign 2026."                      |
| **Library Statistics** | “Provide a breakdown of asset counts by file type in a table.” |
| **Usage Analytics**    | “List the top 10 most-viewed videos.”                          |

<figure><img src="https://2446612760-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F0sjg5GpIPr5HzBTJYqYe%2Fuploads%2FwaAYI3diJFHsU6QI7NgG%2FGroup%20821%20(1).png?alt=media&amp;token=673a2634-e0ae-4e58-b5ba-3d4d43b54b2e" alt=""><figcaption></figcaption></figure>

Please contact the HIVO Admin if you encounter any difficulties during the installation process.


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation by asking a question.

Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:

```
GET https://support.hivo.co/integrations/hivo-mcp-connection.md?ask=<question>&goal=<user_goal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.
