How to Use MCP Tools for Ticket Actions in BoldDesk

The BoldDesk MCP Server enables AI Agents to perform supported BoldDesk actions through natural language prompts. Once the MCP Server is configured and connected to an AI Agent, users can interact with BoldDesk conversationally to retrieve ticket details, list and create tickets, update ticket information, add messages, search contacts, and create approval requests.

This article explains how to use a connected BoldDesk MCP Server with an AI Agent. It covers prerequisites, prompt-writing best practices, supported scenarios, and troubleshooting steps to help you get the most out of the integration.

Prerequisites

Before using the AI Agent with the BoldDesk MCP Server, verify the following:

The AI Agent is created, configured, and deployed to the required channel. Explore How to Create an AI 2.0 Agent in BoldDesk.
The required MCP Server is already created and available in the MCP Server list. Learn How to Create and Configure an MCP Server in BoldDesk.
The MCP Server uses the correct BoldDesk MCP endpoint, such as https://your-subdomain.bolddesk.com/mcp.
The MCP Server uses the required x-api-key custom header. Learn to Generate and Manage API Keys in BoldDesk.
The API key owner has permission to perform the required actions, such as reading tickets, creating tickets, updating tickets, or adding messages.

Do not expose the API key in screenshots, articles, public documentation, or shared environments. Use masked values when documenting or demonstrating the MCP Server setup.

Configure and Use MCP Tools with the AI Agent

After creating the BoldDesk MCP Server, link it to the required AI Agent and select the MCP tools the AI Agent can use. Once the tools are linked, users can enter natural language prompts to perform supported BoldDesk actions, such as retrieving ticket details, creating tickets, updating ticket information, or adding ticket messages.

Follow these steps to link the MCP Server, select the required tools, and use effective prompts with the AI Agent.

Step 1. Link the MCP Server and Select Tools for the AI Agent

Before users can ask the AI Agent to perform BoldDesk actions through the MCP Server, the MCP Server must be linked to the AI Agent, and the required MCP tools must be selected. This configuration must be completed by an agent whose AI Agent Builder Access is set to Full Access. Learn more on How to Configure AI Agent Builder Access Permission.

The AI Agent can use only the MCP tools that are selected while linking the MCP Server. To link the MCP Server to the AI Agent:

Select the AI module from the left navigation menu.
Under AI Agent & Tools, select AI Agents.
On the AI Agents page, select the required AI Agent.
In Agent Builder, select AI Tools from the left menu. The MCP Tools tab opens by default.
Click Link MCP server.
In the Link MCP Servers panel, find the required MCP Server.
Click Select tools for the MCP Server that must be linked to the AI Agent.
Select the required tools that the AI Agent can use.
Click Link.
Publish the AI Agent if the updated configuration is still in draft.

The AI Agent can only use the MCP tools selected when the MCP Server is linked to the AI Agent. If a required tool is not selected, the AI Agent cannot perform actions that depend on that tool, even if the MCP Server is linked. Explore List of Available MCP Tools in BoldDesk.
When an AI Agent uses a direct MCP action to add a public message, the update is posted under the ticket requester/customer’s name, as MCP tool calls execute using the requester’s permissions. If the response should appear as an agent reply, use the Ticket Response Agent with the appropriate orchestration trigger instead of a direct MCP public message action. The Ticket Response Agent still retrieves data using the requester’s permissions but publishes the final response as an agent reply.

Step 2. Use MCP Tools Through a Deployed AI Agent Channel

After the MCP Server is linked and the required tools are selected, users can ask the AI Agent to perform supported BoldDesk actions through the channels where the AI Agent is deployed, such as Live Chat.

The AI Agent can use the linked MCP Server to perform supported actions when the required tools are selected. To get accurate results, users should write prompts that clearly state the action to perform and include the required details, such as ticket ID, requester email, subject, description, group, status, or message content.

Start with prompts that retrieve or confirm information before asking the AI Agent to create or update data. This helps reduce mistakes and ensures the AI Agent is working with the correct ticket, contact, group, or status.

Example:

The following GIF shows an example of using natural language prompts with an AI Agent to retrieve ticket information and perform a supported MCP tool action in BoldDesk.

Users can use MCP tools only through channels where the AI Agent is deployed. If the AI Agent is not deployed to a channel, users cannot use that channel to ask the AI Agent to perform MCP-supported BoldDesk actions.

Common MCP Actions and Prompt Examples

Use the following examples to understand how to phrase prompts clearly.

Action	Example Prompt
Retrieve a list of tickets based on available filters.	List all open tickets assigned to me.
Retrieve details of a specific ticket.	Show me the details of ticket #[ticket ID].
Create a new support ticket.	Create a ticket for `[email protected]` with subject “Password reset issue” and description “The customer cannot reset their password.”
Update ticket fields such as priority, assignee, group, category, or custom fields.	Update ticket #[ticket ID] and set the priority to High.
Change a ticket’s status.	Update ticket #[ticket ID] to Resolved.
Add a public reply, public note, or internal note to a ticket.	Add an internal note to ticket #[ticket ID] saying the issue has been escalated to engineering.
Retrieve the conversation history of a ticket.	Show the latest messages for ticket #[ticket ID].
Retrieve the current user context, accessible brands, and available forms.	Show my available BoldDesk brands.
Retrieve available ticket fields and allowed values.	Show the available ticket fields.
Retrieve valid ticket status transitions.	What statuses are available for ticket #[ticket ID]?
Find contacts by email, name, phone, or ID.	Search for the contact with email `[email protected]`.
Find agents for assignment or follow-up.	Search for agent Maria.
Find groups or teams for ticket routing.	Search for the Support group.
Find contact groups by name or identifier.	Search for the contact group Acme Inc.
Create an approval request for a ticket.	Create an approval request for ticket #[ticket ID] to approve a refund exception.

What to Include in Your Prompt

The AI Agent works best when your prompt includes enough information for the requested action. Use the following checklist before submitting a prompt.

Action	Details to Include
Create ticket	Requester email, subject, description, and priority if needed.
Update ticket	Ticket ID, field to update, and new value.
Update ticket status	Ticket ID and target status.
Add ticket message	Ticket ID, message type, and message content.
Retrieve ticket details	Ticket ID.
List tickets	Ticket condition, such as open, assigned to me, pending, high priority, or unresolved.
Search contact	Contact email, name, phone, or ID.
Search agent	Agent name, email, or identifier.
Search group	Group or team name.
Create approval request	Ticket ID and approval reason.

Best Practices

Follow these best practices when using a linked BoldDesk MCP Server with an AI Agent:

Start with read-only prompts before using actions that create or modify data.
Use exact ticket IDs when updating, replying to, retrieving, or closing tickets.
Use requester email when creating tickets or searching contacts.
Confirm contacts, agents, or groups before using them in ticket updates.
Ask the AI Agent to confirm ticket details before high-impact changes.
Avoid vague prompts such as “update this ticket” unless the ticket context is clear.
Use human confirmation for replies, status changes, ticket closure, and approval requests.
Review the AI Agent response after every action.
Test each tool type separately before using the AI Agent in live customer conversations.

The AI Agent may ask for missing details before using a tool.

Troubleshooting

The AI Agent does not use the expected MCP tool

Use a more direct prompt and include required details such as ticket ID, requester email, subject, description, status, group, or agent.

The AI Agent asks for more details

The prompt may be missing required information. Provide the missing ticket ID, requester email, subject, description, or status.

The AI Agent may update the wrong ticket

Use the exact ticket ID and ask the AI Agent to confirm the ticket details before making changes.

No result is returned

Start with a read-only prompt, such as Show me the details of ticket #[ticket ID], to confirm that the AI Agent can retrieve data.

Frequently Asked Questions

Can I use natural language to create tickets through the AI Agent?
Yes. If the linked MCP Server includes the ticket creation tool, you can ask the AI Agent to create tickets using natural language.
What details should I provide when creating a ticket?
Provide the requester email, subject, and description. You can also include priority, category, group, or assignee if needed.
Can the AI Agent update existing tickets?
Yes. If the required update tool is available, you can ask the AI Agent to update supported ticket fields or status.
Can the AI Agent reply to ticket conversations?
Yes. If the message tool is available, provide the ticket ID and the reply or note content.
Why does the AI Agent ask for missing details?
The AI Agent asks for missing details when a tool requires specific values before it can complete the request.

Was this article useful?

Dislike

Comments (0)