PD MCP Hub for WHMCS: Connect AI Assistants and Business Tools

Your AI assistant can write an answer. The more useful question for a hosting business is whether it can work with the services, domains, DNS records and support information behind that answer.

PD MCP Hub is a free WHMCS MCP server that connects compatible AI assistants to WHMCS, tools contributed by your installed PD modules and approved third-party MCP servers. Developed by Projektif Digital Ltd, it brings those capabilities into one hub while letting you decide which assistant can use which tools.

That creates opportunities beyond asking AI to summarise text. A staff member can investigate an account with its service context, maintain a knowledge base, inspect DNS across contributing providers or use approved external tools in the same conversation. Teams using PD AI Assistant can also bring the hub’s capabilities into their in-panel co-pilot.

This guide explains the product’s practical value, the workflows it supports and how its different parts fit together.

What is a WHMCS MCP server, and why would a hosting business use one?

The Model Context Protocol gives AI applications a standard way to connect to tools and information. In a WHMCS integration, the assistant can request the tools made available by your installation instead of relying solely on information someone pastes into a chat.

For a hosting operator, that changes the starting point of a conversation. “Help me understand this service” can lead to an actual service lookup. “Read the DNS records for this zone” can use an available DNS tool. “Find our existing guidance before drafting a new article” can begin with the knowledge base you already maintain.

Your assistant still interprets the request and produces the answer. PD MCP Hub supplies the connection to the relevant tools and enforces the authority you give that connection. The two work together: the AI handles the conversation, and the hub connects that conversation to supported business operations.

The benefit is especially clear when a task crosses several screens or systems. You can give the assistant relevant capabilities together, instead of treating every system as a separate island.

Three layers of capability in one hub

PD MCP Hub brings together three sources of tools. Understanding the distinction helps you see what it can add to your own installation.

This is a broader proposition than exposing a list of standard WHMCS API calls. It connects business context, module-specific capabilities and external MCP tools, with separate controls for the assistants using them.

The actual Catalogue screen identifies each tool’s source, toolset, risk and confirmation requirement. This example installation includes PD module contributions and attached Cloudflare tools. Your catalogue reflects the modules and servers you use; the totals in the screenshot are not a fixed product limit.

Bring account context into the conversation

A customer’s question often makes little sense in isolation. The relevant answer may depend on the service they bought, its status, the domain associated with it and the previous support conversation.

PD MCP Hub includes Everything about one client, a composite tool that returns client details, services, domains, invoices, tickets and orders together. Separate tools provide detailed context for an individual service or domain. Other tools read invoices, orders and ticket conversations.

Consider this illustrative staff request: “Find this client and summarise the services and domains relevant to their enquiry.” The assistant can locate the account and gather its context through the tools available to its connection. An operator can then work from an account-specific explanation rather than a generic AI answer about hosting.

Composite tools also make the integration easier to understand. A client overview is a business-level task, not a sequence of unrelated requests that the operator must design themselves. The hub also provides tool discovery for other supported work, so a deliberately focused connection can remain useful as the conversation develops.

For hosting teams, the opportunity is better preparation for decisions: assemble the relevant facts, identify the service being discussed and give the person handling the enquiry a clearer starting point.

Turn knowledge-base maintenance into an AI-assisted workflow

A hosting knowledge base grows through everyday work: new product guidance, clearer explanations, updated configuration instructions and answers to questions that keep coming back.

PD MCP Hub provides dedicated tools to list and search KB content, read articles, manage categories, and create or edit articles. The tools support article categories, tags, visibility and translations. Announcements and staff notes have their own operations too.

An authorised content assistant could search for existing guidance, read the relevant articles and prepare an update in the correct category. That gives the assistant the context needed to improve the material you already have, rather than generating another disconnected document for someone to copy manually.

A useful brief might be: “Find the existing nameserver guidance and identify what needs updating for our new onboarding instructions.” The operator supplies the approved source information and reviews the proposed content. The connection can then perform the supported editing operation when its permissions allow it.

For a hosting business, this links content work to the support system where the guidance will be used. It is also a distinct capability worth evaluating when looking for WHMCS AI knowledge-base integration, rather than only conversational access to client records.

The KB article creation guide explains the task-level workflow. Editing requires the appropriate connection ceiling and the assigned administrator’s Manage Knowledgebase permission.

Use PD module capabilities beyond the standard API

Your PD modules can extend what the assistant can do. The released MCP Hub documentation describes DNS, SMS and security contributions, making the module ecosystem part of the integration’s value.

DNS tools across contributing providers

DNS tools cover listing zones, reading records and adding, changing or deleting records. Where several installed modules provide DNS capability, the hub shares one set of tools with provider choices rather than presenting a separate near-identical tool family for every provider.

For a hosting team managing several DNS services, that supports a useful read workflow: list zones across the contributing providers, then inspect records for the zone in question. Results identify providers that did not respond, so an incomplete result does not silently look like a complete inventory.

A DNS change names the intended provider explicitly. The assistant uses the supported tool under the connection’s permissions; deletion requires a person’s approval. The practical opportunity is to bring DNS investigation and permitted record work into the same assistant-led conversation as the hosting account context.

SMS and operational communication

Contributing SMS tools let an assistant read recent messages, see which gateways are configured and request a text message. Sending a message requires human approval because it reaches a real phone and may incur a cost.

That creates a route from preparing a communication to an actual supported messaging operation. The operator decides whether the message is appropriate, while the hub provides the tool and review step. It is more useful than an assistant that can only write text someone must transfer to another system.

PD Guard security information

PD Guard contributes tools for an environment check, failing checks and recent sign-ins. An assistant with suitable access can bring that information into a staff investigation or regular review.

These are specific supported contributions, not a promise that every installed PD module exposes every screen or function. The Catalogue shows what your own installation offers, and a connection’s permissions determine what that assistant can use.

Attach third-party MCP servers to connect the wider business

Hosting companies also operate other systems alongside WHMCS. PD MCP Hub also accepts remote MCP servers that support Streamable HTTP over HTTPS.

You can choose a listed server or supply its address, provide its required authorisation, and review the tools it offers. Once you approve those definitions and allow a connection to reach the server, its tools become part of the assistant’s available toolkit.

This real screen shows a Cloudflare server attached to the hub. In the Catalogue screenshot, that example server contributes documentation and migration-guide tools. It does not imply that attaching it grants every Cloudflare operation. The capabilities come from the particular server and tools you review.

That distinction is useful when designing a workflow. A hosting assistant might need an external documentation tool alongside WHMCS service information. Another team might attach a different server for a different business system. You select the actual tools needed for the job instead of assuming an integration name tells you everything it can do.

Attached tools have a name prefix identifying their server. Your installation connects directly to that server using the authorisation you supply; PD Modules does not relay the connection. Approved tools are also checked for changed definitions, with a changed server held for review before its tools return to use.

The hub exposes these tools together, while the assistant selects the relevant calls during a conversation. It does not automatically manufacture new composite operations spanning different vendors.

For the setup steps, see how to attach an MCP server and allow a connection to use its tools.

Use the AI assistant that fits your team

PD MCP Hub connects the hosting tools to an MCP-capable assistant. Your team can choose a client such as Claude Desktop, Claude Code, Codex, Cursor or VS Code, or work with the in-panel PD AI Assistant bridge. The core value stays the same: access to permitted WHMCS business tools, contributions from PD modules and approved attached servers.

For a support lead using Claude, a useful request could be: “Read the confirmed client’s service information and the matching helpdesk conversation, then summarise the questions we should answer.” A content team using Codex could ask: “Find the existing DNS setup article, compare it with these approved instructions and prepare an update.” These illustrative workflows depend on the tools and permissions you configure, rather than on an exclusive relationship with one AI client.

The released connection guidance covers Claude Desktop, Claude Code, Cursor and VS Code. Codex can also connect to WHMCS through PD MCP Hub: its support for remote MCP over Streamable HTTP with bearer-token authentication matches the hub’s connection method. Use the endpoint and dedicated token shown by your hub, following the selected client’s configuration. OpenAI’s MCP documentation explains that client’s current setup.

Create a separate connection for each assistant or role, with access suited to its work. The chosen client handles the conversation and tool calls; MCP Hub supplies the business capabilities and enforces the connection’s limits. That lets you evaluate different assistants while keeping control of the tools they can use.

AI providers, MCP clients and local LLMs: choose how you want to work

There are three different parts to the setup. The model or AI provider produces responses. The MCP-capable client or host manages the conversation and tool calls. PD MCP Hub exposes the permitted WHMCS, PD module and attached-server tools.

That separation gives a hosting business options. You can use an external assistant, an application you build around a provider’s API, a local-model host or the PD AI Assistant bridge. The chosen client and model need to support the tool workflow you want to run.

Use OpenAI models through the Responses API

For a custom application, the OpenAI Responses API supports remote MCP tools. Supply the hub’s server URL and connection authorisation, and use a model supporting that tool. Application approvals and the hub’s own limits are separate controls. This is an API route; it does not imply that every ChatGPT interface accepts the same configuration.

Use a local LLM with WHMCS through an MCP-capable host

A downloaded model needs a host that can call tools. LM Studio supports remote MCP servers, including authenticated connections, and can make their tools available to local models. That makes it a documented compatibility route to evaluate with PD MCP Hub.

For an Ollama-based setup, Open WebUI connects local and cloud models and separately supports MCP over Streamable HTTP. The host connects to the hub; Ollama supplies the selected model. Use the hub’s required bearer authentication and a model capable of reliable tool calls.

These are documentation-based integration routes, not claims that every local model or version has been tested with PD MCP Hub. Local inference also does not make the whole workflow offline: WHMCS and any attached remote systems still exchange the data needed for the calls you permit.

Extend the in-panel assistant with your chosen provider

With PD AI Assistant installed, enabling its MCP Hub bridge gives the in-panel co-pilot the hub’s capabilities under the same limits. Its own tools retain priority.

PD AI Assistant’s released provider guidance covers your own Amazon Bedrock, OpenAI or Google Gemini connection. That provider choice belongs to PD AI Assistant; MCP Hub supplies the tools. It is a different route from using a local model through an external MCP host.

Explore PD AI Assistant for WHMCS if your team prefers assistance inside the billing panel.

Real-world scenarios: connect the context across CRM, billing and support

The following are hypothetical hosting-business scenarios, not customer case studies. Each assumes the external system has a compatible remote MCP server offering the named read tools, that you approve those tools and that the assistant’s connection can reach them. The assistant coordinates the lookups; the hub does not automatically synchronise databases or create cross-system workflows for you.

A CRM account review alongside the actual WHMCS services

Imagine an agency keeps account-management notes in a CRM but bills hosting through WHMCS. Before speaking to a business customer, an account manager wants to know what was discussed previously, what the customer currently uses and whether an opportunity recorded in the CRM is still relevant.

An illustrative request is: “Using this confirmed CRM account reference and WHMCS client ID, summarise the latest account notes, current hosting services and domains. Identify questions for our review call.”

With suitable tools available, the assistant can read the CRM account through the attached server and fetch the matching WHMCS context through PD MCP Hub. It can bring the findings into one summary while identifying their source. The operator supplies or verifies the account relationship rather than relying on a similar company name as proof of a match.

The useful outcome is a prepared account conversation: CRM history beside the services actually being delivered. You are not limited to asking questions about one platform at a time.

An external helpdesk conversation with WHMCS service context

A hosting company might use a separate helpdesk for customer conversations while WHMCS manages services, domains and invoices. A staff member investigating a ticket needs both the conversation and the account it concerns.

An illustrative request is: “Read this helpdesk ticket and the confirmed matching WHMCS service. Summarise the reported problem, the current service information and the questions we should answer before replying.”

If the helpdesk’s approved MCP tools can retrieve the ticket, the assistant can read that conversation and the relevant WHMCS tools. A contributing DNS module may also supply record-reading tools when DNS is relevant. The operator gets a combined view instead of manually moving every piece of context between applications.

A read-and-draft workflow is already useful. Sending a response through the external helpdesk is a separate capability that depends on that server’s tools and permissions; it is not implied by being able to read a ticket.

Compare an external billing record with a WHMCS invoice

A finance team may keep accounting records in another system while WHMCS raises hosting invoices. When a record needs investigation, the team wants to compare the original invoice with the external record using known invoice and account references.

An illustrative request is: “Read WHMCS invoice X and accounting record Y. Compare their references, amounts and recorded status, and list any differences for our team to review.”

If the attached server exposes the relevant accounting reads, the assistant can retrieve both records and organise a comparison. That supports an investigation without presenting AI as the system of record or assuming it can resolve a mismatch by changing financial data.

The same pattern applies to other systems with suitable MCP tools: gather the permitted facts, preserve the relationship between the records and help a person make a better-informed decision. This is where WHMCS CRM integration through MCP and cross-platform AI assistance become concrete business use cases.

Practical control makes broader capability usable

Different jobs need different authority. An assistant researching account context and one maintaining approved KB content do not need an identical connection.

PD MCP Hub gives each assistant its own token, ceiling, tool areas and assigned administrator. The administrator’s WHMCS role is enforced alongside those settings. Optional client scope, expiry, address restrictions and calls-per-minute limits provide further control.

You can require approval for every data-changing operation. Operations with mandatory human steps retain them, and the approval description is built from the actual request rather than the assistant’s explanation of its intentions. The audit trail records calls and refusals so an operator can understand what happened.

These controls support useful delegation: give a connection the capability to perform its job, keep consequential decisions with the right person, and retain a record of the work. They are part of the module’s operating model, rather than the whole reason to use it.

Get started through PD Modules

PD MCP Hub is free, with its features and controls included. It is installed through PD Modules, the control panel used to manage the module family.

Install PD Modules first, connect the account that holds your licence, and add MCP Hub from its Store. Use the hub’s own Catalogue to inspect the available tools and its Connect tab for your installation’s connection details. Configure the intended assistant’s connection, then enable the endpoint when ready.

Choose a workflow whose result you can assess: relevant account context, a correctly updated KB article or the expected DNS records. That gives you a practical way to judge the integration’s value while making use of the modules and tools you actually operate.

Explore PD MCP Hub for WHMCS to see the product, or browse its tool catalogue guide to understand how your installation’s capabilities are organised.

Frequently asked questions

Is PD MCP Hub a free WHMCS MCP server?

Yes. PD MCP Hub is free, with its connection controls, approvals, audit trail and attached-server functionality included. You supply a compatible AI client and any external services you choose to connect.

What makes it more than a WHMCS API connector?

Alongside supported WHMCS operations, it provides composite account-context tools, dedicated content tools, contributions from compatible installed PD modules and approved tools from attached MCP servers. Those additions expand the work available to an assistant beyond a standard API connection alone.

Can PD MCP Hub work with a local LLM or Ollama?

A local model needs an MCP-capable host. LM Studio supports remote MCP servers, and Open WebUI can connect an Ollama-backed model to an authenticated Streamable HTTP MCP server. Check transport, authentication, reachability and tool-calling ability for your chosen setup.

Can an assistant combine CRM, billing and support information?

It can use WHMCS and approved attached MCP tools within one task when those servers expose the required operations. Supply or verify matching account references. This enables cross-system analysis; it is not automatic data synchronisation or a ready-made integration with every CRM or helpdesk.

Can I connect Claude or Cursor to WHMCS?

Yes. The released MCP Hub documentation includes connection guidance for Claude Desktop, Claude Code, Cursor and VS Code. Use the configuration shown on your own Connect tab and a separate connection with appropriate access for each assistant.

Is there an MCP server for connecting Codex to WHMCS?

Yes. PD MCP Hub connects WHMCS to Codex through an authenticated remote MCP connection. Use the hub endpoint and a dedicated connection token; the connection settings and assigned administrator constrain the available operations.

Can an AI assistant manage DNS through PD MCP Hub?

Contributing DNS modules provide tools to list zones, read records and request supported changes. The available providers depend on your installation. Writes identify the intended provider, and deletion requires human approval.

Can it create and update WHMCS knowledge-base articles?

Yes. Dedicated tools cover KB categories, article search, reading, creation and updates, including supported visibility and translation operations. Editing needs the appropriate connection ceiling and the assigned administrator’s matching permission.

Can I attach other MCP servers?

Yes, remote servers using Streamable HTTP over HTTPS. Review and approve their tool definitions, then grant the intended connection access. The attachment feature exposes tools; it does not accept locally executed servers, prompts or resources.

Does PD MCP Hub work with PD AI Assistant?

Yes. Enable the assistant bridge in MCP Hub Settings to make the hub’s capabilities available to the in-panel co-pilot under the same limits. PD AI Assistant’s own tools keep priority.

Does PD Modules host the AI model or relay our external connections?

No. Your installation connects to your chosen assistant and attached servers directly. The AI client and external services receive the information needed for the operations you permit; the hub is not a promise that data never leaves your server.