Frontu has a built-in Model Context Protocol (MCP) server. MCP is an open standard that lets AI assistants such as Claude and ChatGPT connect securely to business systems. Once connected, you can ask your assistant in plain language to look up tasks, customers, technicians and inventory, and to make everyday changes such as creating a task, assigning a technician, logging work or rescheduling a job.
No coding and no API keys are required. You add Frontu to your assistant, sign in with your normal Frontu login, click Allow, and you are connected.
How it works
The MCP server is part of your Frontu account. When you connect an assistant, it acts as you: it sees exactly the data you can see in Frontu and can only do what your user permissions allow, within your own company account.
Read tools let the assistant search and open records.
Write tools let it create, update or append data. There are no delete tools.
Every change goes through the same validation, status rules and notifications as the web app, so your team sees it as if you had made it in Frontu directly.
What you can ask
Phrase requests naturally; the assistant picks the right tool. Some examples:
Tasks (work orders)
"Show me all unfinished tasks for customer Acme this month."
"Open task FS-1042 and show me its notes and logged work."
"Create a maintenance task for the boiler at the Riga warehouse next Tuesday."
"Assign this job to technician Jonas and move it to Friday morning."
"Add a note: customer requested a callback." / "Log 2 hours of pump replacement."
"Start this task", "Pause it, waiting for parts", "Mark it done", "Approve it", "Reopen it".
Technicians
"List all technicians." / "Where was Jonas last seen?"
"What is Jonas working on right now, and what is left in his queue today?"
Customers, sites and equipment
"Find the customer with VAT code LT123456789."
"List all sites for customer Acme and the equipment installed at each."
"Update the phone number for this customer."
Manager analytics
"How are we doing today?" returns queue stats, SLA risk, alerts, technician availability and workload, the same numbers as the live dashboard.
"Which tasks breached SLA?" / "Which tasks have been stopped for over two hours?"
"How many tasks per technician this month?" / "Total work price by customer last quarter."
Calendar, evaluations and stock
"What is scheduled for Jonas next week?" / "Schedule a team meeting on Thursday at 10:00."
"Show the completed evaluation form for task FS-1042."
"Is part X in stock at the Riga warehouse?"
Before you start
You need two things:
Your Frontu server address. It is the address you use to log in, for example
https://yourcompany.frontu.com. The MCP address is that address followed by/client/api/mcp:https://YOUR-SUBDOMAIN.frontu.com/client/api/mcpYour Frontu username and password. You will be asked to sign in during setup. Nothing else is needed: the assistant registers itself with Frontu automatically.
Whichever assistant you use, the last step is always the same. A Frontu window opens, you sign in if you are not already, and you see a screen titled Authorize [assistant name] listing the requested access. Click Allow to connect, or Deny to cancel. You can re-run the connection at any time.
Connect ChatGPT
Custom MCP connectors in ChatGPT are available on Plus, Pro, Business, Enterprise and Edu plans and are set up from the web app at chatgpt.com. On Business, Enterprise and Edu workspaces an administrator must first allow connectors and developer mode in the workspace settings.
1. Turn on developer mode
Open chatgpt.com and click your profile picture in the bottom-left corner, then Settings.
Open Apps & Connectors (called Connectors in some versions).
Scroll to Advanced settings and switch on Developer mode. This unlocks the option to add your own MCP servers.
2. Add Frontu as a connector
Back in Apps & Connectors, click Create in the top-right corner.
Fill in the form:
Name: Frontu
Description (optional): Frontu field service management, tasks, customers and technicians.
MCP Server URL:
https://YOUR-SUBDOMAIN.frontu.com/client/api/mcpAuthentication: OAuth. Leave the Client ID and Client Secret fields empty. Frontu issues these automatically.
Tick I trust this application and click Create.
ChatGPT opens a Frontu window. Sign in with your Frontu credentials if asked, then click Allow on the Authorize ChatGPT screen.
The window closes and the Frontu connector shows as connected, with the list of available tools.
3. Use Frontu in a chat
Start a new chat and click the + button next to the message box.
Choose More, then Developer mode, and tick Frontu. In some versions the connector appears directly in the + menu under Connectors.
Ask your question, for example "Which of our tasks are unassigned for tomorrow?"
ChatGPT runs read tools on its own. For write tools (create, update, assign, start, complete and so on) it shows you what it is about to do and asks you to confirm before it changes anything in Frontu.
Disconnect: Settings, then Apps & Connectors, click Frontu and choose Disconnect or Delete.
Connect Claude
Custom connectors are available on Claude Pro, Max, Team and Enterprise plans. On Team and Enterprise plans an organization owner may need to enable custom connectors first.
Open claude.ai, click your initials in the bottom-left corner, then Settings.
Open Connectors and click Add custom connector.
Enter Name: Frontu, and Remote MCP server URL:
https://YOUR-SUBDOMAIN.frontu.com/client/api/mcp. Leave the advanced OAuth fields empty and click Add.Click Connect next to the new Frontu connector. Sign in to Frontu if asked and click Allow on the Authorize Claude screen.
In a new chat, open the Search and tools menu (the sliders icon under the message box) and make sure Frontu is switched on. Then ask your question.
The connector is tied to your Claude account, so it also works in the Claude desktop app without extra setup.
Claude Code (for developers): run the command below, then type /mcp, select frontu and choose Authenticate to complete the sign-in in your browser.
claude mcp add --transport http frontu https://YOUR-SUBDOMAIN.frontu.com/client/api/mcp
Other MCP clients
Any client that supports remote MCP servers over Streamable HTTP with OAuth sign-in (for example Cursor, Windsurf or custom agents) can connect. Add a remote server with the URL above and choose OAuth as the authentication method. Frontu supports automatic client registration, so you do not need to enter a client ID or secret. Complete the sign-in and Allow step when your browser opens.
Once connected, a good first question is "What task types do we have in Frontu?"
Full tool reference
These are the exact capabilities exposed to the assistant. Searches return compact summaries in pages of up to 100 records; use the get tools for full detail of one record.
Tasks
Tool | Type | What it does |
search_tasks | Read | Search tasks by status, customer, site, type, billing, planned dates, office, region, tags or code |
get_task | Read | Full detail of one task: works, notes, materials, times |
get_task_status_history | Read | Status-change history |
search_task_requests | Read | Browse intake / backlog task requests |
create_task | Write | Create a new task |
update_task | Write | Update fields on an existing task |
assign_task | Write | Assign or re-assign a task to a technician |
reschedule_task | Write | Set a task's planned start (and optional end) time |
create_task_request | Write | Create an intake request, for example a follow-up visit |
add_task_note | Write | Append a note to a task |
log_work | Write | Log a performed work record against a task |
start_task / stop_task | Write | Start work, or pause it with an optional reason |
complete_task / approve_task | Write | Mark a task done (optionally with a note), then approve it |
abort_task / reopen_task | Write | Cancel a task with an optional reason, or reopen a closed one |
Manager analytics
get_operations_snapshot | Read | Live-dashboard snapshot: queue, alerts, SLA risk, technician status, workload |
get_problem_tasks | Read | Task list for one alert category, such as SLA overdue or stopped over 2 hours |
get_technician_overview | Read | One technician's current task, next task, today's summary and queue |
aggregate_tasks | Read | Counts, price and duration sums grouped by status, technician, type, customer, site, office, day or month |
Technicians and users
search_users / get_user | Read | Find users and open one user's profile |
get_user_task_times | Read | Time logged by users against tasks |
get_user_locations | Read | Last-known and recent locations, for one user or all |
Customers, sites and equipment
search_clients / get_client | Read | Find customers by name, email, phone, company or VAT code; open one customer |
update_client | Write | Update fields on a customer |
search_workspaces / get_workspace | Read | Find sites (objects) by title, serial number, code, type or customer; open one site with its equipment |
search_equipment | Read | Find equipment by title, site or model |
search_equipment_actions | Read | Equipment install, service and repair history |
Calendar events, evaluations and materials
search_events / get_event | Read | Find calendar events by user, date range or type; open one event |
create_event / update_event | Write | Create or update a calendar event |
list_evaluation_forms | Read | Evaluation / checklist templates |
get_task_evaluation | Read | Evaluation recorded for a task |
check_part_availability | Read | Stock of a material at a specific warehouse or site |
search_material_stock | Read | Inventory levels across warehouses |
Reference data
list_task_types, list_offices, list_task_stop_reasons, list_task_templates | Read | Lookups the assistant uses when filtering or creating records |
Tips for good results
Be specific. Names, task codes, dates and site names help the assistant pick the right records. You never need to know internal IDs; the assistant resolves them.
Search first, then drill in. Ask for a list, then ask to "open" a specific record for full detail.
Review changes. For create, update and assign actions the assistant tells you what it changed. Double-check important changes just as you would in the web app.
Security and limits
The assistant signs in as you. It cannot see or change anything you could not see or change yourself in Frontu, and it never sees another company's data.
You never share your password with the assistant. Sign-in happens on Frontu's own page, and the assistant only receives a limited access token that Frontu can revoke.
The assistant can read data and make create, update and append changes. It cannot delete records.
Actions your user is not allowed to perform are refused, and the assistant tells you so.
Results are returned in pages of up to 100 records.
Frontu administrators can review active API access, including access granted to AI assistants, under Settings, then API, and revoke it there. You can also remove the connector from your assistant at any time.
Frequently asked questions
I got "Authorize ChatGPT" (or Claude) but the page asks me to log in first. That is expected. You sign in with your normal Frontu account, then the Authorize screen appears. If you use single sign-on, complete it as usual.
Can several people in my company connect? Yes. Each person connects their own assistant with their own Frontu login, and each connection carries that person's permissions.
The connection stopped working. Open the connector settings in your assistant and connect again. Access is also removed if a Frontu administrator revokes it or your Frontu user is deactivated.
Does this replace the REST API? No. The REST API remains the right choice for system-to-system integrations. MCP is designed for AI assistants working on behalf of a person. Developers building their own agents can also authenticate the MCP endpoint with a Client API token if their client does not support OAuth.
Is my data used to train AI models? Frontu does not send your data anywhere. Your assistant connects to your Frontu account directly. Check your AI provider's own data policy for how it handles conversation content.
Where can I find the full API documentation? At https://developers.frontu.com/.