The Streamtime public MCP is in open beta. The below capabilities are current as of August 2026. Capabilities may change over time, so for the most up to date information, you can connect your AI assistant and ask for the latest capability.
Work with your Streamtime account through an AI assistant, using plain language. Create custom outputs interpreting your data from multiple tools, automate job creation or time logging, or simply query your data: this is a massive unlock.
🕰️ Est. Time: 8 minutes
🎯 Goal: Understand what you can, and can't, do in Streamtime through an AI assistant connected via MCP
🔑 Requirements: An active Streamtime login. The connector acts with your permissions, so you can only do through it what you can already do in the app.
The MCP connector lets you talk to Streamtime through an AI assistant like Claude. You ask a question or give an instruction in normal language, and the assistant carries it out in your account. This guide steps through what it can do today, grouped by the kind of work. Use with caution because the connector can create, change, and delete real data.
🔌 What the connector is
MCP (Model Context Protocol) is an open standard that lets AI assistants connect to outside tools. The Streamtime connector is our MCP server. Once you've connected it, your assistant can read from and write to your Streamtime account on your behalf.
You talk to it the way you'd talk to a colleague: "What jobs are in play for Acme?" or "Add a design item to the Acme rebrand." The assistant works out which Streamtime actions that maps to and runs them. It's two-way. It can look things up, and it can create and change records, including deleting them. Treat connecting it like handing someone access to your account.
The connection is set up in your AI assistant, not inside Streamtime. Instructions for connecting your account are here.
Prefer to watch first? See our webinar for a bit about the how and why, and a live demo of some key workflows.
🔍 Finding and reading your data
This is the safest and most common use: asking the assistant to look things up. Using search, it can:
Search your records: Jobs, Items, To Dos, Companies, Contacts and more, filtered by whatever you describe
Open a specific job and pull its full detail
Check availability for your team across a date range
List To Dos (planned and logged time) for a given period
View Priorities Boards and the cards on them
📊 Reporting on your data
For counts, sums and totals, utilisation, revenue, hours logged, quoted vs used, the connector runs reports, not searches.
This distinction matters more than it sounds. A search can look like it answered a "how many" or "how much" question when it's only shown you a page of results. A report is built to return the actual figure. If you want a number you can put in front of a client or a director, ask the assistant to report on it, and it'll pull the real total.
💼 Creating and Updating Jobs
The Job is the centre of everything in Streamtime, and the connector can build and maintain it end to end. It can:
Create a new job from scratch, with its items and phases, based on your instruction, or even a CSV upload
Duplicate an existing job: a common way to start a new one that's similar to past work
Update a Job's settings and status, add or remove Labels
Add, edit or remove Items, Phases and Milestones
Set dependencies so one Item follows another
In some cases, item ordering isn't controllable when the assistant creates Items. New Items may not land in the order you'd expect. If the sequence matters, reorder them yourself in the app afterwards.
📅 Setting Dates
Streamtime's reporting data is best when you've got dates on your Items, but we know sometimes things move quickly, and dates don't always get shifted and updated as a project changes. They might not even be set in the first place.
Quickly update item dates or set job dates based on To Dos logged, or move scheduled To Dos to match item dates through the MCP — giving you flexibility in both scheduling workflows and reporting ones.
⏱️ Logging, Moving, and Creating To Dos
Your AI assistant can create, update and remove To Dos, including in bulk through the MCP. So you can ask it to plan a week, adjust logged time, or log set of tasks in one go.
💰 Quoting, Invoicing, Expenses, and Purchase Orders
The connector handles the commercial side of a job:
Add, edit and remove Expenses: third-party costs like printing, travel or hosting, so they're captured for quoting and invoicing.
Build a Quote: from the job's plan, from an existing quote, or from scratch
Raise an Invoice: from the job's plan, from a quote, or manually
Create a Purchase Order to record what a supplier is charging you
Everything here is created as a draft. The assistant builds it; you review and send it from Streamtime. For quotes and invoices the assistant works in two steps: it checks the available options on the job first, then creates the draft, so if it asks a clarifying question before creating, that's expected.
📋 Priorities Boards (Kanban)
For task-level workflow, the connector works with Priorities boards. It can view your boards and their cards, create a new board, and update an existing one.
👥 Contacts and Companies
Create, update and delete companies (clients and suppliers), contacts (the people at those companies) and their addresses.
🔑 Permissions and access
The connector inherits your Streamtime permissions exactly: nothing more, nothing less. If you don't have View Job Pricing, for example, the assistant can't show you quote or invoice values, because you can't see them yourself. If an action fails with a permissions message, that's usually the cause, not a fault in the connector.
🚧 What it can't do, and where to be careful
Being upfront about the edges saves you trouble. Remember that connecting your data to an LLM is your responsibility. Once you connect an outside model to read your data, Streamtime isn't responsible for whether or not that data is private or used for training purposes.
Boundary | What it means for you |
When to use the MCP, and when to stay in Streamtime | Some things are easier via MCP; other things are better in app. Batch actions from the list views and CSV exports of reports are still going to be faster coming straight out of Streamtime. Some bulk actions are currently better through MCP (such as large job setups or bulk item date setting), though these may be present in the product itself in the future.
Streamtime MCP has the most power when it's combined with data from other sources — that's the big unlock.
We don't want you using all your precious tokens on things that are simple within Streamtime. Consider where best to aim the MCP. |
It only does what you can do | The connection acts as you. If you lack permission for something in Streamtime, you can't do it through the MCP either. |
Accounting integrations | Your accounting package connections aren't serviced through the MCP. For example, sending an invoice to Xero or QuickBooks needs to be done from within the app. |
It can misread a request | The assistant interprets plain language and can get it wrong. Review anything that matters, especially financials and deletions, before it's final. |
Deletes are permanent | Deleting a job, item or expense is a hard delete and cascades to related records. Just like in Streamtime, there's no trash or undo. Always confirm before you tell the assistant to delete anything. |
The MCP is in open beta. The available actions and their behaviour may change. Report any issues encountered to help@streamtime.net.
We're currently working on recommended skills to help supplement your workflows using AI and our MCP. Got a specific use case? Hit us up.