A Toggl Track MCP built for Toggl's request quota, not against it
57 nodes across time entries, projects, clients, tasks, tags, workspaces, organizations, members and the Reports API. The design decision that shapes everything here is Toggl's hourly request quota, which starts at 30 requests per hour per user per organization on the Free plan. So the reporting nodes answer whole workspace questions in one request, the two bulk nodes edit many records in one request, and a quota node lets a workflow check what is left before it spends it.
What the Toggl Track MCP connector does
The Toggl Track connector for Mattering Automations is 57 nodes on Toggl's public API: 30 that read and 27 that act. Time entries cover listing, reading, creating, updating, stopping and bulk editing. Projects, clients, tasks and tags cover the structure that time is tracked against, including archiving and restoring clients. Workspaces, organizations, workspace members, workspace users and project members cover who can see and do what. Five nodes reach the Reports API for summary, detailed, weekly, project and per user reporting, plus a totals node. Six nodes are destructive and each one says so in its own description.
Every capability is a node on the Automations canvas, and every node speaks MCP, the open protocol for connecting AI assistants to tools. Claude, ChatGPT and any other MCP-capable assistant can use them, alongside Matter, the AI assistant built into the platform. So an assistant can ask what a team tracked last week and get a real answer from Toggl Track: Run Summary Report, and the same node can sit in a scheduled workflow that posts the answer every Monday with nobody present.
Toggl Track is two APIs sharing one hostname, and knowing which one a node sits on explains most of its behaviour. The v9 API is conventional REST with filters in the query string, and it is where every workspace, project, client, tag, task, member and time entry node lives. The Reports API v3 is POST only, even for its reads, takes its filters including the date window in a JSON body, and is scoped to exactly one workspace at a time, since every one of its paths names a single workspace. Both are covered here and the connector assembles either shape for you.
The request quota is the constraint worth understanding before you design anything. Toggl runs a sliding window quota on requests that name a workspace or an organization: 30 per hour per user per organization on Free, 240 on Starter, 600 on Premium, and custom on Enterprise. Requests about the token owner themselves, such as the account and quota reads, sit on a separate allowance that is 30 per hour on every plan, with Toggl stating no plan raises it. Spending the window returns HTTP 402, which is not retryable until the window moves. A second and independent mechanism, a leaky bucket per token and IP address, returns HTTP 429 and Toggl suggests treating one request per second as a safe pace.
That number is why this connector leans on the nodes that answer a lot in one call rather than a loop. Toggl Track: Run Summary Report totals a whole workspace for a window in one request, and Toggl Track: Run Project Users Summary Report answers who worked on what across every project in one request. Toggl Track: Bulk Edit Time Entries applies the same change to up to 100 time entries in a single request, which Toggl documents as the per request limit, so retagging fifty entries costs one call rather than fifty. Toggl Track: Get API Quota reads how many requests are left and how long until the window resets, so a workflow can branch instead of discovering the ceiling halfway through.
One security property is specific to this connector and worth stating plainly. Toggl's account, workspace list and single workspace responses each return a live api_token in the payload: the caller's own token, and each workspace's legacy one. That token is the credential, so anyone holding it holds the account. The connector strips it recursively out of the output of Toggl Track: Get My Account, Toggl Track: List My Workspaces and Toggl Track: Get Workspace, replacing the value with a redaction marker before the node returns anything. Node output routinely lands in a run log, a spreadsheet, a chat message or an AI step's context, and an assistant driving these tools over MCP forwards whatever it is handed. If you pipe raw Toggl responses around through some other route, that token goes with them.
Several features these nodes touch are paid plan features in Toggl, and the honest version is that some nodes simply will not work on a Free workspace. Toggl lists tasks, described as sub projects, from the Starter plan, so the six task nodes here fail on Free. Billable rates and project time estimates also start at Starter. Labor costs, fixed fee projects and project templates start at Premium, which affects the labor cost fields on the member and project nodes, the fixed fee field on project creation, the template option on Toggl Track: Create Project, and the labor cost figures the totals and project reports return. Toggl's own API reference says the client archive endpoint is for premium workspaces only, so Toggl Track: Archive Client and Toggl Track: Restore Client belong to the same group. Nothing here works around a plan boundary, and a node that hits one usually answers 402.
Two things this connector does not do. There are no trigger nodes: Toggl publishes a webhooks API, but nothing here listens for a push, so Toggl workflows run on a schedule or off a trigger from another connector. And Toggl's Insights and profitability API, saved reports, email and password authentication and session cookie authentication are all deliberately outside the catalog, the last two because a stored integration credential should never be an account password. Toggl does not publish an MCP server of its own, so there is no first party alternative to weigh this against.
One connection, three surfaces
- Visual automations. Drag Toggl Track nodes onto the canvas and wire them to triggers, conditions, and other apps.
- AI agents, yours or ours. These nodes speak MCP, the open protocol, so Claude, ChatGPT, and any MCP-capable assistant can use them, and Matter, the agent built into the platform, uses them natively.
- The whole platform. Combine with native tasks, projects, and other connectors, no middleware in between.
Every Toggl Track node, ready to drop on the canvas
57 nodes, generated from the connector itself so this list always matches what ships. Expand a category to browse.
Read and query nodes 30
Toggl Track: Get API Quota
How many requests are left in the current sliding window, per organization, and how many seconds until it resets. Toggl answers HTTP 402 once the window is spent, and a Free organization only allows 30 requests per hour per user. Read this before a fan-out and branch on data[].remaining rather than discovering the limit halfway through a loop. The quota response headers are not visible to a canvas node; this body is. Reading it costs one request.
Toggl Track: Get Client
Get one client by id, with its notes, archived state and external reference.
Toggl Track: Get My Account
Get the account behind the API token: user id, name, email, timezone, default workspace and week start. Use it to resolve who the token acts as before writing anything. This is the connector's auth probe. The live response also contains the account's own api_token, so treat this node's output as secret. User-scoped, so it costs one of 30 requests per hour on every plan.
Toggl Track: Get Organization
Get an organization's name, pricing plan, workspace allowance and user count. The pricing plan is the thing to read here: it determines the hourly request quota for every workspace underneath it, so a workflow can branch on it before deciding how much fan-out to attempt.
Toggl Track: Get Project
Get one project with its client, dates, colour, billable flag, rate, estimated hours and hours actually tracked.
Toggl Track: Get Project Statistics
Get the earliest and latest time entry on a project, which is the cheapest way to ask how long a project has actually been worked on. Toggl states this endpoint IGNORES running entries, so a timer started today will not move the latest value until it is stopped.
Toggl Track: Get Report Totals
The totals behind a detailed report, without the rows: seconds tracked, billable amount, labour cost, how many days had any tracking, and the rates that made up the money. Optionally returns a bucketed graph at day, week or month granularity. This is the cheapest node in the connector for answering how much time or money a window contains, because it is one request no matter how many entries are in it.
Toggl Track: Get Running Time Entry
Get the timer that is running right now for the token owner, or null if nothing is running. The entry's duration is negative while it runs; its start time is the useful field. Pair with Toggl Track: Stop Time Entry to close whatever is open.
Toggl Track: Get Task
Get one task with its estimate in seconds, the seconds actually tracked against it, whether it is still open, and which project and client it belongs to.
Toggl Track: Get Time Entry
Get one of the token owner's time entries by id. Scoped to the account behind the token: another user's entry is not readable here even for an admin, which is what the Reports API is for.
Toggl Track: Get Workspace
Get one workspace's settings: name, organization, plan flags, default currency and hourly rate, rounding rules and who may create projects or tags. Read this before writing rates or billable flags, since those are paid-plan features and are silently ignored on a Free workspace. The live response carries the workspace's legacy api_token, so treat this node's output as secret.
Toggl Track: Get Workspace Statistics
Get a workspace's admin list, member count and group count in one request. A cheap headcount check that does not spend a request per member the way listing them does.
Toggl Track: List Clients
List a workspace's clients, optionally filtered to active or archived ones or by a name fragment. Clients are what projects hang off, so this is usually the first read in a billing workflow.
Toggl Track: List My Organizations
List every organization the token's user belongs to, with the pricing plan name, workspace allowance and user count. The plan matters beyond billing: it sets the hourly request quota for every workspace under it, and gates billable rates, labor costs and saved reports.
Toggl Track: List My Projects
List every project the token owner can see, ACROSS all their workspaces, in one request. Useful when a workflow does not know which workspace a project lives in. This is user-scoped, so it draws on the flat 30 per hour user quota rather than a workspace's.
Toggl Track: List My Workspaces
List every workspace the token can reach, with each workspace's organization, plan flags, default currency and the token owner's role in it. This is the lister behind the workspace picker on every other node. The live response also carries each workspace's legacy api_token, so treat this node's output as secret.
Toggl Track: List Organization Users
List everyone in an organization, across its workspaces, including people who are only INVITED and have not joined yet. This is the wider directory: use Toggl Track: List Workspace Users when the question is who can track time in one workspace.
Toggl Track: List Project Members
List who is assigned to which projects in a workspace, with each assignment's manager flag, rate and labor cost. Rows carry ids only, no names: join them against Toggl Track: List Workspace Users on user_id if a human-readable answer is needed.
Toggl Track: List Project Tasks
List one project's tasks as a plain array, with each task's estimate, tracked seconds and assignee. This is the lister behind the task picker, chosen over the workspace-wide one because it is already scoped to the chosen project and its rows are a flat array.
Toggl Track: List Tags
List a workspace's tags. This is the lister behind the tag picker, and the way to turn tag names into the tag IDs that the reporting nodes filter on.
Toggl Track: List Time Entries
List the token owner's own time entries, optionally within a date window. This is the only entry lister outside the Reports API and it covers ONE user: use Toggl Track: Run Detailed Report for a whole team. A running entry comes back with a null stop and a NEGATIVE duration (minus the start time in UNIX seconds), which is a marker, not a length, so never add it to real durations.
Toggl Track: List Workspace Members
List the MEMBERSHIP rows for a workspace: role, billable rate, labor cost, group memberships, timezone and working hours, plus each row's invite state. Each row's id is the workspace-user id used when updating or removing a member, which is NOT the same number as the user id in the same row.
Toggl Track: List Workspace Projects
List a workspace's projects with their client, colour, billable flag, rate, estimated and actual hours. This is the lister behind the project picker on every project-scoped node. All filters are optional despite what Toggl's Swagger marks as required.
Toggl Track: List Workspace Tasks
List every task in a workspace, across projects, with paging. This is the ONE lister in this connector that does not return a bare array: its rows live under data.data[], alongside page, per_page and total_count. Tasks are a paid-plan feature in Toggl Track.
Toggl Track: List Workspace Users
List a workspace's users as flat rows of id, full name, email, role and active state. This is the lister behind every user picker in this connector. For the membership record itself, with rates, labor costs and group assignments, use Toggl Track: List Workspace Members.
Toggl Track: Run Detailed Report
Every time entry in a window, for the WHOLE workspace rather than one user, with project, client, task, tags, user and billable amount on each row. This is the node to reach for when building an invoice or exporting a timesheet. Each row carries a nested time_entries array holding the individual entries and their start, stop and seconds. Paging is driven by response HEADERS that a canvas node cannot see, so raise the page size to cover the window in one call and treat a result AT the page size as truncated.
Toggl Track: Run Project Summary Report
One project's totals for a window: seconds tracked, billable amount, labour cost, how many days saw any tracking, the rates involved, and a bucketed graph. Use it for a project health check without pulling every entry. Toggl returns 402 if the workspace's plan does not include the feature and 403 if the project is not accessible.
Toggl Track: Run Project Users Summary Report
Tracked and billable seconds per project per user for a window, as flat rows of project_id, user_id, tracked_seconds and billable_seconds. The cheapest way to answer who worked on what: one request for the whole workspace. Rows carry ids only, so join them against Toggl Track: List Workspace Projects and List Workspace Users for names.
Toggl Track: Run Summary Report
Total tracked time for a date window, grouped and sub-grouped the way the Toggl web report does it: by project, client, user, task or tag. One request answers a question that would otherwise take one call per entity, which is the difference between fitting inside a 30 per hour quota and not. The report covers ONE workspace; the Reports API does not do cross-workspace reports.
Toggl Track: Run Weekly Report
A seven-day grid of tracked time or billable amounts, grouped by user and project. Each row's seconds and billable_seconds are ARRAYS of one value per day of the week, in order, with no dates attached, so index 0 means the first day of the requested week. Good for a standing weekly status message; use the detailed report when individual entries are needed.
Action nodes 27
Toggl Track: Add Project Member
Assign a user to a project, optionally as its manager and with a per-project rate and labor cost. On a private project this is what makes the project visible to that person at all. Toggl's authorization layer is eventually consistent, so allow two to five seconds after creating a project before assigning people to it.
Toggl Track: Archive Client
Archive a client AND its projects, hiding them from pickers while keeping every time entry intact. This is the safe alternative to deleting a client, and it is reversible with Toggl Track: Restore Client. Toggl documents archiving as a premium-workspace feature, so a Free workspace will refuse it.
Toggl Track: Bulk Edit Time Entries
Apply the same JSON Patch operations to up to 100 time entries in ONE request, which matters a lot at 30 requests per hour: retagging fifty entries is one call, not fifty. Toggl executes this PARTIALLY on error, with no transaction and no rollback, so the response's success and failure arrays are the record of what actually happened.
Toggl Track: Bulk Update Projects
Apply the same JSON Patch operations to several projects in ONE request, which is how to archive or recolour a batch without spending a request each at 30 per hour. Like the bulk time-entry edit, this can apply partially, so read the success and failure arrays.
Toggl Track: Create Client
Create a client in a workspace. Toggl does not deduplicate by name, so running this twice produces two clients with the same name; check with Toggl Track: List Clients first if the workflow can retry.
Toggl Track: Create Project
Create a project in a workspace, optionally under a client and from a template. Billable, rate, currency, fixed fee and estimates are paid-plan features. Toggl's authorization layer is eventually consistent, so wait a couple of seconds before assigning users to a project you just created.
Toggl Track: Create Tag
Create a workspace tag. Usually unnecessary: creating or updating a time entry with tag NAMES creates any missing tags automatically. Use this when the tag must exist before any time is tracked, or when the workspace restricts tag creation to admins.
Toggl Track: Create Task
Create a task inside a project. Tasks are a paid-plan feature in Toggl Track, so this fails on a Free workspace. The estimate is in SECONDS, not hours.
Toggl Track: Create Time Entry
Log time, either finished or still running. For a finished entry send a start and either a stop or a positive duration in seconds; the two must agree if you send both. To START a running timer, send the start time and a duration of -1. Tags named here that do not exist yet are created automatically. Reversible only in the sense that the entry can be deleted again.
Toggl Track: Create Workspace
Create a workspace inside an existing organization. The organization's plan decides how many workspaces are allowed (a Free organization allows one). Toggl's authorization layer is eventually consistent: the vendor recommends waiting two to five seconds before using a newly created workspace, so put a delay before the next node rather than chaining straight into it.
Toggl Track: Delete Client
DESTRUCTIVE AND IRREVERSIBLE: permanently deletes a client. The client's projects lose their client association, and every report grouped or filtered by that client stops resolving it. There is no undo. Use Toggl Track: Archive Client instead unless the record genuinely has to disappear.
Toggl Track: Delete Project
DESTRUCTIVE AND IRREVERSIBLE: permanently deletes a project. Its blast radius is larger than the project row, because the time entries deletion mode decides what happens to every hour ever tracked against it: unassign detaches them and keeps them, delete DESTROYS THEM. This node defaults to unassign for that reason. There is no undo either way. To retire a project reversibly, update it with active set to false instead.
Toggl Track: Delete Tag
DESTRUCTIVE AND IRREVERSIBLE: permanently deletes a tag and strips it from every time entry that carried it. The entries survive, but the classification is gone from all of them at once and recreating a tag with the same name does not reattach it. Reports filtered by that tag will return nothing afterwards. Rename the tag instead if the goal is tidying.
Toggl Track: Delete Task
DESTRUCTIVE AND IRREVERSIBLE: permanently deletes a task. Time entries tracked against it lose their task association and any report broken down by task stops resolving it, so the shape of past reporting changes. There is no undo. To close a task out without losing that structure, update it with active set to false instead.
Toggl Track: Delete Time Entry
DESTRUCTIVE AND IRREVERSIBLE: permanently deletes a time entry. Toggl has no trash and no undo for time entries, so the recorded work is gone along with its description, tags and billable amount, and any report already run stops matching. Read the entry first if you might need to recreate it. To detach an entry from a project without losing it, update it with a blank project instead.
Toggl Track: Invite Users To Organization
Invite one or more people to an organization by email, optionally placing them into named workspaces, user groups and a project in the same call. Adding a seat can change what the organization is billed, so treat this as a spending action. The response returns an invite URL per person, which is what to send if email delivery is skipped.
Toggl Track: Remove Project Member
Remove someone's assignment to a project. Their time entries on that project are NOT touched; only the assignment goes, and adding them back restores access. On a private project this removes their visibility of it. Note this takes the ASSIGNMENT id, not the user id.
Toggl Track: Remove Workspace Member
DESTRUCTIVE: removes a person from the workspace. Their membership record goes with its role, rate, labor cost and group assignments, and they lose access immediately. Restoring it means a fresh invitation and reconfiguring everything by hand. Deactivate them instead with Toggl Track: Update Workspace Member if the goal is to stop them tracking time.
Toggl Track: Rename Tag
Rename a tag. The tag keeps its id, so every time entry carrying it follows the new name; any saved filter written against the OLD NAME rather than the id will stop matching.
Toggl Track: Restore Client
Un-archive a client, and either all of its projects or a named subset. The inverse of Toggl Track: Archive Client.
Toggl Track: Stop Time Entry
Stop a running timer, which fills in its stop time and turns its negative placeholder duration into a real length in seconds. Get the id from Toggl Track: Get Running Time Entry. Stopping an entry that is already stopped is an error, not a no-op.
Toggl Track: Update Client
Rename a client or change its notes and external reference. Archiving is a separate call: use Toggl Track: Archive Client rather than trying to set a flag here.
Toggl Track: Update Project
Change a project in place. Setting active to false ARCHIVES the project, which is the reversible way to retire one; deleting is not. When changing the rate, the rate change mode decides whether already-tracked time is repriced, so override-all can silently rewrite historical invoicing amounts.
Toggl Track: Update Task
Change a task: rename it, re-estimate it, reassign it, or mark it done by setting active to false. Marking a task done keeps every time entry attached to it, so this is the safe way to close work out.
Toggl Track: Update Time Entry
Change an existing time entry: its description, times, project, task, billability or tags. Toggl's own guidance is to send only the fields that changed. Tag edits are controlled by the tag action field, which decides whether the tags you send are added to or removed from the entry.
Toggl Track: Update Workspace
Change a workspace's settings. Send only the fields you are changing, per Toggl's own guidance. Do not send paid-plan fields (default currency, default hourly rate, rounding, rate change mode) on a Free workspace: the vendor asks clients not to, and the request can be rejected rather than ignored.
Toggl Track: Update Workspace Member
Change a member's role, rate, labor cost, working hours or billing visibility. Setting inactive to true DEACTIVATES the member, which is the reversible alternative to removing them: their history stays and they can be reactivated. Rates and labor costs are paid-plan features.
Connecting Toggl Track takes one credential
Your API token is on your own Toggl Track profile page at track.toggl.com/profile, at the bottom, under API Token. There is no separate developer account, no application to register and no review process: the Free plan has API access and the token is already there. The token acts as the user it belongs to, with that user's role in every workspace and organization they belong to, so the account you take the token from decides what every node can see and do. For unattended automation that argues for a Toggl user the team controls rather than an individual's personal login, because the connection follows the person out of the door otherwise. Toggl also accepts an email and password pair and a session cookie as authentication; neither is supported here on purpose, since an integration credential should be rotatable without changing an account password.
Decide which Toggl user holds the connection. Every node acts as that user, inherits their role in each workspace, and draws on that user's hourly request quota, which is counted per user per organization.
Sign in to Toggl Track as that user, open the profile page at track.toggl.com/profile and copy the API token from the bottom of the page. Treat it as a password: it grants full access to every workspace and organization the user belongs to, and it is the same value the connector redacts out of node output.
In Mattering Automations, open Connectors, click Add Connector and choose Toggl Track, then create a new credential and paste the token into the single API Token field. There is no second field to fill in. Toggl's scheme puts the constant string api_token in the password position, and the connector supplies it, because asking a person to remember a fixed protocol constant would be a form that lies about what the secret is.
Click Test Connection. Mattering calls the account endpoint, which is the cheapest authenticated call on the API. Note that a failed authentication in Toggl answers HTTP 403 rather than 401, so a 403 at this step means the token is wrong or revoked, not that a permission is missing.
Confirm the plan boundaries you care about before building. Tasks, billable rates and project estimates need Starter or above; labor costs, fixed fee projects and project templates need Premium; and Toggl documents client archiving as premium workspaces only. A node that crosses one of those lines answers 402, which reads like a quota error and is not one.
All 57 Toggl Track nodes appear in the canvas palette. Run Toggl Track: Get My Account first to confirm which user the token acts as, then Toggl Track: List My Workspaces to populate the workspace picker every other node uses. Both come back with the api_token field already redacted.
Common uses for the Toggl Track MCP
A weekly time report that costs one request, not one per person
Toggl Track: Run Summary Report totals a date window for a whole workspace and groups it the way the Toggl web report does, by project, client, user, task or tag. That is one request for a question that a naive integration answers with one call per entity, which is the difference between fitting inside a 30 requests per hour window and spending it before the report is finished. Toggl Track: Run Project Users Summary Report is the flat version, one row per project per user with tracked and billable seconds, and Toggl Track: Run Weekly Report gives the seven day grid. Its rows carry arrays of one value per day with no dates attached, so index zero means the first day of the requested week rather than a named weekday.
Invoice preparation that a person signs off
Toggl Track: Run Detailed Report returns every entry in a window for the whole workspace, with project, client, task, tags, user and billable amount on each row, which is the shape an invoice or a timesheet export actually needs. Toggl Track: Get Report Totals answers the same window without the rows when all you want is the number. Watch the paging on the detailed report: Toggl returns the next cursor in response headers that a canvas node cannot read, so raise the page size to cover the window in one call and treat a result sitting exactly at the page size as truncated rather than complete. Billable amounts depend on billable rates, which are a Starter feature, and labor cost figures need Premium.
Cleaning up mistracked time in one request
Toggl Track: Bulk Edit Time Entries applies the same JSON Patch operations to up to 100 time entries in a single request, which is Toggl's documented per request limit. Reassigning a week of entries to the right project or retagging them is one call rather than one per entry, and at 30 requests per hour that is the difference between possible and not. Read the response rather than the status code: Toggl states there is no transaction and no rollback, so a partially applied patch still answers 200 and the success and failure arrays are the only record of what happened. Toggl Track: Bulk Update Projects works the same way for archiving or recolouring a batch of projects.
Closing out timers somebody left running
Toggl Track: Get Running Time Entry returns whatever timer is open for the token owner, or nothing if none is. Toggl Track: Stop Time Entry closes it, filling in the stop time and turning the placeholder duration into a real length. Two details save an evening of debugging here. A running entry stores its duration as the negative of its start time in UNIX seconds, so it is a marker rather than a length and summing a window that contains one produces a number nine digits wrong. And stopping an entry that is already stopped is an error rather than a no operation, so branch on whether one is running instead of stopping unconditionally.
Opening the project structure from the systems around Toggl
A signed deal, an approved quote or a new job in another tool can create the structure to track against without anybody retyping it. Toggl Track: Create Client adds the client, and it is worth checking with Toggl Track: List Clients first because Toggl does not deduplicate by name and a retried workflow will happily create two. Toggl Track: Create Project opens the project under it, and Toggl Track: Add Project Member assigns the people, which on a private project is what makes it visible to them at all. Leave a gap between those last two: Toggl states its authorization layer is eventually consistent and recommends waiting two to five seconds, with up to three retries, before using a newly created project, so chaining straight through produces intermittent 403s that look like a permissions bug.
A directory sync that knows which id it is holding
Toggl Track: List Workspace Users returns flat rows of id, full name, email, role and active state, and is the lister behind every user picker here. Toggl Track: List Workspace Members returns the membership record instead, with role, rate, labor cost, group memberships and invite state. The distinction is not cosmetic and it is the sharpest trap in this connector: a membership row's own id and the user id inside the same row are different numbers, and the membership id is the one Toggl Track: Update Workspace Member and Toggl Track: Remove Workspace Member take. Passing the wrong one either 404s or reaches a different membership. When someone leaves, prefer setting them inactive with Toggl Track: Update Workspace Member over removing them, because removal takes the role, rate, labor cost and group assignments with it.
Example: A Monday morning time review that stays inside the quota
Built on the Automations canvas in a few minutes, or described in plain language to Matter and built for you.
Trigger: a weekly schedule fires. This connector has no trigger nodes, so Toggl workflows run on a schedule or off another connector's trigger rather than listening for a push from Toggl.
Toggl Track: Get API Quota reads how many requests are left in the window and how long until it resets. A condition on that number decides whether to run now or wait, which is cheaper than discovering HTTP 402 in the middle of the workflow.
Toggl Track: Run Summary Report totals last week for the workspace, grouped by project, in one request. Toggl Track: Run Project Users Summary Report adds tracked and billable seconds per project per user, also in one request.
Toggl Track: List Workspace Projects and Toggl Track: List Workspace Users resolve the ids in those rows into names, because the report rows carry ids only. Two more requests, and both results are reusable for the rest of the run.
An AI step turns the numbers into a short written summary and flags anything worth a human look, such as a project with tracked time well past its estimate. Nothing has been written back to Toggl at this point.
The summary is posted to a channel and a task is opened in Mattering Productivity for anything flagged. Corrections stay with a person, and when one is agreed, Toggl Track: Bulk Edit Time Entries applies it to every affected entry in one request rather than one per entry.
Toggl Track MCP questions, answered
What can this connector actually do in Toggl Track?
57 nodes: 30 read and 27 act. Reads cover the account behind the token, organizations, workspaces and workspace statistics, the request quota, time entries including the one running right now, projects and project statistics, project members, clients, tags, tasks at both project and workspace scope, workspace users and workspace members, organization users, and five Reports API nodes plus a report totals node. Writes cover creating and updating time entries, projects, clients, tags, tasks and workspaces, stopping a running timer, bulk editing time entries and projects, archiving and restoring clients, adding and removing project members, updating and removing workspace members, inviting people to an organization, and six deletes. Not included, and worth knowing before you plan: trigger nodes of any kind, so nothing here fires when something changes in Toggl; saved reports; the Insights and profitability API; and file or attachment handling, which Toggl's time tracking API does not have.
Does this work with Claude, ChatGPT, or other AI assistants?
Yes. Mattering exposes these nodes over MCP, an open protocol for connecting AI assistants to tools, so any MCP-capable assistant, including Claude and ChatGPT, can connect to your workspace and work with Toggl Track through them. Matter, the assistant built into the platform, uses the same nodes natively. The same nodes are also steps on the automation canvas, so a workflow an assistant helps you design can then run on a schedule with nobody watching. Given the quota, the sensible pattern with an assistant is the same as with a workflow: reach for the reporting nodes that answer a whole workspace question in one request rather than letting an assistant walk a list one record at a time.
Does Toggl publish its own MCP server?
No. Toggl does not ship a first party MCP server, and its integrations directory lists no Model Context Protocol or AI assistant connector. So this is not a page arguing against a vendor server; it is the practical way to put Toggl Track behind an MCP-capable assistant today. What the connector adds beyond a chat session is that the same 57 nodes run on a schedule with nobody present, chain with every other connector in one workflow, hold the API token once in a shared workspace credential store rather than in each person's local config, and carry destructive flags on the six nodes that delete something permanently.
What are Toggl's rate limits, and how should that change what I build?
Two independent mechanisms, and the first one is the binding constraint. Toggl runs a sliding window request quota on calls that name a workspace or organization: 30 per hour per user per organization on Free, 240 on Starter, 600 on Premium, and custom quotas on Enterprise. Calls about the token owner themselves, such as the account read and the quota read, sit on a separate allowance of 30 per hour that Toggl states is the same on every plan. Exceeding the quota returns HTTP 402, which is not retryable until the window moves. Separately, a leaky bucket limit per token and IP address returns HTTP 429, and Toggl suggests one request per second as a safe pace and asks clients to back off on a 429. Practically, 30 per hour is one request every two minutes, so a loop that makes one call per project or per person is not a viable design on Free. Use Toggl Track: Run Summary Report and its siblings, which answer whole workspace questions in one request, and Toggl Track: Bulk Edit Time Entries, which changes up to 100 entries in one. Toggl Track: Get API Quota lets a workflow read what is left and branch before it spends it.
Which nodes will not work on a Free Toggl workspace?
Toggl gates several of the things these nodes touch behind paid plans, and it is better to know which before a workflow fails. Toggl lists tasks, which it describes as sub projects, from the Starter plan, so all six task nodes need Starter or above. Billable rates and project time estimates also begin at Starter, which affects the billable amounts in the reports and the estimate fields on projects and tasks. Labor costs, fixed fee projects and project templates begin at Premium, which affects the labor cost fields on the member and project nodes, the fixed fee field on Toggl Track: Create Project, its template option, and the labor cost figures in Toggl Track: Get Report Totals. Toggl's API reference states the client archive endpoint is for premium workspaces only, so Toggl Track: Archive Client and Toggl Track: Restore Client sit in that group too. Toggl also asks clients not to send paid plan fields such as a default currency or an hourly rate on a workspace that does not include them, and warns that such a request can be rejected rather than quietly ignored, so a create node with a rate filled in can fail where the same node without it succeeds.
Is it safe to hand this to an AI assistant given that Toggl returns an API token in some responses?
This is the one thing to read carefully, and the connector already handles the sharp edge. Toggl's account response, its workspace list and its single workspace read each contain a live api_token: the caller's own token, and each workspace's legacy one. That value is the credential, so anyone holding it has the user's full access everywhere they belong. The connector recursively strips it out of Toggl Track: Get My Account, Toggl Track: List My Workspaces and Toggl Track: Get Workspace, replacing it with a redaction marker before the node returns, precisely because node output ends up in run logs, spreadsheets, chat messages and AI context, and an assistant forwards whatever it is handed. A warning in a description would not have been enough. The corollary applies outside Mattering: if you call those Toggl endpoints yourself and pipe the raw response somewhere, you are moving your own token with it. Beyond that, the ordinary discipline holds, which is to give an assistant the read nodes and put the six destructive ones behind a condition or an approval step.
Why do my time totals look absurd when a timer is running?
Because Toggl stores a running entry's duration as the negative of its start time in UNIX seconds rather than as a length. It is a marker meaning still running, and a window that happens to contain one running timer produces a sum that is nine digits wrong if it is added like a duration. Toggl Track: List Time Entries says so in its own description and deliberately declares no additive measure on that field for the same reason. Use the stop time or the start time to decide what is open, close it with Toggl Track: Stop Time Entry, and take totals from the reporting nodes, which count real durations. A related detail worth knowing: Toggl Track: Get Project Statistics ignores running entries entirely, so a timer started today does not move a project's latest entry date until it is stopped.
What happens when a workflow deletes something in Toggl?
It is gone. Toggl has no trash and no restore endpoint for the six destructive nodes here, and each one names a reversible alternative in its own description because that is usually what the workflow actually wanted. Deleting a time entry destroys the recorded work along with its description, tags and billable amount. Deleting a tag strips it from every entry that carried it at once, and recreating the name does not reattach it. Deleting a task or a client leaves the time entries but breaks the association, so reports grouped by them stop resolving. Removing a workspace member takes the role, rate, labor cost and group assignments with it. The largest blast radius is Toggl Track: Delete Project, because Toggl's time entries deletion mode decides what happens to every hour ever tracked against the project: one value detaches them and keeps them, the other destroys them. This node defaults to detaching for that reason, and the reversible way to retire a project is to update it as inactive instead, which archives it.
Can I report across several workspaces at once?
Not through the Reports API, and this is Toggl's boundary rather than the connector's. Every path on Toggl's Reports API names a single workspace, so each of the five report nodes takes one workspace and answers for that one only. Two v9 nodes do read across boundaries: Toggl Track: List My Workspaces returns every workspace the token can reach, and Toggl Track: List My Projects returns every project the token owner can see across all of them in one request, which is useful when a workflow does not know where a project lives. Building a multi workspace view means running a report node per workspace and combining the results, and the quota is the thing to plan around when you do, since each of those requests is workspace scoped and draws on the per organization allowance.
Put Toggl Track to work inside your workspace
Connect once, then use every node in visual automations and AI agent workflows across the Mattering platform.
All product names, logos, and brands are property of their respective owners and are used for identification purposes only. Use of these names, logos, and brands does not imply endorsement or affiliation. Toggl Track is a product of its respective owner; Mattering is an independent platform.