Guides

Send the Weekly Traffic Report With a Manus Cue Connector for Analytics Reporting

Taras Shynkarenko
Taras Shynkarenko
•Updated: •8 min read
A Manus Cue connector for analytics reporting assembling a weekly Flowsery traffic report from read-only callsA Manus Cue connector for analytics reporting assembling a weekly Flowsery traffic report from read-only calls

TL;DR, Quick Answer

8 min read

A weekly Flowsery report is six read calls: list_websites once, get_overview for both periods, get_timeseries for the shape, then get_channels and get_breakdown for the cause. Manus documents scheduled and triggered Automations; Cue documents no scheduling as of 29 September 2026. Name the dates, the timezone and every recipient in the prompt, because a Cue agent has its own email address.

The weekly traffic report is the safest job for a Manus Cue connector for analytics reporting, because every Flowsery call it needs is a read. Cue is the personal-agent app Manus launched on 28 September 2026, and it runs on the same infrastructure as Manus. The harder question is not which calls to make. It is which product runs the report every Monday, and as of 29 September 2026 the answer is Manus, not Cue.

The short version: the report is six reads and no writes. Manus Automations can start a task on a schedule or on an event. Cue has no documented scheduler yet, so run the report on demand in Cue, or build it as a Manus Automation and test that the Flowsery connector answers there.

Setup, the token role and the MCP form sit in the Manus Cue connector for web analytics guide. This post starts at the report.

Which Flowsery calls does a weekly report need?

A weekly Flowsery report needs six read calls, and the order matters more than the count. list_websites returns the IDs every later call carries. get_overview runs once per period. get_timeseries with interval=day shows whether the change was one bad Tuesday or a trend. get_channels and get_breakdown name the source that moved.

The six reads, in order
Setup
  • list_websites (GET /websites), first run only
The comparison
  • get_overview, last week
  • get_overview, the week before
  • get_timeseries, interval=day
The cause
  • get_channels
  • get_breakdown, dimension=channel or campaign
Each MCP tool maps to one GET endpoint in Flowsery's public spec. None of them changes data.

GET /overview returns visitors, sessions, bounce rate, average session duration, revenue and conversion rate as a single row. GET /breakdown groups visitors by any of 25 dimensions, from channel and campaign to utm_source and entry_page. That one call replaces a dozen dashboard clicks.

A person checking a weekly planner at a desk, matching the question of what runs the report every Monday.

Can Cue run the report on a schedule?

Cue has no documented scheduler as of 29 September 2026, so scheduling belongs to Manus. Manus 2.0 turned Scheduled Tasks into Automations, which start work at a set time or when something happens in a connected service. Manus lists three kinds: Schedule, Triggered task and Advanced automation, which is in beta.

ProductWhat it documents for recurring workStatus on 29 September 2026
ManusSchedule, Triggered task, Advanced automation (Beta)Documented
ManusCloud Computer, "a permanent home for an automation", sold separatelyDocumented
CueScheduling, triggers, memoryNot documented
CueWhether a Manus connector appears inside CueNot documented

Manus has not documented whether an Automation can call a custom MCP server either. Test it before you rely on it: create a Schedule automation that calls list_websites and returns the names, and let it run once. Manus's Automations docs tell you to "review the generated timing, conditions, actions, and accounts before activating it", which is the only review step Manus documents for recurring work. Read the actions list for any Flowsery tool that writes.

Until that test passes, run the report in Cue by hand on Monday morning. One prompt, six reads, and a person awake to read the answer.

Test Manus before you trust it with Monday
1
Create. Make a Schedule automation that calls list_websites and returns the names.
2
Review. Read the generated timing, conditions, actions and accounts before activating, and look for any Flowsery tool that writes.
3
Run once. Let it fire one time and check the result.
4
Decide. If the names come back, move the report to Manus. If not, run it by hand in Cue on Monday morning.
Manus has not documented whether an Automation can call a custom MCP server, so the test comes first.

What should the report prompt say?

The report prompt names both date ranges, the timezone and the recipients in plain words. Flowsery buckets startAt and endAt against the website's own timezone unless the call names one, and it defaults startAt to 30 days ago when the agent leaves it out. Two silent defaults, both able to shift the report by a day.

Report on flowsery.com for 21 to 27 September 2026 against
14 to 20 September 2026, timezone Europe/Berlin.
 
Call list_websites on the first run only and reuse the websiteId.
Then get_overview once per period and get_timeseries with interval=day.
If sessions or conversions moved more than 10%, call get_breakdown with
dimension=channel for both periods and name the channel.
 
Numbers first, then one paragraph. No recommendations.
Send the report only to me. Do not email anyone else.
Do not call any tool that records, deletes or updates data.

"No recommendations" earns its line. An agent asked for analysis produces analysis, and this one has traffic numbers and nothing else: no release log, no roadmap, no idea you paused a campaign. The Grok bot template for analytics reporting carries the same instruction for the same reason.

Who receives the report when the agent has its own email?

A Cue agent sends the report wherever you tell it to, because Manus gives each agent "its own email, phone number, wallet, and computer, so it can send messages". That is useful. The report lands in an inbox from an address that belongs to the agent, not to you. It is also the part of Cue with the least documentation: Runtimewire reported on launch day that Manus did not explain how users "restrict what agents can do with their email or phone identity".

Name every recipient in the prompt. Keep the report aggregate. A weekly report never needs GET /visitors/{visitorId}, which returns name and email for identified visitors, so leave that call out of the prompt entirely.

Group chats give you a cleaner split. Agents in a Cue group chat share a goal and "hand work to each other", while you "set the direction and make the final call". Put two agents in one chat:

  1. The numbers agent holds the Flowsery connector and returns the six reads as raw figures.
  2. The writer agent has no connector. It turns the figures into one paragraph and sends it to the names you gave.

The writer never touches your analytics, and the numbers agent never touches an outbox. If a figure in the email looks wrong, you know which agent to question.

Flowsery
Flowsery

Start FREE Trial

Real-time dashboard

Goal tracking

Cookie-free tracking

Can a trigger start the report instead of a clock?

A Manus Triggered task starts work "when something happens in a connected service", and Manus lists Gmail, Outlook Mail, Notion, Google Calendar, GitHub, Shopify, advertising platforms, RSS and Webhook as sources. Manus's own examples include "a change in ad performance". Flowsery's public spec documents no outbound webhook, so Flowsery itself does not fire the trigger.

Two triggers fit a Flowsery report. A Google Calendar event for your weekly review starts the report an hour before the meeting, so the numbers match the week everyone is about to discuss. An ad-platform trigger fires when spend or performance shifts, and the task pulls get_breakdown with dimension=campaign to check whether the traffic moved with it. Both run on Manus, not Cue, and both need the connector test from the previous section to pass first.

A closed padlock on a wooden door, standing for a read-only token that cannot change data.

What can a reporting connector change?

A reporting connector built on a Contributor token changes nothing. Flowsery's spec lists 28 operations as of 29 September 2026: 23 read and 5 write, and the writes cover goals, payments and issue status. A Contributor token gets 403 permission_denied on all five, because only Admin and Editor hold the flowsery.write permission.

That matters more for a scheduled job than for a chat. Nobody watches a Monday run. If a prompt goes wrong at 7am, the worst a Contributor token does is read the wrong week. The Meta Muse connector for analytics reporting leaned on Meta's approval prompts for the same protection. Cue documents none, so the token role does the job instead.

What breaks on a recurring report?

Four failures show up on a recurring report, and none of them is the first authentication.

The timezone. startAt and endAt without timezone bucket against the website's timezone set in Flowsery. The report still arrives and still looks right, and it covers a different seven days than your calendar.

The missing selector. A workspace token spans every website in the workspace, so each call needs websiteId or domain. Flowsery's API reference lists that miss under 400 Bad Request. Tell the agent the selector is required before it tries to fix the error by rewriting the query.

The dead token. Flowsery answers 401 with token_issuer_lost_access when the member who created the key has left the workspace. A scheduled report fails silently from that day on. Create the token from an account that is going to stay.

The retry loop. Flowsery allows 600 requests a minute per key and returns 429 with Retry-After past that. Six reads a week never gets near the limit. An agent retrying one failing call in a loop does, so tell it to stop after one retry and report the error.

For the report you read yourself, the website analytics reporting view in Flowsery shows the same numbers the agent pulls.

Frequently Asked Questions

Can Cue send the weekly report on its own?

Cue documents no scheduler as of 29 September 2026. Manus documents Schedule and Triggered task automations, and Cue runs on the same infrastructure. Build the report as a Manus automation, test that it reaches the Flowsery connector, and run it by hand in Cue until the test passes.

Can a reporting connector change my Flowsery data?

A reporting connector on a Contributor token changes nothing. Every call in the weekly report is a GET, and Flowsery returns 403 permission_denied to a Contributor token on the five write calls. Create the token under Contributor in the Create Token form at flowsery.com/api-tokens.

Why does the second period come back as zeros?

Flowsery defaults startAt to 30 days ago and endAt to now when either is missing, and a window with no traffic returns zeros instead of an error. Check the dates the agent sent. Give both periods as literal dates in the prompt, never as "the week before".

Can the report compare more than two periods?

Each period is one get_overview or get_timeseries call with its own startAt and endAt, so four quarters cost four calls. Give every call the same timezone, or the periods start and end at different midnights.

Flowsery
Flowsery

Start FREE Trial

Real-time dashboard

Goal tracking

Cookie-free tracking

Should I use Cue or the Grok bot template for reporting?

The Grok bot template for analytics reporting installs from a share link with the instructions already written. Cue connects through a Manus custom MCP connector and can email the report from the agent's own address. Both read the same Flowsery data, and every reporting call either one makes is a read.

Does Cue cost anything to run a weekly report?

Cue is free during invite-only early access as of 29 September 2026, and Manus has not said what it costs afterwards. Manus sells Cloud Computer separately for automations that need a permanent machine. Flowsery includes the API and MCP server on both Team at $250 a month and Pro at $500 a month.

How many API calls does the weekly report make?

The report makes six read calls, and list_websites is only needed on the first run. Flowsery allows 600 requests a minute per key, so a weekly report never gets near the limit.

Which Flowsery token role should the report use?

Use Contributor. It can read everything the report needs, and it gets 403 permission_denied on the five write calls, because only Admin and Editor hold the flowsery.write permission. Create it in the Create Token form at flowsery.com/api-tokens.

Why did the report stop arriving after a teammate left?

Flowsery answers 401 with token_issuer_lost_access when the member who created the key has left the workspace. A scheduled report then fails silently. Create the token from an account that is going to stay.

Should the report include visitor names and emails?

Leave them out. GET /visitors/{visitorId} returns name and email for identified visitors, and a weekly report never needs it. Keep the prompt to aggregate calls and name every recipient, because a Cue agent has its own email address.

Create a Contributor token at flowsery.com/api-tokens, run the prompt above in Cue this Monday, and read the session analysis guide when the report points at a broken page.

Sources: Introducing Manus 2.0 (28 September 2026), What is new in Manus 2.0, Manus Automations docs, Manus custom MCP docs, Runtimewire (28 September 2026), the Flowsery OpenAPI spec and the Flowsery API reference. Checked 29 September 2026.

Was This Article Helpful?

Let us know what you think!

See us more often in Google

One click marks Flowsery as a preferred source, so our articles sit higher in your Top Stories, AI Mode, and AI Overviews.

Before you go...

Flowsery

Flowsery

Revenue-first analytics for your website

Track every visitor, source, and conversion in real time. Simple, powerful, and cookie-free.

Real-time dashboard

Goal tracking

Cookie-free tracking

Related Articles