PostHog MCP connector for Manus

Ask PostHog what changed, then check the flag, the error and the query

Ask “Did payment success change after yesterday’s checkout launch?” Manus checks your PostHog data and returns a chart, a one-line answer and the SQL it used. Ask which feature flags are on and for how many users, what an experiment found, or which errors hit the most people this week. Before it changes a flag or creates an experiment, ask it to show the change, and it waits for your OK.

First, connect PostHog in Manus
ManusIllustration

Did payment success change after we shipped the new checkout page yesterday? Also show me which flags are on in PostHog.

Yes. Payment success is up about 2 points since the launch: 96.2% yesterday and 96.0% so far today, against 94.1% on average in the 7 days before. Check again after a full week.

Payment success rateOct 13 to Oct 21
97%95%93%
New checkout shipped
Oct 13Oct 17Oct 21
Query I used
SELECT toDate(timestamp) AS day,  countIf(event = 'payment_succeeded')  / countIf(event = 'checkout_started') AS success_rateFROM eventsWHERE timestamp >= '2026-10-13'GROUP BY day ORDER BY day
Feature flagsIllustration
  • new-checkoutNew checkout pageAll users
  • saved-cardsSaved cards at checkout10% of users
    Suggestion: go to 25% next and watch payment errors for a day. I’ll change it after your OK.
  • pricing-page-v2New pricing page · experiment running50% of users
  • onboarding-checklistOnboarding checklist50% of users
  • legacy-cartOld cart pageOff
Recent changes
  • new-checkout · Set to all usersBy Leo Park on Oct 20
  • saved-cards · Set to 10% of usersBy Maya Chen on Oct 14

Get PostHog SQL insights from a plain-English question

Get PostHog AI analytics without writing SQL. Manus writes the PostHog SQL query, runs it and shows the result with the query attached, so you can check it or open it in the PostHog SQL editor.

SQL insightIllustration

Write a SQL query for me: people who signed up in the last 30 days but haven’t finished onboarding, by signup week. Then save it as an insight.

SELECT toStartOfWeek(timestamp, 1) AS week,  count(DISTINCT person_id) AS stuck_usersFROM eventsWHERE event = 'user_signed_up'  AND timestamp > now() - INTERVAL 30 DAY  AND person_id NOT IN (SELECT person_id FROM events    WHERE event = 'onboarding_completed')GROUP BY week ORDER BY week
weekstuck_users
  • Sep 21412
  • Sep 28388
  • Oct 5455
  • Oct 12521
  • Oct 19 (so far)203

521 people who signed up the week of Oct 12 still haven’t finished onboarding, the most of any week in the last 30 days.

Saved as insight: Signups stuck in onboarding

Turn a question into a saved PostHog SQL insight

Say what you want to count. You get the SQL, the result and a saved insight.

Find where users drop off in a PostHog funnel

Run a PostHog funnel analysis from signup to first payment and see which step loses the most people.

Feature flag management for PostHog: rollouts, change history and stale flags

Ask which PostHog feature flags are on, roll out a feature to 10% of users or more, or clean up old flags. Ask Manus to show each change first, and it waits for your OK.

Change to reviewIllustration
saved-cardsSaved cards at checkout
Rollout10%25%
Release condition
All users
Users affected
About 4,800 more users (estimate)
Last changed
By Maya Chen on Oct 14
Waiting for your OK
Flags not used in 90 daysIllustration
  • legacy-cartLast used Jul 2
  • old-onboarding-tourLast used Jun 18
  • holiday-banner-2025Last used Jan 6
These 3 flags are off, and none has been checked since early July. I’ll delete them only after you confirm.

Roll out a PostHog feature to more users

Go from 10% to 25% of users when the numbers look good. Ask to see the change and who it reaches first.

See who changed a PostHog feature flag and when

Get every change to a flag in order, with the person, the time and what changed.

Clean up PostHog flags no one uses

Find flags that haven’t been checked in 90 days. Ask for the list first, and tell it which ones to remove.

Ask which PostHog experiment variant won, and by how much

Get experiment results with the numbers and a plain answer: which variant is ahead, by how much and whether you can trust it yet. Ask Manus to show a new A/B test before it launches, and it waits for your OK.

Get PostHog experiment results in plain words

See which version is ahead, by how much and whether you can trust it yet.

Set up a PostHog A/B test, checked before launch

Manus drafts the experiment, its flag and the goal. Ask to see the setup first, and it launches only after your OK.

Experiment resultsIllustration
pricing-page-v2Running since Oct 14 · goal: reach checkout
  • Current page6,240 users4.0%
  • New pricing page6,198 users4.6%
Checkout conversion by day
5.0%3.5%
Oct 14Oct 17Oct 20
The new pricing page is ahead: 4.6% of visitors went on to checkout, against 4.0% on the current page. That gap isn’t reliable yet. At this traffic, give it about 6 more days before you decide.

See the top PostHog errors this week and who they hit

Ask which errors hit the most people, read the stack trace and turn the worst one into a fix or an issue. Ask Manus to check with you before it changes an error’s status.

Top errors this weekSince Oct 19Illustration
  • TypeErrorCannot read properties of undefined (reading 'total')1,284 times · 312 users · 4 min ago
    at CartSummary (src/checkout/CartSummary.tsx:42:18)at renderWithHooks (react-dom.production.min.js:160:2)at CheckoutPage (src/checkout/CheckoutPage.tsx:88:7)
    This one first showed up on Oct 20, the day new-checkout went to all users. It comes from line 42 of CartSummary.tsx, where the order summary can load before the cart total.
  • ChunkLoadErrorLoading chunk 41 failed640 times · 205 users · 1 hr ago
  • ErrorRequest failed with status code 502318 times · 97 users · 25 min ago
  • RangeErrorInvalid time value142 times · 38 users · 3 hr ago

Read the stack trace behind a PostHog error

Get the full trace, when it first showed up and the likely cause in plain words.

Find the code behind a PostHog error in GitHub

Manus finds the file in your repo and drafts a fix as a pull request you review.

More about GitHub in Manus

Mark a PostHog error resolved once the fix ships

Ask Manus to check with you first, and it changes the status only after you confirm.

A weekly product report from PostHog, sent where your team works

Use your PostHog integrations in one place. Manus posts the weekly numbers to Slack, turns errors into Linear issues and saves findings to a project.

Weekly product reportIllustration
Sent Monday Oct 19, 9:00 AM · for Oct 12 to Oct 18
  • Weekly active users18,420+3.2% vs last week
  • Signups2,140+1.8% vs last week
  • Checkout conversion4.3%Same as last week
Weekly active users, last 8 weeks
Aug 24Oct 12
  • pricing-page-v2 went to 50% of users on Oct 14.
  • saved-cards went to 10% of users on Oct 14.
  • The pricing-page-v2 experiment is running. It’s too early to call.
Top 3 errors
  • ChunkLoadErrorLoading chunk 41 failed188 users
  • ErrorRequest failed with status code 50264 users
  • RangeErrorInvalid time value31 users
Posted to #product in SlackSaved to the Growth project

Get a weekly product report from PostHog every Monday

Set it up once. Manus runs it as an automation and posts the numbers, flag changes, experiments and top errors.

See Automations

Turn the top PostHog error into a Linear issue

The issue includes the stack trace and the number of users affected. Ask to see it before it’s created.

More about Linear in Manus

Connect the PostHog MCP server to Manus

PostHog is already in Manus Connectors. Sign in once, and any task can use PostHog MCP analytics, feature flags, experiments and error tracking.

  1. Open Connectors

    In Manus, open Connectors, choose PostHog and select Connect.

  2. Sign in to PostHog

    Follow the steps to sign in with your PostHog account and allow access. Manus then works with the projects and data your PostHog account can already see.

  3. Ask in any task

    Mention PostHog and the metric, flag, experiment or error. Before a change, ask Manus to show it first and wait for your OK.

PostHogPerform product analytics, manage feature flags, and run experiments

Transform your product analytics with AI-powered insights. Easily perform data analytics, manage feature flags, and run experiments using natural language commands.

ConnectIllustration
  • Sign in with your PostHog account and allow access
  • Sees and changes only what your PostHog account can

More ways to use PostHog with Manus

Questions about PostHog AI and MCP

How do I connect PostHog to Manus?

Open Connectors in Manus, choose PostHog and select Connect. Follow the steps to sign in with your PostHog account and allow access. After that, mention PostHog in any task, for example “Which feature flags are on in PostHog?”

Will Manus change my feature flags by mistake?

Manus changes a flag or an experiment in PostHog when your task asks it to. Ask it to show the change first, and it waits for your OK before it changes a rollout, creates an experiment or ships a variant. The examples on this page that change a flag or an experiment all ask for this.

Can I check the numbers Manus gives me?

Yes. Ask it to include the query it ran. For SQL questions you get the SQL itself, so you can run it again in the PostHog SQL editor or save it as an insight.

Will this cost more in PostHog?

Connecting to the PostHog MCP server is free, according to PostHog. Your normal PostHog usage still applies to the data your queries read, and a few PostHog tools that use AI count toward PostHog AI usage. Check PostHog’s own pricing for the details.

Can Manus send me a weekly product report from PostHog?

Yes. Ask for the report once and say when to run it, for example every Monday at 9 AM. Manus runs it as an automation and puts the key numbers, flag changes, experiment results and top errors in one message.

How is Manus different from PostHog AI, PostHog’s own AI?

PostHog AI is the AI analyst you use inside PostHog. Manus works with your PostHog data through the connector and takes the answer to the next step: a weekly report, a doc or a task, together with the other tools you’ve connected.

Is this like the PostHog Claude connector?

Yes, both use the PostHog MCP server. The PostHog Claude integration lets Claude work with your PostHog data, and PostHog in Manus is an MCP connector too. Connect it once in Manus Connectors, and any task can use it alongside the other tools you’ve connected.

Can Manus send PostHog errors to Linear or results to Slack?

Yes, if those apps are connected in Manus. It can turn the top PostHog error into a Linear issue or post a summary to a Slack channel, and you can ask to see it first. This works across your PostHog integrations in Manus, with no separate PostHog Slack integration to set up.

What can Manus see in PostHog?

Only what your PostHog account can see. If you can’t open a project, an insight or a flag in PostHog, Manus can’t either. You can disconnect PostHog in Manus Connectors at any time.

Do I need to write SQL or code?

No. PostHog is already in Manus Connectors. Ask in plain words, for example “Write a SQL query for me that finds users who signed up but didn’t finish onboarding,” and Manus writes and runs it.

Use Manus as your PostHog AI agent

Connect PostHog in Manus

Less structure,
more intelligence.

Product

PricingWeb appAI designAI slidesAI image generatorAI music generatorVideoWide ResearchSlack integrationAll modelsMore features

Resources

DownloadBlogDocsUpdatesHelp centerTrust centerAPITeam planStartupsPlaybookBrand assets

Community

EventsFellows

Compare

VS ChatGPTVS LovableVS Replit

Download

Desktop appMy Browser

Business

Team planSSOAPI

Company

About usCareersFor mediaTerms of servicePrivacy policy
© 2026 Manus AI