Product analytics and marketing analytics have always been two silos. Your ad campaigns drive users to sign up — but whether those users activate, convert to paid, or churn lives in PostHog, not in Google Ads or GA4. Until today there was no clean way to bring those two data sets together in one agent conversation. PostHog is now a live MCP integration in Cogny.

One personal API key, your instance host, and the agent has read access to everything PostHog knows about your users.

TL;DR

PostHog is live now. Connect it from your workspace Settings → Cogny Managed MCPs.
13 read-only tools: get_current_user, list_projects, list_insights, get_insight, list_dashboards, list_events, list_event_definitions, list_property_definitions, list_persons, get_person, list_cohorts, list_session_recordings, list_feature_flags. No write tools are registered.
Works with EU Cloud (eu.posthog.com), US Cloud (us.posthog.com), and self-hosted instances — the instance host is configurable per workspace.
Auth is a personal API key (phx_…) you create in PostHog with read-only scopes. No OAuth app registration needed.
Rate limits are org-wide in PostHog (240 req/min for analytics endpoints); the proxy surfaces HTTP 429 with a clear message instead of silently retrying.
Available on any paid Cogny plan (Solo $9/mo and above). The agent pays for its own inference; Cogny handles the proxy and credential storage.

What you can do with it

Tool	What Claude can do
`list_projects`	List all projects the key can access — use to resolve `project_id`
`list_insights` + `get_insight`	Read saved funnels, trends, retention, and lifecycle charts
`list_dashboards`	List dashboards and their tiles
`list_events`	Query recent events by name, date, `distinct_id`, before/after
`list_event_definitions`	Discover which event names exist in the project
`list_property_definitions`	Discover available event and person properties
`list_persons` + `get_person`	Search user profiles; filter by email or `distinct_id`
`list_cohorts`	Read cohort definitions
`list_session_recordings`	Read recording metadata with date filters
`list_feature_flags`	Read feature flag keys, rollout rules, and status
`get_current_user`	Validate credentials and see project access

In concrete terms: "Which acquisition source produces users who complete onboarding within 3 days?" "Show me the funnel drop-off from signup to first purchase." "Which cohort had the best 30-day retention last month?" "Is the beta feature flag live for more than 20% of users?" All of those come from PostHog. The Cogny agent can now answer them in the same conversation where it checks your ad spend and GA4 traffic.

How to connect it

In PostHog, go to Settings → Personal API keys and click Create personal API key.
Name it (e.g. "Cogny") and grant these read-only scopes: project:read, insight:read, dashboard:read, event_definition:read, person:read, cohort:read, session_recording:read, feature_flag:read.
Copy the key — it starts with phx_ and is only shown once.
In your Cogny workspace, go to Settings → Cogny Managed MCPs, click Connect on the PostHog card, and paste:
- The personal API key
- Your instance host: https://eu.posthog.com (EU Cloud), https://us.posthog.com (US Cloud), or your self-hosted domain — no trailing slash.

The key is stored encrypted, sent only as Authorization: Bearer <key> to your PostHog instance, and scoped to the single workspace. Disconnect at any time.

On rate limits

PostHog's rate limits are shared across your entire organization, not per key: 240 requests/minute for analytics endpoints (insights, events) and 480 requests/minute for everything else. In most interactive agent sessions this is well within limits. The proxy returns a clear error message on HTTP 429 — "Rate limit exceeded. PostHog rate limits are shared across your entire organization. Wait before retrying." — so the agent knows to pause rather than loop.

On read-only trust

This integration is strictly read-only v1. The proxy registers no create/update/delete tools, so an agent cannot post events, modify feature flags, or touch any PostHog data. If write support turns out to be useful (e.g. creating or toggling feature flags from an automation loop), it will ship behind an explicit approval gate.

The pairing of ad-platform data with in-product behavioral data is where growth analytics gets interesting — and it's now one agent session. Connect PostHog from your Cogny dashboard and ask your first question.