PostHog MCP for Analytics

The PostHog MCP server lets your AIML.chat agents query events, insights, and funnels so they can answer product and usage questions with real numbers — read-only, so your analytics are never altered. It installs from the MCP marketplace in a few clicks and lets your agent cite live product data.

What your agents can do with PostHog

Once connected, an internal or product agent can answer "how many signups this week," "what's the drop-off in this funnel," or "which feature flag is rolled out" by querying PostHog and reporting the result with context. For product teams, that turns a knowledge agent into a fast way to check the numbers without opening a dashboard. Because analytics should never be mutated by a chatbot, this connector is read-only by design — there are no write tools, so there's nothing to approve and nothing that can change your data. Grounded in your own content plus your PostHog metrics, the agent gives specific, data-backed answers rather than vague ones.

Tools this MCP server exposes

The PostHog MCP server exposes read tools only — run an events or HogQL query, read an insight or funnel, look up a metric over a date range, list events and feature flags. There are deliberately no write tools: analytics is a read surface, so the agent can report numbers but never alter them. You scope the connection to the right project and decide which agents may use it. Because everything is read-only, you can enable it broadly without approval friction.

Connect it in three steps

Because the connector is read-only, there's no write-confirm step here — just accurate, cited numbers. You start on a real free tier and scale as usage grows — see pricing for the agent and message limits, and the MCP docs for the server contract. When you're ready, start free.