Posthog

What PostHog/posthog MCP server does

PostHog/posthog MCP server gives compatible AI agents access to PostHog as a product context layer. It is intended for working with data and operations held in PostHog, including product analytics, web and group analytics, session recordings, heatmaps, errors, logs, traces, feature flags, experiments, surveys, support, workflows, warehouse data, and AI observability.

The available product areas cover both investigation and iteration. An agent can use PostHog context to understand user behavior, examine errors and recordings, inspect product data, and work with rollout or experimentation workflows. The README describes the MCP connection as a way to query and act on PostHog data; it does not provide a complete per-tool catalog in the supplied material.

How it works

PostHog exposes the MCP service remotely at https://mcp.posthog.com/mcp. An MCP-compatible client connects to that endpoint, and the user authenticates with PostHog. The service selects the correct data region, either US or EU, when the user signs in.

This design avoids running the MCP service locally for the standard setup. The client remains the place where an agent is used, while PostHog supplies the product data and actions available through the connection. The README names Claude Code, Cursor, Codex, VS Code, Zed, and PostHog Desktop as installation targets; PostHog Desktop is not included in the compatible-client list here because it is not one of the supported directory slugs.

Setup and configuration

PostHog/posthog MCP server can be added with the PostHog wizard:

npx @posthog/wizard mcp add

The repository points to client-specific setup guides in the PostHog documentation. A PostHog instance and user authentication are required before the connection can provide account data. The supplied material describes PostHog Cloud US and EU, as well as self-hosted PostHog deployments, but it does not specify whether every MCP feature or endpoint behavior is identical across those deployment types.

No required environment variables are documented for the remote connection. The endpoint, authentication flow, and regional routing are handled through the client setup and PostHog login process described above.

Tools and capabilities

PostHog/posthog MCP server is suited to agent workflows involving:

• Querying product, web, group, and warehouse analytics
• Investigating errors, logs, traces, and AI application observability data
• Finding and examining session replays and heatmap information
• Working with feature flags, experiments, and early-access features
• Reviewing surveys, support context, and workflows
• Using PostHog data while developing in an MCP-compatible editor or coding agent

The broader PostHog platform also includes data pipelines, endpoints, and MCP analytics. The supplied README establishes these as PostHog capabilities, but does not state that each one is exposed as a distinct MCP tool.

Limitations and notes

PostHog/posthog MCP server requires a compatible MCP client and a PostHog login. The remote endpoint routes to the US or EU region based on the authenticated user, so regional account selection is part of the connection behavior. The README labels MCP analytics as beta, but does not label the entire MCP server beta.

The provided material does not list exact tool names, argument schemas, permission requirements, or a full matrix of supported client versions. Developers who need those details should consult the linked MCP documentation after installation. PostHog’s free usage limits and product-specific pricing also apply to the connected PostHog account.