Base44
Build and edit Base44 apps from an AI assistant, and query data your app's users submit.
Run HogQL, build dashboards, ship feature flags and triage errors.
Opens Agent Studio, where connecting is one click. The connector URL below works in any MCP client.
Paste it into any MCP client. Setup docs
The PostHog MCP server connects a PostHog product-analytics project to Claude, Cursor, Codex and any MCP-compatible agent. It runs HogQL and SQL queries, builds insights and dashboards, creates and rolls out feature flags, manages experiments and surveys, and triages errors and logs. Sign-in is OAuth, and the server advertises 139 scopes — the most granular consent surface in this catalogue.
Verified connector
Listed by Anthropic as a partner connector in its Connectors Directory.
Connection checked by Agentman on .
Anthropic states this reflects the level of review a connector received, not a security audit.
Tool names from Anthropic's directory listing. This server requires sign-in, so we could not read tool descriptions or parameter schemas.
PostHog describes it as a hosted Model Context Protocol endpoint that exposes PostHog's products — feature flags, product analytics, error tracking, experiments, SQL queries, CDP, surveys and more — as function-calling tools to any MCP-compatible agent. It runs at https://mcp.posthog.com/mcp over streamable HTTP, and PostHog's authentication server routes you to the correct data region automatically.
Connecting to the server and calling its tools is free, per PostHog's MCP FAQ. Standard PostHog usage and billing still apply to the underlying data your queries consume, so an agent running large HogQL queries spends against your normal plan. Some tools run LLMs internally and may be billed as PostHog AI spend.
Yes, but no separate MCP limit. PostHog states that tool calls execute against its API and inherit the same API rate limits. Analytics endpoints allow 240 requests per minute and 1,200 per hour; the query endpoint allows 2,400 per hour. These limits apply per team, across every user and key in the organization.
No. PostHog states OAuth is the recommended path and works out of the box with its wizard. If your client does not support OAuth, you can use a personal API key with the MCP Server preset, which scopes access to a specific project, and pass it as an Authorization Bearer header in your MCP configuration.
PostHog answers this directly and does not simply reassure. Its FAQ warns that language models can be tricked into following untrusted commands, so you should review tool calls before executing them. It then names three controls: restrict the session to read-only tools, filter which tools are exposed, and pin the agent to one organization or project.
To your own PostHog instance. PostHog states the MCP server acts as a proxy and is routed automatically to the correct region, US or EU, based on the account you sign in with. It does not store your analytics data — queries execute against your PostHog project and results return directly to your AI client.
Yes, on an enterprise plan. PostHog documents enterprise-managed authorization using ID-JAG, which lets you control MCP access through an identity provider such as Okta or Entra ID instead of per-user OAuth or personal API keys. On other plans, access is granted per user through the OAuth consent screen or an API key.
Connect PostHog once and your agents call these tools on their own: on a schedule, in a workflow, with nobody at the keyboard.