Automate Mixpanel Data Analysis with Rube MCP

This skill automates Mixpanel product analytics through Composio's Mixpanel toolkit via Rube MCP, covering events, segmentation, funnels, cohorts, user profiles, and JQL queries. Its own rule is to always call RUBE_SEARCH_TOOLS first to get current tool schemas before running a workflow.

Prerequisites are Rube MCP connected with RUBE_SEARCH_TOOLS available, and an active Mixpanel connection through RUBE_MANAGE_CONNECTIONS with toolkit mixpanel. Setup means adding https://rube.app/mcp as an MCP server in your client configuration; it states no API keys are needed, then verifying RUBE_SEARCH_TOOLS responds, calling RUBE_MANAGE_CONNECTIONS with toolkit mixpanel, and following the returned auth link if the connection is not yet ACTIVE.

It documents six core workflows with tool sequences:

• Aggregate event data: MIXPANEL_GET_ALL_PROJECTS to get a project ID, then MIXPANEL_AGGREGATE_EVENT_COUNTS with event, from_date/to_date, unit, type, and a where filter.
• Segmentation queries: MIXPANEL_QUERY_SEGMENTATION, segmenting by a property such as properties["country"].
• Funnel analysis: MIXPANEL_LIST_FUNNELS to resolve a funnel_id, then MIXPANEL_QUERY_FUNNEL with a date range, segmentation property, and conversion window (length).
• User profiles: MIXPANEL_QUERY_PROFILES to search and filter profiles with pagination via page and session_id, and optionally MIXPANEL_PROFILE_BATCH_UPDATE using $set, $unset, $add, and $append operations, chunked to the API's per-request maximum.
• Cohorts: MIXPANEL_COHORTS_LIST, which returns each cohort's id, name, description, and count; cohorts themselves are created and managed in the Mixpanel UI.
• JQL and insight queries: MIXPANEL_JQL_QUERY for custom JavaScript-like JQL scripts, and MIXPANEL_QUERY_INSIGHT for a saved insight by bookmark_id; the source flags JQL as a legacy feature to check against current Mixpanel documentation.

It also documents Mixpanel's expression syntax for where and on parameters (property references as properties["name"], equality, comparison, boolean, contains, and AND/OR), and known pitfalls: dates must be 'YYYY-MM-DD' and are inclusive on both ends, string values in expressions must be quoted while numbers and booleans are not, large segmentation queries may hit rate limits or time out, response data may be nested under a data key, numeric values may come back as strings, and empty date ranges return empty objects rather than empty arrays.

Use it when a user wants to count or trend events, break down events by property, analyze conversion funnels, query or batch-update user profiles, list cohorts, or run a custom JQL or saved insight query in Mixpanel.

Do not use it to create funnels or cohorts, since it states these are created and managed in the Mixpanel UI and the API only reads or queries existing ones.

Inputs are the Mixpanel project (resolved via MIXPANEL_GET_ALL_PROJECTS), event names, date ranges, filter and segmentation expressions, funnel or bookmark IDs, and for batch updates, profile update objects keyed by $distinct_id. Outputs are event aggregations and trends, segmented breakdowns, funnel conversion results, profile query results and pagination tokens, cohort lists, and JQL or insight query results.

It is for product analysts and growth or data engineers who want to drive Mixpanel analytics programmatically through an MCP-connected agent instead of the Mixpanel UI.