mixpanel-data

Manage Mixpanel authentication — check status, add/switch/test accounts, OAuth login. Use with no arguments for a quick status check.

Use this agent for general-purpose Mixpanel data analysis, answering product analytics questions, building dashboards, or investigating metrics. This is the orchestrator agent that delegates to explorer, diagnostician, or narrator when needed. <example> Context: User wants to understand their product metrics user: "How are our key metrics trending this month?" assistant: "I'll use the analyst agent to pull and analyze your key product metrics." <commentary> General analytics question about product health — the analyst orchestrator handles this, querying multiple metrics and synthesizing. </commentary> </example> <example> Context: User asks about a specific metric user: "How many signups did we get last week broken down by country?" assistant: "I'll use the analyst agent to query your signup data segmented by country." <commentary> Specific data question requiring a segmentation query — analyst handles directly. </commentary> </example> <example> Context: User wants to create or modify Mixpanel entities user: "Create a new cohort of users who signed up in the last 30 days and made a purchase" assistant: "I'll use the analyst agent to create that cohort using the Mixpanel App API." <commentary> Entity management via App API — analyst handles CRUD operations. </commentary> </example>

Use this agent for root cause analysis when a metric has changed unexpectedly. Specializes in diagnosing "why did X drop/spike?" questions through systematic segmentation and correlation analysis. <example> Context: User notices a metric changed user: "Why did our signup conversion drop last week?" assistant: "I'll use the diagnostician agent to systematically investigate the conversion drop across multiple dimensions." <commentary> Classic "why did X change?" question — diagnostician segments across dimensions to isolate the root cause. </commentary> </example> <example> Context: User sees an unexpected spike user: "We're seeing a huge spike in error events since Tuesday. What happened?" assistant: "I'll use the diagnostician agent to investigate the error spike, find the inflection point, and identify the affected segments." <commentary> Unexpected metric change needing root cause analysis with temporal and dimensional investigation. </commentary> </example> <example> Context: User reports metric divergence user: "Signups are up but activation is down. What's going on?" assistant: "I'll use the diagnostician agent to investigate the divergence between signup and activation metrics." <commentary> Metric divergence requiring correlation analysis and segment-level investigation. </commentary> </example>

Use this agent for open-ended or vague analytics questions that need systematic decomposition before querying. Specializes in schema discovery, hypothesis generation, and GQM (Goal-Question-Metric) analysis. <example> Context: User asks a vague question about their product user: "What's going on with our mobile app?" assistant: "I'll use the explorer agent to systematically investigate your mobile app metrics using GQM decomposition." <commentary> Vague, open-ended question — explorer decomposes into specific measurable sub-questions before querying. </commentary> </example> <example> Context: User wants to understand their data landscape user: "I'm new to this Mixpanel project. What data do we have?" assistant: "I'll use the explorer agent to discover and map your event schema, properties, and saved entities." <commentary> Schema exploration and data discovery — explorer's primary strength. </commentary> </example> <example> Context: User has a broad goal without specific metrics user: "Are our users getting value from the product?" assistant: "I'll use the explorer agent to decompose this into measurable questions across the AARRR framework." <commentary> Broad business question needing framework-based decomposition into specific queries. </commentary> </example>

Use this agent to synthesize data findings into polished executive summaries, stakeholder reports, and strategic narratives. Takes raw analysis output and transforms it into business-ready documentation. <example> Context: User needs to present analytics findings to leadership user: "Can you put together an executive summary of our Q1 metrics for the board?" assistant: "I'll use the narrator agent to compile a polished executive summary with key metrics, trends, and strategic recommendations." <commentary> Executive-level reporting request — narrator queries key metrics and synthesizes into a board-ready narrative. </commentary> </example> <example> Context: User wants a comprehensive product health report user: "Generate a monthly product health report for March" assistant: "I'll use the narrator agent to create a comprehensive report covering all key product metrics for March." <commentary> Structured report generation — narrator pulls data across AARRR stages and creates a formatted report. </commentary> </example> <example> Context: User has analysis results and needs them formatted user: "Take these findings and write them up as a report I can share with the product team" assistant: "I'll use the narrator agent to transform these findings into a structured, shareable product report." <commentary> Synthesis and formatting of existing analysis into stakeholder-ready documentation. </commentary> </example>

Translate natural language analytics questions into typed Workspace.query() calls. Use when the user describes what they want to measure and needs the exact Python code — DAU trends, conversion formulas, filtered breakdowns, rolling averages, property aggregations, and multi-metric comparisons. <example> Context: User wants to measure daily active users user: "Show me DAU over the last 30 days" assistant: "I'll use the query agent to translate that into a query() call with math='dau'." <commentary> Direct insights query — maps to a single query() call with math="dau". </commentary> </example> <example> Context: User wants a conversion rate formula user: "What's the signup to purchase conversion rate by country?" assistant: "I'll use the query agent to build a multi-metric formula query with group_by." <commentary> Multi-metric formula with breakdown — requires Metric objects, formula, and group_by. </commentary> </example> <example> Context: User wants a property aggregation user: "What's the average revenue per user by plan type?" assistant: "I'll use the query agent to build a per_user aggregation query." <commentary> Per-user property aggregation with breakdown — math="total", per_user="average", math_property, group_by. </commentary> </example>

This skill should be used when the user asks about Mixpanel product analytics, event data, funnel analysis, retention curves, cohort analysis, segmentation queries, JQL, user behavior, conversion rates, churn, DAU/MAU, ARPU, revenue metrics, feature adoption, A/B test results, or any request to query, explore, visualize, or analyze Mixpanel data using Python.

This skill installs mixpanel_data, pandas, numpy, matplotlib, and seaborn, then verifies Mixpanel credentials. It should be invoked when setting up a new environment for Mixpanel data analysis, when dependencies are missing, or when configuring service account or OAuth credentials for the first time.