Resources & Prompts¶

MCP resources provide read-only data access, and prompts offer guided workflow templates.

Explore on DeepWiki

Ask about specific resources, explore prompt workflows, or get help with MCP capabilities.

Resources¶

Resources expose data through MCP's resource protocol. AI assistants can read these directly without invoking tools.

Static Resources¶

Always-available resources that reflect current state.

URI	Description
`workspace://info`	Workspace configuration (project ID, region, account)
`workspace://tables`	List of local DuckDB tables
`schema://events`	All tracked event names
`schema://funnels`	Saved funnel definitions
`schema://cohorts`	Saved cohort definitions
`schema://bookmarks`	Saved report bookmarks

Usage example:

Read the workspace://info resource to see my configuration

Dynamic Resource Templates (v2)¶

Parameterized resources that accept arguments.

URI Template	Description
`analysis://retention/{event}/weekly`	12-week retention curve for an event
`analysis://trends/{event}/{days}`	Daily event counts for N days
`users://{id}/journey`	Event journey for a specific user

Usage examples:

Read analysis://retention/Signup/weekly
Read analysis://trends/Purchase/30
Read users://user@example.com/journey

Prompts¶

Prompts provide structured workflow templates for common analysis patterns. They guide the AI assistant through multi-step analysis processes.

analytics_workflow¶

Complete analytics exploration workflow.

Description: A comprehensive guide for exploring Mixpanel data, from discovery through analysis.

Workflow steps:

Discover available events and properties
Understand data volume and patterns
Run exploratory queries
Fetch data for deeper analysis
Execute SQL queries locally
Synthesize findings

When to use: Starting a new analytics exploration with no specific goal in mind.

funnel_analysis¶

Funnel conversion analysis workflow.

Description: Structured approach to analyzing conversion funnels.

Workflow steps:

List available funnels
Select funnel to analyze
Query conversion rates over time
Identify drop-off points
Segment by user properties
Recommend optimizations

When to use: Investigating conversion rates or funnel performance.

retention_analysis¶

User retention analysis workflow.

Description: Systematic retention and cohort analysis.

Workflow steps:

Define cohort (born event)
Define return criteria
Run retention query
Analyze by time intervals
Segment by acquisition source
Compare cohorts

When to use: Understanding user retention patterns or comparing cohort performance.

local_analysis_workflow¶

Local SQL analysis workflow.

Description: Guide for working with locally stored data.

Workflow steps:

List available tables
Inspect table schemas
Sample data for understanding
Build SQL queries iteratively
Extract insights
Export results if needed

When to use: Analyzing data that's already been fetched locally.

gqm_decomposition¶

Goal-Question-Metric investigation framework.

Description: Structured problem investigation using the GQM methodology.

Workflow steps:

Define the business goal
Formulate specific questions
Identify metrics for each question
Execute queries for each metric
Synthesize findings
Recommend actions

When to use: Investigating a specific business problem or goal.

Example: "Why is user activation declining?"

aarrr_analysis¶

Pirate metrics (AARRR) analysis.

Description: Comprehensive product health analysis using the AARRR framework.

Metrics covered:

Metric	Description
Acquisition	How users find the product
Activation	First value experience
Retention	Repeat usage patterns
Revenue	Monetization metrics
Referral	Viral growth indicators

Workflow steps:

Define events for each stage
Query acquisition metrics
Measure activation rates
Analyze retention curves
Track revenue events
Identify referral patterns

When to use: Getting a holistic view of product health.

experiment_analysis¶

A/B test and experiment analysis.

Description: Framework for analyzing experiment results.

Workflow steps:

Identify experiment cohorts
Define success metrics
Query metric values per cohort
Calculate statistical significance
Analyze segment performance
Make recommendations

When to use: Analyzing A/B tests or feature experiments.

MCP Capabilities¶

The server leverages advanced MCP features for enhanced functionality.

Feature	Usage	Graceful Degradation
Sampling	`ctx.sample()` for LLM analysis of query results	Returns raw data with hints
Elicitation	`ctx.elicit()` for interactive workflows	Proceeds with warning
Tasks	Progress reporting via `ctx.report_progress()`	Synchronous execution
Middleware	Request interception for caching, rate limiting, audit	N/A

Sampling (ctx.sample)¶

Used by Tier 3 intelligent tools to synthesize query results. When the AI assistant processes tool results, it can understand and explain complex patterns.

Example flow:

diagnose_metric_drop queries multiple dimensions
Results are passed to ctx.sample() with analysis prompt
LLM synthesizes findings into actionable insights

Elicitation (ctx.elicit)¶

Used by interactive tools to request user confirmation or input.

Example flow:

safe_large_fetch estimates data volume
If large, ctx.elicit() asks for confirmation
User confirms or cancels
Fetch proceeds based on response

Progress Reporting (ctx.report_progress)¶

Long-running operations like fetch_events report progress:

Fetching events: 15,234 / ~50,000 (30%)

Middleware Layer¶

Cross-cutting concerns handled transparently.

Caching¶

Discovery tools cache responses to reduce API calls.

Scope	TTL	Tools Affected
Schema	5 minutes	list_events, list_properties, list_funnels, etc.

Rate Limiting¶

Respects Mixpanel API limits automatically.

API	Rate Limit	Concurrent Limit
Query API	60/hour	5 concurrent
Export API	60/hour, 3/sec	100 concurrent

When limits are reached, requests are queued and wait time is reported.

Audit Logging¶

All tool invocations are logged with:

Tool name
Parameters
Execution time
Result summary
Errors (if any)