Mixpanel AI Plugins

Plugins that give AI agents Mixpanel expertise. Built on the Agent Skills open standard.

The mixpanel plugin is engine-agnostic: its skills work through whichever engine you set up — the Mixpanel MCP server (any region) or the mixpanel-headless Python SDK. Run the install skill once; every other skill takes it from there.

Getting started

Claude Code

claude plugin marketplace add mixpanel/ai-plugins
claude plugin install mixpanel

Then, in your project, run /mixpanel:install to set up an engine and region.

Cursor

Install the plugin from the Cursor marketplace, or have a team admin import this GitHub repository as a team marketplace (Dashboard → Settings → Plugins → Import).

Skills appear in Cursor Settings → Rules under Agent Decides and can be invoked with /skill-name in chat. Run the install skill first to set up an engine.

Skills

Skill	Description
install	Sets up a Mixpanel engine for a project — MCP server (US/EU/India), mixpanel-headless SDK, or your own integration.
analyze-report	Reads and explains an existing report or chart — what it shows, what's notable, what's worth a closer look.
create-dashboard	Creates a well-designed dashboard with validated data, text cards, and narrative layout.
deep-research	Structured metric investigation — why a metric changed, what's driving a trend, root-cause deep dives.
learn-mcp	Guided, interactive onboarding to the Mixpanel MCP server (applies when the engine is MCP).
manage-boards	Full dashboard lifecycle — create, template, clone, clean up, inventory, and update across projects and teams.
manage-experiment	Coaches through every phase of an experiment — design (hypothesis, metrics, sizing, statistical model) and interpretation (results, ship/iterate/kill, health checks, segments, replays).
manage-feature-flags	Feature-flag guidance — Feature Gate vs Dynamic Config vs Experiment, staged rollouts, kill switch, exposure debugging, SDK patterns.
manage-lexicon	Audits, scores, enriches, and cleans up Lexicon metadata — descriptions, tags, resets, data-quality triage.
monitor-metrics	Monitors and diagnoses a metric — anomaly detection, drift detection, and root-cause attribution with charts and a diagnosis board.
prepare-ai-readiness	Sets up business context and delegates Lexicon enrichment so Mixpanel's AI assistants understand your data.
tracking-implementation	Guides analytics implementation — Quick Start, Full Implementation, Add Tracking, and Audit modes.

Internal skills

Skill	Description
review-skill	Reviews a skill against a weighted quality rubric and produces a score with actionable issues. Run /review-skill <skill-name> before requesting a code review.

Migrating from the legacy plugins

The region-specific plugins — mixpanel-mcp, mixpanel-mcp-eu, mixpanel-mcp-in — are retired: region is now a configuration choice inside mixpanel.

Most users — run /plugin marketplace update mixpanel. Your install is renamed to mixpanel automatically (Claude Code ≥ 2.1.193). The old plugin's bundled MCP server doesn't carry over, so run /mixpanel:install once to reconnect your engine and region.

Seeing plugin-not-found? Your Claude Code predates rename support. Update it, or migrate by hand:

/plugin uninstall mixpanel-mcp@mixpanel   # or mixpanel-mcp-eu / mixpanel-mcp-in
/plugin install mixpanel@mixpanel
/mixpanel:install

Prefer the old plugins? Pin the pre-restructure state (tag v0.1.1):

claude plugin marketplace add mixpanel/ai-plugins@v0.1.1

Contributing

Contributions are welcome — see CONTRIBUTING.md.

License

Apache 2.0 — see LICENSE.