Keen.io

@keen list all event collections in our production project and show me the schema for each one

Found 4 event collections in your Keen project: 'user_signups' (properties: email [string], plan [string], timestamp [datetime]), 'page_views' (properties: url [string], user_id [string], duration_ms [number]), 'api_requests' (properties: endpoint [string], status_code [number], latency_ms [number]), and 'feature_flags' (properties: flag_name [string], enabled [boolean], user_segment [string]). Each collection includes inferred property types based on ingested events.

This example uses the Inspect All Event Collections tool with schema enabled. Requires a valid API key with read permissions. The schema shown reflects Keen's type inference from actual event data — if your events have inconsistent property types, the schema may show 'mixed' or the most recent type.

@keen show me all unique user segments that have triggered the 'checkout_completed' event in the last 7 days

Querying 'checkout_completed' collection for distinct 'user_segment' values from the past 7 days. Found 5 unique segments: 'enterprise', 'pro', 'starter', 'trial', and 'free'. This represents 2,847 total checkout events across these segments. The 'enterprise' segment accounts for 61% of completed checkouts in this timeframe.

This example demonstrates the Select Unique tool with timeframe filtering. Useful for segmentation analysis or building dynamic filters. Be aware of Keen's query limits — projects on lower tiers may have daily query caps, and results are constrained by your event retention window.

@keen analyze our 'api_requests' collection and summarize which endpoints have the highest error rates this month

Inspecting 'api_requests' collection schema shows 'endpoint' and 'status_code' properties. Querying events from the past 30 days: '/v1/auth/login' has a 12.3% error rate (status >= 400), '/v1/data/export' shows 8.7%, and '/v1/webhooks/receive' has 4.2%. The login endpoint's errors spiked on March 15th, suggesting a potential incident. Total API volume: 1.2M requests this month.

This example pairs Keen's schema inspection with the AI's analytical reasoning to surface actionable insights. The AI synthesizes data from multiple queries (schema + filtered aggregations). Requires read access and sufficient query quota. Error rate calculations assume your events include HTTP status codes.

When Keen.io wins for event tracking cleanup

A 5-person product team ships a new feature every two weeks and logs 40+ custom events to Keen.io. After six months, no one remembers which properties live in which collections or what types they expect. The Inspect All Event Collections tool surfaces every collection name and schema in one call, then Inspect Single Event Collection drills into the messy ones. This MCP is the right call if your team already uses Keen.io and needs to audit or document event schemas without opening the web console. If you're evaluating analytics platforms or your events live in Mixpanel or Amplitude, this MCP won't help. Use Switchy to run the audit once a quarter and paste the schema into your team wiki.

When this MCP helps debug user-reported issues

A 3-person support team fields tickets about missing data in a SaaS dashboard. The engineering team logs all user actions to Keen.io, but support can't query events without asking a developer. The Select Unique tool lets support agents pull distinct values for properties like user_id or session_id, filtered by timeframe, to confirm whether an event fired. This works if your support team already has read-only Keen.io API keys and the event volume is under 100k per day. Above that threshold, queries slow down and you need a dedicated BI tool. If your events live in a different analytics backend, this MCP won't connect. Use Switchy to let support run one-off queries during live troubleshooting calls.

When Keen.io MCP handles key rotation at scale

A 10-person engineering team rotates Keen.io API keys every quarter for compliance. Developers occasionally revoke keys by mistake, breaking staging environments until someone logs into the Keen.io console to unrevoke them. The Unrevoke Access Key tool restores a key in one call, and List Cached Dataset Definitions confirms which datasets still reference old keys. This MCP is the right call if your team manages more than five Keen.io projects and rotates keys on a schedule. If you only have one project or rotate keys once a year, the web console is faster. If your analytics platform isn't Keen.io, this MCP won't apply. Use Switchy to script key checks into your quarterly rotation runbook.