Analytics

The Analytics page gives you a usage overview for all MCP servers in your workspace. All charts are scoped to your workspace and updated from pre-aggregated snapshots for performance.

What's tracked

• Top tools — which tool names are called most, by volume
• Volume over time — call counts per hour as a heatmap
• Error rates — percentage of tool calls that failed, per server or globally
• Latency — p95 response time histogram, per server or globally

Data freshness

Analytics are pre-aggregated into hourly snapshots and cached in Redis with a 5-minute TTL. This means:

• Analytics data can be up to 5 minutes behind real-time
• Very recent tool calls (last few minutes) may not appear in charts
• Historical data going back up to 30 days is available

Filters

All analytics charts support filtering by:

• Time range — last 24h, 7 days, 30 days
• Server — focus on a specific MCP server

In this section

• Top tools and volume
• Errors and latency
• How aggregation works