Problem
The list_events tool currently hardcodes dataset to only ["errors", "logs", "spans"]:
https://github.com/getsentry/sentry-mcp/blob/main/packages/mcp-core/src/tools/list-events/index.ts#L67
However, on self-hosted Sentry instances with metrics-only configuration:
- The
spans dataset returns empty (no event-level span storage)
- The Sentry UI uses
dataset=metricsEnhanced to display transaction metrics
- The
metricsEnhanced dataset works perfectly via direct API calls
Impact
Users cannot access transaction performance data (p50, p75, p95, p99) through MCP, even though:
- The data exists and is visible in the UI
- The API endpoint supports it
- The
list_events tool already supports aggregate functions
Proposed Solution
Add "metricsEnhanced" to the dataset enum in list-events/index.ts:
dataset: z
.enum(["errors", "logs", "spans", "metricsEnhanced"])
.default("errors")
Problem
The
list_eventstool currently hardcodes dataset to only["errors", "logs", "spans"]:https://github.com/getsentry/sentry-mcp/blob/main/packages/mcp-core/src/tools/list-events/index.ts#L67
However, on self-hosted Sentry instances with metrics-only configuration:
spansdataset returns empty (no event-level span storage)dataset=metricsEnhancedto display transaction metricsmetricsEnhanceddataset works perfectly via direct API callsImpact
Users cannot access transaction performance data (p50, p75, p95, p99) through MCP, even though:
list_eventstool already supports aggregate functionsProposed Solution
Add
"metricsEnhanced"to the dataset enum inlist-events/index.ts: