MCP servers that query logs, metrics & traces
Let your AI investigate incidents across Datadog, Grafana, Prometheus & more.
14 servers · Last updated August 1, 2026
TL;DR: These servers connect your agent to observability data — querying logs, metrics, traces and alerts so it can help investigate incidents in plain English. They're the foundation of an on-call/SRE assistant. Coverage varies by backend, so match the server to the stack you actually run.
Bottom line: if you only try one, PostHog MCP Server (Official Remote) is the most popular, verified option for this (350★). 13 more compared below.
Compare 14 servers
| Server | Transport | Auth | Verified | Stars | Tools for this |
|---|---|---|---|---|---|
| PostHog MCP Server (Official Remote) | Local (stdio) | API key | 350 | insight-query, query-error-tracking-issues-list, query-llm-traces-list | |
| Prometheus MCP Server | Local (stdio) | No auth | 340 | execute_query, execute_range_query, list_metrics +1 | |
| Datadog MCP Server (Official Remote) | Remote (HTTP) | OAuth | 250 | search_datadog_logs, get_logs / query log analytics, query_timeseries_data +4 | |
| Honeycomb MCP Server | Local (stdio) | API key | 250 | run_query, get_trace_link | |
| Dynatrace MCP Server | Local (stdio) | OAuth | 200 | create_dynatrace_notebook | |
| Axiom MCP Server (Official Remote) | Local (stdio) | OAuth | 130 | queryApl | |
| VictoriaMetrics MCP Server | Local (stdio) | API key | 130 | query, query_range, metrics +4 | |
| Last9 MCP Server | Local (stdio) | API key | 90 | get_logs, get_service_logs, get_traces +3 | |
| Grafana Tempo MCP Server | Local (stdio) | No auth | — | 90 | tempo_query |
| PagerDuty MCP | Local (stdio) | API key | — | 72 | list_incidents, get_incident, manage_incidents +19 |
| New Relic MCP Server | Local (stdio) | API key | 60 | query_nrql, nerdgraph_query, list_alert_policies +1 | |
| Amplitude MCP | Remote (HTTP) | OAuth | — | 45 | query_chart, query_experiment, query_dataset +1 |
| Inspektor Gadget MCP Server | Local (stdio) | No auth | — | 26 | gadget_trace_dns, gadget_trace_tcp, gadget_trace_exec +4 |
| Mixpanel MCP | Local (stdio) | API key | — | 19 | query_insights_report, query_funnel_report, query_retention_report +6 |
The servers
Official PostHog server: product analytics, feature flags, experiments, error tracking and SQL.
Run PromQL queries and analyze Prometheus metrics from any MCP client.
Datadog's managed remote server: query logs, metrics, traces, monitors and incidents.
Honeycomb observability via AI: query datasets, alerts and boards (community + official).
Official Dynatrace server: run DQL over logs, events, spans and metrics.
Query Axiom logs and events with APL via Axiom's hosted remote MCP server.
Official VictoriaMetrics server: MetricsQL/PromQL queries plus embedded docs search.
Connect AI to Last9 production observability: logs, metrics, traces, exceptions and alerts.
Query distributed traces in Grafana Tempo with TraceQL (archived; Tempo now has embedded MCP).
PagerDuty's official MCP server — agents triage incidents, manage on-call schedules, and query services from your IDE.
Query New Relic via NerdGraph: entity search, NRQL, deployments and infrastructure.
Amplitude's hosted MCP server — query charts, dashboards, experiments, flags, and session replays in plain English.
Debug Kubernetes & container workloads with eBPF kernel telemetry through an AI interface.
Community MCP server for the Mixpanel API — query events, funnels, retention, and user profiles from Claude or Cursor.
Use these in a stack
FAQ
Which observability tools have MCP servers?
Datadog, Grafana, Prometheus, New Relic, Dynatrace, Honeycomb, Sentry and more — pick the one matching your existing stack.
Can an AI actually triage incidents with these?
It can fetch and correlate logs/metrics/traces and summarize, which speeds triage. Keep humans in the loop for remediation actions.