Tempo MCP Server Resources Reference

The TalkOps Tempo MCP Server exposes 11 read-only resources (tempo:// URIs). These endpoints provide dynamic cluster states, operational incident runbooks, and static TraceQL language cheat sheets.

---

Dynamic System Resources

Resource URI	Description	Example Target
tempo://system/backends	List of all configured Tempo backends and health states.	Backend health JSON
tempo://deployment/overview	Cluster deployment topology, multi-tenancy mode, and active querier endpoints.	Monolithic vs. Microservices state

---

Operational Incident Runbooks

Resource URI	Description
tempo://runbooks/latency-spike	Step-by-step diagnostic workflow for tracking down latency regressions and critical-path bottlenecks.
tempo://runbooks/error-burst	Investigation procedures for sudden error rate increases across microservice boundaries.
tempo://runbooks/no-traces-found	Triage steps when expected traces are missing (ingestion lag, drop rules, sampler conflicts).
tempo://runbooks/cross-tenant-access	Guide to cross-tenant querying and tenant header formatting.

---

Static Language References

Resource URI	Description
tempo://reference/traceql	Comprehensive TraceQL syntax reference, intrinsic attributes, and structural queries.
tempo://reference/traceql-metrics	Metrics functions (rate, count_over_time, quantile_over_time) and aggregations.
tempo://reference/k8s-attributes	Standard OpenTelemetry semantic conventions for Kubernetes environments.
tempo://examples/common-queries	Curated library of real-world TraceQL search expressions.

---

Resource Usage Examples

1. Reading Backend Diagnostics (tempo://system/backends)

{
  "uri": "tempo://system/backends",
  "mimeType": "application/json",
  "text": [
    {
      "id": "default",
      "display_name": "Production Tempo",
      "url": "http://tempo.monitoring:3200",
      "status": "ready",
      "version": "2.9.0",
      "multi_tenant": true,
      "default_tenant": "production"
    }
  ]
}

---

2. Loading Operational Runbook (tempo://runbooks/latency-spike)

AI assistants load this runbook during live performance investigations to follow standardized SRE procedures:

# Runbook: Latency Spike Diagnostic Workflow
1. Calculate P95 and P99 latency percentiles using TraceQL metrics.
2. Search for candidate traces exceeding the SLO threshold with tempo_traceql_search.
3. Perform critical-path extraction using tempo_summarize_trace.
4. Distinguish database row lock delays from async timers using time-gap analysis.
5. Generate Prometheus alerting rules with tempo_generate_alerting_expression.