mshegolev/jaeger-mcp

**MCP server for Jaeger distributed tracing.**
Give Claude (or any MCP-capable agent) read access to your trace data — search traces, inspect spans, compare traces, compute span statistics, map service dependencies, predict performance issues, and forecast capacity needs — without leaving the conversation.
Why another Jaeger MCP?
The existing Jaeger integrations require a running UI or custom scripts. This server:
- Speaks the standard Model Context Protocol over **stdio** — works with Claude Desktop, Claude Code, Cursor, and any MCP client.
- Is **read-only**: all 15 tools carry readOnlyHint: true — zero risk of modifying trace data.
- Returns **dual-channel output**: structured JSON (structuredContent) for programmatic use + Markdown (content) for human-readable display.
- Has **actionable error messages** that name the exact env var to fix and suggest a next step.
- Supports **Bearer token**, **HTTP Basic auth**, or **no auth** (common for internal deployments).
- Includes **OpenAPI specification** documenting the underlying Jaeger Query API (openapi.yaml).
Tools
| Tool | Endpoint | Description |
|------|----------|-------------|
| jaeger_list_services | GET /api/services | List all instrumented services |
| jaeger_list_operations | GET /api/services/{service}/operations | List operation names for a service |
| jaeger_search_traces | GET /api/traces | Search traces with rich filters |
| jaeger_get_trace | GET /api/traces/{traceID} | Full trace detail with span tree |
| jaeger_get_dependencies | GET /api/dependencies | Service-to-service call graph |
| jaeger_compare_traces | GET /api/traces/{traceID} ×2 | Structural diff between two traces |
| jaeger_span_statistics | GET /api/traces | Per-operation latency and error stats |
| jaeger_critical_path | GET /api/traces/{traceID} | Longest-duration span chain and bottleneck ranking |
| jaeger_compare_windows | GET /api/traces ×2 | Aggregate trace behavior diff between two time periods |
| jaeger_detect_anomalies | GET /api/traces ×2 | Statistical latency/error-rate spike detection per operation |
| jaeger_predict_degradation | GET /api/traces | Predict performance degradation 2-24 hours in advance |
| jaeger_forecast_capacity | GET /api/traces | Forecast throughput demands and resource requirements |
| jaeger_find_test_traces | GET /api/traces | Correlate a test run to its traces by tag query (Allure/pytest/custom) |
| jaeger_regression_diff | GET /api/traces ×2 | Classify per-operation regressions between two time windows |
| jaeger_test_profile | GET /api/traces | Per-operation latency hotspots for a tagged test run |
Installation
Or run directly without installing:
Configuration
All configuration is via environment variables:
| Variable | Required | Default | Description |
|----------|----------|---------|-------------|
| JAEGER_URL | **Yes** | — | Jaeger query service URL, e.g. https://jaeger.example.com |
| JAEGER_TOKEN | No | — | Bearer token (takes precedence over Basic auth) |
| JAEGER_USERNAME | No | — | HTTP Basic auth username |
| JAEGER_PASSWORD | No | — | HTTP Basic auth password |
| JAEGER_SSL_VERIFY | No | true | Set false for self-signed certificates |
| JAEGER_TIMEOUT | No | 30 | HTTP request timeout in seconds |
| JAEGER_RETRY_ATTEMPTS | No | 3 | Retry count for transient failures (0 to disable) |
| JAEGER_CACHE_TTL | No | 120 | TTL in seconds for discovery endpoint cache (0 to disable) |
Copy .env.example to .env and fill in your values.
Claude Desktop / Claude Code setup
Add to your MCP config (claude_desktop_config.json or .claude/mcp.json):
Or with uvx (no install required):
Docker
Example queries
Once configured, ask Claude:
- "What services does Jaeger know about?"
- "Find traces with HTTP 500 errors in order-service from the last hour"