Tracea: Self-hosted monitoring and debugging tool for AI agents with automatic RCA and team memory

Summary

Tracea is a self-hosted monitoring and debugging tool for AI agents, similar to Datadog but purpose-built for agent workflows. It captures all tool calls, LLM responses, and cost data; provides automatic root cause analysis (RCA) for failures; uses YAML-based detection rules to catch errors before production; and features a "Company Brain" that turns session data into reusable team knowledge. It deploys via a single Docker command with data staying entirely on-premises.