Blog

Six metrics for judging AI agent traces in production: tool failures, hallucination, logic errors, task completion, safety, and a custom metric you define.

traceroot and Langfuse both trace your agent. traceroot is agent-native: a CLI your coding agent can drive, plus Signals and root cause analysis to improve it.

A LangGraph SQL agent got a $200 MRR drop right and its cause wrong. Follow the trace to the reporting view that shaped its answer.

An answer that crosses no alert threshold. A traceroot Detector, an LLM-as-a-judge check written in plain English, catches it in production and opens a Signal.

A LangGraph SQL agent misread a revenue drop. Follow the Signal to a faulty subscription join, fix the query, and rerun the agent.

An agent evaluation tutorial in traceroot: both runs report the same total, and four code checks on one saved case tell the faulty and corrected runs apart.

A tool call succeeds, but the agent gives the wrong answer. See how traceroot Detectors judge the trace and group repeat failures into Signals.

pi is an open-source agent runtime. Here is how to trace its sessions, turns and tool calls, and how we run our own production agents on pi.

What 1,085 Reddit comments reveal about finding failed agent runs, reconstructing context, and turning production mistakes into regression tests.

The agent in the traceroot app reads your traces and your code at the production commit, and comes back with a change instead of a chart.

Connect TypeSafe AI's Jev to traceroot detectors to screen agent traces and inspect the decisions behind each finding.

The traceroot CLI puts your traces where the work already happens: in the coding agent's terminal, not a browser tab.

Use Pi’s four built-in tools to fix a failing TypeScript test, then customize the coding agent with project instructions and extensions.