agentevals

Evaluate and score agent behavior against a golden reference. Use this skill whenever the user wants to run evaluation, check pass/fail status, understand metric scores, compare sessions for regressions, validate agent behavior, or score a trace from a file or a live session. Trigger on phrases like "eval this trace", "check my agent output", "did my agent do the right thing", "compare runs", "did my agent regress", "score session X", "evaluate against golden", "run evals". Works with both local trace files and live streaming sessions.

Inspect and debug live streaming agent sessions to understand what the agent did. Use this skill when the user wants to see what happened in a recent agent run, trace through the tool calls step by step, debug unexpected agent behavior, or get a readable narrative of a session. Trigger on phrases like "show me what my agent did", "inspect session", "what happened in that run", "debug my agent", "trace through session", "walk me through the last run", "what tools did it call". Requires `agentevals serve --dev` to be running. For scoring/pass-fail evaluation, use /eval instead.