mastra-ai/mastra · error · Error

Root agent span has no output

Error message

Root agent span has no output

What it means

transformTraceToScorerInputAndOutput throws this when the root agent_run span exists but has no `output`. The scorer output is derived directly from the root span's output payload, so scoring cannot proceed without it.

Source

Thrown at packages/core/src/evals/scoreTraces/utils.ts:274

  // Find the root agent run span
  const rootAgentSpan = spanTree.rootSpans.find(span => span.spanType === 'agent_run') as SpanRecord | undefined;

  if (!rootAgentSpan) {
    throw new Error('No root agent_run span found in trace');
  }

  return { spanTree, rootAgentSpan };
}

export function transformTraceToScorerInputAndOutput(trace: TraceRecord): {
  input: ScorerRunInputForAgent;
  output: ScorerRunOutputForAgent;
} {
  const { spanTree, rootAgentSpan } = prepareTraceForTransformation(trace);

  if (!rootAgentSpan.output) {
    throw new Error('Root agent span has no output');
  }

  // Build input
  const primaryLLMSpan = findPrimaryLLMSpan(spanTree, rootAgentSpan);
  const inputMessages = extractInputMessages(rootAgentSpan);
  const systemMessages = extractSystemMessages(primaryLLMSpan);

  // Extract remembered messages from LLM span (excluding current input)
  const currentInputContent = inputMessages[0]?.content.content || '';
  const rememberedMessages = extractRememberedMessages(primaryLLMSpan, currentInputContent);

  const input = {
    inputMessages,
    rememberedMessages,
    systemMessages,
    taggedSystemMessages: {}, // Todo: Support tagged system messages
  };

View on GitHub (pinned to 75dd419e61)

Solutions

1. Only trigger scoring after the agent run completes successfully and output is persisted
2. Check the storage adapter persists span output payloads (including large ones)
3. Re-run the generation and score the new trace

Example fix

// before
await scoreTrace({ trace: runningTrace });
// after
await agentRunPromise; // wait for completion with output
await scoreTrace({ trace: await getTrace(traceId) });

const { rootAgentSpan } = prepareTraceForTransformation(trace);
if (!rootAgentSpan.output) throw new Error('Agent run incomplete: no output to score');

function hasRootOutput(t: TraceRecord): boolean {
  const root = buildSpanTree(t.spans).rootSpans.find(s => s.spanType === 'agent_run');
  return !!root && root.output != null;
}

try {
  const { output } = transformTraceToScorerInputAndOutput(trace);
} catch (e) {
  if (e instanceof Error && e.message === 'Root agent span has no output') {
    logger.warn('Skipping scoring: agent run produced no output', { traceId: trace.traceId });
  } else throw e;
}

When it happens

Trigger: Scoring a trace whose root agent span has output === null/undefined — e.g. the agent run was cancelled or errored before producing output, output persistence failed, or spans were trimmed before the run finished.

Common situations: Scoring aborted/failed agent runs; storage adapters dropping large output payloads; running scorers against in-flight traces before output is written.

Related errors

• Trace is null or undefined
• Trace must have a spans array
• Trace has no spans
• No model generation span found in trace
• Span ${span.spanId} references non-existent parent ${span.pa

AI-assisted analysis of mastra-ai/mastra@75dd419e61 (2026-08-30). Data as JSON: /api/errors/65a3e227844d69a4. Report an issue: GitHub.