Reflector produced empty output after ${attemptNumber} attem

Error message

Reflector produced empty output after ${attemptNumber} attempt(s)${parsed.degenerate ? ' (degenerate repetition)' : ''} — refusing to commit an empty reflection over ${originalTokens} observation tokens

What it means

After exhausting the reflection ladder, if the model returned empty or degenerate (repetitive) observations for non-empty input, the reflector throws instead of committing. Committing empty output would silently wipe the active observations slice, so this is a hard safety guard.

Source

Thrown at packages/memory/src/processors/observational-memory/reflector-runner.ts:572

          config: this.getObservationMarkerConfig(),
        });
        streamContext.startedAt = startMarker.data.startedAt;
        // Stream OM lifecycle markers as transient so the OutputWriter does not persist standalone data-only messages; OM persists the durable marker explicitly.
        await streamContext.writer.custom({ ...startMarker, transient: true }).catch(() => {});
        await this.persistMarkerToStorage(startMarker, streamContext.threadId, streamContext.resourceId);
      }

      currentLevel = Math.min(currentLevel + 1, maxLevel) as CompressionLevel;
    }

    // A reflection of non-empty observations must never come back empty: the
    // caller commits the result as the new activeObservations (sync path) or
    // as the bufferedReflection replacing the reflected slice (buffered path),
    // so returning '' here silently wipes memory. Empty output only happens
    // when every ladder attempt was degenerate (parseReflectorOutput discards
    // degenerate text) or the model returned nothing — both are failures.
    if (observations.trim().length > 0 && parsed.observations.trim().length === 0) {
      throw new Error(
        `Reflector produced empty output after ${attemptNumber} attempt(s)${parsed.degenerate ? ' (degenerate repetition)' : ''} — refusing to commit an empty reflection over ${originalTokens} observation tokens`,
      );
    }

    const structuredExtraction = await extractStructuredValues({
      agent,
      source: 'reflector',
      extractors: activeExtractors,
      memory: temporaryMemory?.options,
      priorExtractedValues,
      requestContext: internalRequestContext,
      observabilityContext,
      abortSignal,
    });
    const parsedExtractedValues = mergeExtractedValues(parsed.extractedValues, structuredExtraction.values);
    const parsedExtractionFailures = mergeExtractionFailures(parsed.extractionFailures, structuredExtraction.failures);
    const hookedValues = await applyExtractorHooks({
      source: 'reflector',

View on GitHub (pinned to 75dd419e61)

Solutions

Use a stronger model for the reflector (check reflector model config in ObservationalMemory options)
Review the reflector prompt/instructions for token limits or constraints causing empty replies
Reduce the observation batch size so the reflection fits within the model's output budget
Retry the reflection — transient model failures may resolve on a later call

Example fix

// before
new ObservationalMemory({ reflector: { model: 'tiny-model' } })
// after
new ObservationalMemory({ reflector: { model: 'openai/gpt-4o' } })

Defensive patterns

Strategy: retry

Validate before calling

const obs = observations.trim(); if (!obs) return; // nothing to reflect

Type guard

const isEmptyReflection = (out: string) => out.trim().length === 0;

Try / catch

try { result = await reflect(...) } catch (e) { if (String(e).includes('empty output')) { logAndRetryOrSkip(e); } else throw e; }

Prevention

Configure a capable model for the reflector
Keep reflection input within model output budget
Monitor for degenerate repetition and alert
Retry reflections on transient failures

When it happens

Trigger: All retry/ladder attempts of the reflector agent produced whitespace-only or degenerate repetition output while the original observation tokens were non-empty (observations.trim().length > 0 but parsed.observations empty).

Common situations: Misconfigured/low-quality model for the reflector; model hitting token limits and returning nothing; model loops producing degenerate repetition; reflector agent prompt overridden incorrectly.

Related errors

AI-assisted analysis of mastra-ai/mastra@75dd419e61 (2026-08-30). Data as JSON: /api/errors/40130a5d94b57d6e. Report an issue: GitHub.