mastra-ai/mastra · error · Error

No task: provide targetType+targetId or task

Error message

No task: provide targetType+targetId or task

What it means

runExperiment needs a task to run per item: either a registered target (targetType+targetId) or an inline `task` function. If neither is present the config cannot produce an execFn, so this plain Error is thrown during setup and the run is marked failed before execution.

Source

Thrown at packages/core/src/datasets/experiment/index.ts:353

        throw new Error(`Target not found: ${targetType}/${targetId}`);
      }
      const { target } = resolved;
      execFn = (item, itemSignal) => {
        // Merge global request context with per-item request context (item takes precedence)
        const mergedRequestContext =
          globalRequestContext || item.requestContext ? { ...globalRequestContext, ...item.requestContext } : undefined;
        return executeTarget(target, targetType, item, {
          signal: itemSignal,
          requestContext: mergedRequestContext,
          experimentId,
          versions,
          toolMocks: targetType === 'agent' ? item.toolMocks : undefined,
          unmockedToolPolicy:
            targetType === 'agent' ? (item.unmockedToolPolicy ?? config.unmockedToolPolicy ?? 'allow') : undefined,
        });
      };
    } else {
      throw new Error('No task: provide targetType+targetId or task');
    }
  } catch (err) {
    await markFailedOnSetupError(err);
    throw err; // unreachable, but satisfies TS control flow
  }

  // Tool mocks only apply to agent targets. If a dataset carrying toolMocks is reused
  // against a task/workflow/scorer target, the mocks are silently ignored — warn once
  // (not per item) so the misconfiguration is visible without log spam.
  const itemsWithToolMocks = items.filter(item => item.toolMocks?.length).length;
  if (targetType !== 'agent' && itemsWithToolMocks > 0) {
    mastra
      .getLogger()
      ?.warn(
        `Experiment target is "${config.task ? 'task' : targetType}" but ${itemsWithToolMocks} of ${items.length} dataset items declare toolMocks. ` +
          `Tool mocks only apply to agent targets and will be ignored.`,
      );
  }

View on GitHub (pinned to 75dd419e61)

Solutions

1. Provide a `task` function (item, signal) => result for inline execution
2. Or provide both targetType and targetId referencing a registered agent/tool/workflow/scorer
3. Validate the config before calling: ensure (targetType && targetId) || typeof task === 'function'
4. Trace why neither branch was populated (e.g. a flag or lookup returning undefined)

Example fix

// before
await runExperiment({ datasetId: 'ds_123' }); // no task, no target
// after
await runExperiment({
  datasetId: 'ds_123',
  targetType: 'agent',
  targetId: 'my-agent',
});

if (!((config.targetType && config.targetId) || typeof config.task === 'function')) {
  throw new Error('runExperiment requires targetType+targetId or a task function');
}

function hasTask(c: { targetType?: string; targetId?: string; task?: unknown }): boolean {
  return (typeof c.targetType === 'string' && typeof c.targetId === 'string') || typeof c.task === 'function';
}

try {
  await runExperiment(config);
} catch (err) {
  if (err instanceof Error && err.message.includes('No task')) {
    console.error('Experiment config must define targetType+targetId or task');
  } else throw err;
}

When it happens

Trigger: Calling runExperiment (or startExperiment/startExperimentAsync/executeRun) with a config containing neither targetType+targetId nor a task function — e.g. both fields undefined after config assembly.

Common situations: Programmatic config building where task was expected to be injected; copying an example that used datasetId+task but dropping the task; conditional logic that skips both branches (e.g. only setting targetType when some flag is true).

Understand the failure class

Background: "missing required argument" and "the following required arguments were not provided": what required-argument errors mean and how to fix them — this error's family across 20 libraries.

Related errors

• Unknown target type: ${targetType}
• No data source: provide datasetId or data
• AGENT_DURABLE_METHOD_NOT_AVAILABLE
• Unsupported output format: ${format}
• Experiment not found: ${experimentIdA}

AI-assisted analysis of mastra-ai/mastra@75dd419e61 (2026-08-30). Data as JSON: /api/errors/27056009d1f6d815. Report an issue: GitHub.