mastra-ai/mastra · error · MastraError

INVALID_DATA_ITEM

INVALID_DATA_ITEM

Error message

Invalid data item at index ${i}: must have 'input', 'inputs', or 'turns' property

What it means

Each item in an eval dataset must be an object containing at least one of `input`, `inputs`, or `turns`. `validateEvalsInputs` iterates the data array and throws a MastraError (id INVALID_DATA_ITEM, category USER) at the first index that is null/non-object or lacks all three keys. This is a dataset shape contract enforced before running the experiment.

Source

Thrown at packages/core/src/evals/run/index.ts:690

): void {
  const hasGates = !!gates && gates.length > 0;
  if (data.length === 0) {
    throw new MastraError({
      domain: 'SCORER',
      id: 'RUN_EXPERIMENT_FAILED_NO_DATA_PROVIDED',
      category: 'USER',
      text: 'Failed to run experiment: Data array is empty',
    });
  }

  // Tracks whether any data item carries per-turn gates/scorers, which (like
  // top-level scorers/gates) satisfies the "at least one scorer or gate" rule.
  let hasAnyTurnAssertions = false;

  for (let i = 0; i < data.length; i++) {
    const item = data[i];
    if (!item || typeof item !== 'object' || (!('input' in item) && !('inputs' in item) && !('turns' in item))) {
      throw new MastraError({
        domain: 'SCORER',
        id: 'INVALID_DATA_ITEM',
        category: 'USER',
        text: `Invalid data item at index ${i}: must have 'input', 'inputs', or 'turns' property`,
      });
    }
    if ('inputs' in item) {
      if (!Array.isArray(item.inputs) || item.inputs.length === 0) {
        throw new MastraError({
          domain: 'SCORER',
          id: 'INVALID_DATA_ITEM',
          category: 'USER',
          text: `Invalid data item at index ${i}: 'inputs' must be a non-empty array`,
        });
      }
      if (isWorkflow(target)) {
        throw new MastraError({
          domain: 'SCORER',

View on GitHub (pinned to 75dd419e61)

Solutions

Give every data item an `input` key (or `inputs` for multi-input, `turns` for conversations)
Normalize the dataset before calling runEvals: map raw records into `{ input: ... }` shape
Validate/parse the dataset with a schema (zod) before running to catch bad indices early
Check for holes/undefined entries created by `map`/`filter` on sparse arrays

Example fix

// before
const data = [
  { prompt: 'What is 2+2?' }, // wrong key -> INVALID_DATA_ITEM
];
// after
const data = [
  { input: 'What is 2+2?' },
];

Defensive patterns

Strategy: validation

Validate before calling

function isValidDataItem(item) {
  return !!item && typeof item === 'object'
    && ('input' in item || 'inputs' in item || 'turns' in item);
}
data.forEach((item, i) => { if (!isValidDataItem(item)) throw new Error(`Bad eval item at index ${i}`); });

Type guard

function isEvalDataItem(item: unknown): item is { input?: unknown; inputs?: unknown; turns?: unknown } {
  return typeof item === 'object' && item !== null
    && ('input' in item || 'inputs' in item || 'turns' in item);
}
const valid: EvalItem[] = data.filter(isEvalDataItem);

Try / catch

try {
  await mastra.runEvals({ data, scorers, target });
} catch (e) {
  if (e instanceof MastraError && e.id === 'INVALID_DATA_ITEM') {
    console.error(e.message); // names the offending index — fix that record's shape
  } else throw e;
}

Prevention

Standardize dataset items on the `input` key in a shared type/zod schema
Validate the whole dataset with a schema before calling runEvals
When migrating formats, normalize records to the current shape instead of mixing keys

When it happens

Trigger: Calling runEvals with items like `null`, strings/numbers, `{}` (missing keys), or objects using the wrong key such as `{ prompt: ... }` or `{ messages: ... }` — thrown at the first offending index `i`.

Common situations: Hand-written datasets with inconsistent key names; migrating from an older eval format (`inputs` -> `input` or the new `turns` for multi-turn); mapping over an array that produced `undefined` entries; JSON parse producing scalars for some lines of a JSONL file.

Understand the failure class

Background: Schema validation failed / invalid input schema: payload rejected because its shape doesn't match the expected schema — this error's family across 28 libraries.

Related errors

AI-assisted analysis of mastra-ai/mastra@75dd419e61 (2026-08-30). Data as JSON: /api/errors/16743bb41c55f496. Report an issue: GitHub.