mastra-ai/mastra · error · Error

Semantic recall requires an embedder to be…

Error message

Semantic recall requires an embedder to be configured.

https://mastra.ai/en/docs/memory/semantic-recall

What it means

Memory's semantic recall feature stores message embeddings in a vector database and retrieves them by similarity, which requires an embedding model (embedder) to convert text into vectors. Memory throws this Error in its constructor when semantic recall is enabled but config.embedder is not provided. It fails fast at construction time rather than at first recall, so misconfiguration is caught immediately.

Solutions

Pass an embedder in the Memory config, e.g. embedder: new FastEmbed() or an AI SDK embedding model via embedMany-compatible wrapper.
If you do not need semantic recall, remove the semanticRecall option from memory options.
Verify the config object shape: semanticRecall requires BOTH vector (vector store) and embedder.

Example fix

// before
const memory = new Memory({
  options: { semanticRecall: { topK: 5 } },
  vector: new PgVector connectionString,
});

// after
const memory = new Memory({
  embedder: new FastEmbed(),
  vector: new PgVector(connectionString),
  options: { semanticRecall: { topK: 5 } },
});

Defensive patterns

Strategy: validation

Validate before calling

function assertEmbedder(config) {
  if (config?.options?.semanticRecall && !config.embedder) {
    throw new Error('semanticRecall requires an embedder in Memory config');
  }
}

Prevention

Always pass embedder alongside vector when enabling semanticRecall.
Centralize Memory construction in one factory function so required fields are not dropped.
Add a unit test that constructs Memory with your production config.

When it happens

Trigger: new Memory({ options: { semanticRecall: true/..., vector: <store> } }) without an embedder property in the config object.

Common situations: Developers add a vector store (e.g. PgVector) for semantic recall but forget to also pass an embedder like fastembed or an AI SDK embedding model; copying old snippets where the embedder defaulted or was optional; switching Memory configs and dropping the embedder field during refactors.

Understand the failure class

Background: "Must pass :limit option" / "Missing required option" — required option errors explained — this error's family across 41 libraries.

Related errors

AI-assisted analysis of mastra-ai/mastra@75dd419e61 (2026-08-30). Data as JSON: /api/errors/26a36099ce6f609f. Report an issue: GitHub.

Appendix: source

Thrown at packages/core/src/memory/memory.ts:181

      );
    }
    if (config.storage) {
      this._storage = augmentWithInit(config.storage);
      this._hasOwnStorage = true;
    }

    if (this.threadConfig.semanticRecall) {
      if (!config.vector) {
        throw new Error(
          `Semantic recall requires a vector store to be configured.

https://mastra.ai/en/docs/memory/semantic-recall`,
        );
      }
      this.vector = config.vector;

      if (!config.embedder) {
        throw new Error(
          `Semantic recall requires an embedder to be configured.

https://mastra.ai/en/docs/memory/semantic-recall`,
        );
      }

      // Convert string embedder to ModelRouterEmbeddingModel
      if (typeof config.embedder === 'string') {
        this.embedder = new ModelRouterEmbeddingModel(config.embedder);
      } else {
        this.embedder = config.embedder;
      }

      // Set embedder options (e.g., providerOptions for Google models)
      if (config.embedderOptions) {
        this.embedderOptions = config.embedderOptions;
      }
    } else {

View on GitHub (pinned to 75dd419e61)