Embedder returned no vector for knowledge search query.

Error message

Embedder returned no vector for knowledge search query.

What it means

`search` in packages/memory/src/processors/observational-memory/subconscious/semantic-index.ts:73 embeds the knowledge-search query via `this.#embedder.doEmbed`. The library throws when the embedder resolves successfully but returns an empty or missing `result.embeddings[0]`, since a query vector is mandatory to search the vector index.

Source

Thrown at packages/memory/src/processors/observational-memory/subconscious/semantic-index.ts:73

  async drain(scope?: KnowledgeScope): Promise<number> {
    const key = scope?.join('\u001f') ?? '*';
    const active = this.#draining.get(key);
    if (active) return active;
    const draining = this.#drain(scope).finally(() => {
      this.#draining.delete(key);
    });
    this.#draining.set(key, draining);
    return draining;
  }

  async search(query: string, scope: KnowledgeScope, limit = 10) {
    await this.drain(scope);
    const result = await this.#embedder.doEmbed({
      values: [query],
      ...(this.#embedderOptions ?? {}),
    } as never);
    const embedding = result.embeddings[0];
    if (!embedding?.length) throw new Error('Embedder returned no vector for knowledge search query.');

    const indexName = this.#indexName(embedding.length);
    if (!(await this.#knowledgeIndexes()).includes(indexName)) {
      throw new StaleKnowledgeSemanticIndexError(
        `Knowledge semantic index ${indexName} is unavailable. Capture or index knowledge before searching.`,
      );
    }

    const visibleScopeKeys = scope.map((_, index) => scope.slice(0, index + 1).join('\u001f'));
    const batches = await Promise.all(
      visibleScopeKeys.map(scopeKey =>
        this.#vector.query({
          indexName,
          queryVector: embedding,
          topK: limit,
          filter: { scope_key: scopeKey },
        }),
      ),

View on GitHub (pinned to 75dd419e61)

Solutions

Fix the embedder so it returns `embeddings: number[][]` with at least one non-empty vector per input value.
If using a mock/stub, update it to return a realistic vector (e.g. `[[0.1, 0.2, ...]]`).
Verify embedder credentials/model configuration; switch to a known-good embedder implementation.
Log raw doEmbed results to confirm the provider actually returns vectors for your query text.

Example fix

// before (broken stub)
const fakeEmbedder = { doEmbed: async () => ({ embeddings: [] }) };
// after
const fakeEmbedder = { doEmbed: async ({ values }) => ({ embeddings: values.map(() => Array(1536).fill(0.1)) }) };

Defensive patterns

Strategy: try-catch

Type guard

function hasEmbedding(result) {
  return Array.isArray(result?.embeddings) && Array.isArray(result.embeddings[0]) && result.embeddings[0].length > 0;
}

Try / catch

try {
  return await knowledgeSearch(scope, query);
} catch (e) {
  if (e.message === 'Embedder returned no vector for knowledge search query.') {
    // fall back to lexical-only search or surface embedder misconfiguration
    return lexicalFallbackSearch(scope, query);
  }
  throw e;
}

Prevention

Unit-test that your embedder wrapper returns { embeddings: number[][] } with matching length.
Validate embedder credentials and model IDs at startup with a smoke embed call.
Never let embedder adapters swallow API errors into empty results.

When it happens

Trigger: `doEmbed({ values: [query] })` resolves with `{ embeddings: [] }` or an empty first vector — a custom/mock embedder that never populates embeddings, an adapter that swallows API failures and resolves with empty results, or provider-side filtering returning no embedding for the query text.

Common situations: Misconfigured custom embedder wrappers; revoked API keys where the SDK resolves empty instead of rejecting; test doubles returning the wrong shape; provider content filtering of the query.

Related errors

AI-assisted analysis of mastra-ai/mastra@75dd419e61 (2026-08-30). Data as JSON: /api/errors/f267fb83083421fa. Report an issue: GitHub.