mastra-ai/mastra · error · HTTPException

Workspace does not have search configured

Error message

Workspace does not have search configured

What it means

The index route (POST /workspaces/:workspaceId/index) throws this 400 when the target workspace has neither BM25 nor vector search enabled (workspace.canBM25 and workspace.canVector are both false). Content can only be indexed if the workspace was configured with at least one search backend.

Source

Thrown at packages/server/src/server/handlers/workspace.ts:852

  summary: 'Index content for search',
  description: 'Indexes content for later search operations',
  tags: ['Workspace'],
  handler: async ({ mastra, path, content, metadata, workspaceId }) => {
    try {
      requireWorkspaceV1Support();

      if (!path || content === undefined) {
        throw new HTTPException(400, { message: 'Path and content are required' });
      }

      const workspace = await getWorkspaceById(mastra, workspaceId);
      if (!workspace) {
        throw new HTTPException(404, { message: 'No workspace configured' });
      }

      const canSearch = workspace.canBM25 || workspace.canVector;
      if (!canSearch) {
        throw new HTTPException(400, { message: 'Workspace does not have search configured' });
      }

      await workspace.index(path, content, { metadata });

      return {
        success: true,
        path,
      };
    } catch (error) {
      return handleWorkspaceError(error, 'Error indexing content');
    }
  },
});

// =============================================================================
// Skills Routes (under /workspaces/:workspaceId/skills)
// =============================================================================

View on GitHub (pinned to 75dd419e61)

Solutions

1. Configure the workspace with BM25 and/or vector search enabled (per workspace options) before indexing.
2. If search is not needed, don't call the index endpoint.
3. Verify server-side why canBM25/canVector are false (missing index path, missing embedder/vector store config, env vars).
4. Restart/redeploy the server after adding search configuration.

Example fix

// before
new Mastra({ workspaces: { ws1: createWorkspace({ filesystem: fs }) } });
// after
new Mastra({ workspaces: { ws1: createWorkspace({ filesystem: fs, search: { bm25: { enabled: true } } }) } });

// Capability check via the search endpoint's configured workspace
const list = await fetch(`/api/workspaces/${wsId}/search?query=probe`).then(r => r.json());
// or check server config exposes canBM25/canVector before indexing

function supportsSearch(ws: { canBM25?: boolean; canVector?: boolean }): boolean {
  return Boolean(ws.canBM25 || ws.canVector);
}

try {
  await indexContent(wsId, path, content);
} catch (e) {
  if (isHTTPException(e, 400) && /search configured/.test(e.message)) {
    // skip indexing or enable BM25/vector in workspace config and redeploy
  } else throw e;
}

When it happens

Trigger: POSTing content to /workspaces/:id/index on a workspace constructed without a BM25 index directory or vector store; indexing before search backends are initialized; calling index on a filesystem-only workspace.

Common situations: Deploying a workspace config where search options (BM25/vector) were omitted; assuming default search is enabled; environment lacks the index directory or vector DB config so the workspace reports no search capability; stale client docs assuming old defaults.

Related errors

• Workspace filesystem not available
• Skill not found: ${invocation.skillName}.
• Project path is required
• A session workspace must be a valid Workspace instance.
• MASTRA_ADD_WORKSPACE_MISSING_AGENT_METADATA

AI-assisted analysis of mastra-ai/mastra@75dd419e61 (2026-08-30). Data as JSON: /api/errors/4ecd888b8b3ae9b8. Report an issue: GitHub.