fix(search): validate Notion search terms before dispatch (#8358)

Author: waleedlatif1

apps/sim/lib/api/contracts/mothership-assistant-tools.ts

@@ -4,6 +4,9 @@ import { LIVE_SEARCH_PROVIDER_IDS } from '@/lib/sim-search/live/provider-catalog
 export const liveSearchProviderSchema = z.enum(LIVE_SEARCH_PROVIDER_IDS)
 export type LiveSearchProvider = z.output<typeof liveSearchProviderSchema>
 
+export const NOTION_SEARCH_TERMS_REQUIRED =
+  'Notion requires search terms. Add keywords or a concise question.'
+
 /**
  * Native queries one call may send to the same provider account. Alternatives run as separate
  * provider searches and fuse into one ranking, so the bound keeps a call within the provider's
@@ -42,6 +45,14 @@ export const nativeSearchQuerySchema = z
     keywordOnly: z.boolean().optional(),
   })
   .strict()
+  .superRefine((input, context) => {
+    if (input.provider === 'notion' && !input.query)
+      context.addIssue({
+        code: 'custom',
+        path: ['query'],
+        message: NOTION_SEARCH_TERMS_REQUIRED,
+      })
+  })
 export type NativeSearchQuery = z.output<typeof nativeSearchQuerySchema>
 
 export const nativeSearchQueriesSchema = z
@@ -161,15 +172,15 @@ export const searchWorkspaceInputSchema = workspaceSearchFiltersSchema
     nativeQueries: nativeSearchQueriesSchema
       .optional()
       .describe(
-        `Live search only: queries in a provider's own language (Drive q, Gmail operators, JQL, CQL, GitHub qualifiers, Slack RTS, plain Linear/Fireflies terms, Granola natural-language questions, Notion keywords or AI questions when available). Up to ${MAX_NATIVE_QUERIES_PER_ACCOUNT} per account run separately and merge; GitHub and GitLab take one per kind. Write them from the returned live guidance and account IDs; each account status names the queryIndex its cursor belongs to. Omit for simple cross-provider terms.`
+        `Live search only: queries in a provider's own language (Drive q, Gmail operators, JQL, CQL, GitHub qualifiers, Slack RTS, plain Linear/Fireflies terms, Granola natural-language questions, Notion keywords or AI questions when available). Notion requires nonempty search terms even with dates or sorting. Up to ${MAX_NATIVE_QUERIES_PER_ACCOUNT} per account run separately and merge; GitHub and GitLab take one per kind. Write them from the returned live guidance and account IDs; each account status names the queryIndex its cursor belongs to. Omit for simple cross-provider terms.`
       ),
     query: z
       .string()
       .trim()
       .max(2000)
       .default('')
       .describe(
-        'Search terms, without dates already supplied as filters. May be empty for a live listing with a date bound or sortBy newest or oldest.'
+        'Search terms, without dates already supplied as filters. May be empty for a live listing with a date bound or sortBy newest or oldest where supported; Notion requires search terms.'
       ),
     topK: z
       .number()

apps/sim/lib/mothership/generated/sim-assistant-tools.generated.ts

@@ -20,6 +20,9 @@ export const liveSearchProviderSchema = z.enum([
 ])
 export type LiveSearchProvider = z.output<typeof liveSearchProviderSchema>
 
+export const NOTION_SEARCH_TERMS_REQUIRED =
+  'Notion requires search terms. Add keywords or a concise question.'
+
 /**
  * Native queries one call may send to the same provider account. Alternatives run as separate
  * provider searches and fuse into one ranking, so the bound keeps a call within the provider's
@@ -58,6 +61,14 @@ export const nativeSearchQuerySchema = z
     keywordOnly: z.boolean().optional(),
   })
   .strict()
+  .superRefine((input, context) => {
+    if (input.provider === 'notion' && !input.query)
+      context.addIssue({
+        code: 'custom',
+        path: ['query'],
+        message: NOTION_SEARCH_TERMS_REQUIRED,
+      })
+  })
 export type NativeSearchQuery = z.output<typeof nativeSearchQuerySchema>
 
 export const nativeSearchQueriesSchema = z
@@ -177,15 +188,15 @@ export const searchWorkspaceInputSchema = workspaceSearchFiltersSchema
     nativeQueries: nativeSearchQueriesSchema
       .optional()
       .describe(
-        `Live search only: queries in a provider's own language (Drive q, Gmail operators, JQL, CQL, GitHub qualifiers, Slack RTS, plain Linear/Fireflies terms, Granola natural-language questions, Notion keywords or AI questions when available). Up to ${MAX_NATIVE_QUERIES_PER_ACCOUNT} per account run separately and merge; GitHub and GitLab take one per kind. Write them from the returned live guidance and account IDs; each account status names the queryIndex its cursor belongs to. Omit for simple cross-provider terms.`
+        `Live search only: queries in a provider's own language (Drive q, Gmail operators, JQL, CQL, GitHub qualifiers, Slack RTS, plain Linear/Fireflies terms, Granola natural-language questions, Notion keywords or AI questions when available). Notion requires nonempty search terms even with dates or sorting. Up to ${MAX_NATIVE_QUERIES_PER_ACCOUNT} per account run separately and merge; GitHub and GitLab take one per kind. Write them from the returned live guidance and account IDs; each account status names the queryIndex its cursor belongs to. Omit for simple cross-provider terms.`
       ),
     query: z
       .string()
       .trim()
       .max(2000)
       .default('')
       .describe(
-        'Search terms, without dates already supplied as filters. May be empty for a live listing with a date bound or sortBy newest or oldest.'
+        'Search terms, without dates already supplied as filters. May be empty for a live listing with a date bound or sortBy newest or oldest where supported; Notion requires search terms.'
       ),
     topK: z
       .number()

apps/sim/lib/mothership/generated/tool-catalog-v1.ts

@@ -6096,7 +6096,7 @@ export const SearchWorkspace: ToolCatalogEntry = {
       },
       nativeQueries: {
         description:
-          "Live search only: queries in a provider's own language (Drive q, Gmail operators, JQL, CQL, GitHub qualifiers, Slack RTS, plain Linear/Fireflies terms, Granola natural-language questions, Notion keywords or AI questions when available). Up to 4 per account run separately and merge; GitHub and GitLab take one per kind. Write them from the returned live guidance and account IDs; each account status names the queryIndex its cursor belongs to. Omit for simple cross-provider terms.",
+          "Live search only: queries in a provider's own language (Drive q, Gmail operators, JQL, CQL, GitHub qualifiers, Slack RTS, plain Linear/Fireflies terms, Granola natural-language questions, Notion keywords or AI questions when available). Notion requires nonempty search terms even with dates or sorting. Up to 4 per account run separately and merge; GitHub and GitLab take one per kind. Write them from the returned live guidance and account IDs; each account status names the queryIndex its cursor belongs to. Omit for simple cross-provider terms.",
         minItems: 1,
         maxItems: 9,
         type: 'array',
@@ -6140,7 +6140,7 @@ export const SearchWorkspace: ToolCatalogEntry = {
       query: {
         default: '',
         description:
-          'Search terms, without dates already supplied as filters. May be empty for a live listing with a date bound or sortBy newest or oldest.',
+          'Search terms, without dates already supplied as filters. May be empty for a live listing with a date bound or sortBy newest or oldest where supported; Notion requires search terms.',
         type: 'string',
         maxLength: 2000,
       },

apps/sim/lib/mothership/generated/tool-schemas-v1.ts

@@ -6044,7 +6044,7 @@ export const TOOL_RUNTIME_SCHEMAS: Record<string, ToolRuntimeSchemaEntry> = {
         },
         nativeQueries: {
           description:
-            "Live search only: queries in a provider's own language (Drive q, Gmail operators, JQL, CQL, GitHub qualifiers, Slack RTS, plain Linear/Fireflies terms, Granola natural-language questions, Notion keywords or AI questions when available). Up to 4 per account run separately and merge; GitHub and GitLab take one per kind. Write them from the returned live guidance and account IDs; each account status names the queryIndex its cursor belongs to. Omit for simple cross-provider terms.",
+            "Live search only: queries in a provider's own language (Drive q, Gmail operators, JQL, CQL, GitHub qualifiers, Slack RTS, plain Linear/Fireflies terms, Granola natural-language questions, Notion keywords or AI questions when available). Notion requires nonempty search terms even with dates or sorting. Up to 4 per account run separately and merge; GitHub and GitLab take one per kind. Write them from the returned live guidance and account IDs; each account status names the queryIndex its cursor belongs to. Omit for simple cross-provider terms.",
           minItems: 1,
           maxItems: 9,
           type: 'array',
@@ -6114,7 +6114,7 @@ export const TOOL_RUNTIME_SCHEMAS: Record<string, ToolRuntimeSchemaEntry> = {
         query: {
           default: '',
           description:
-            'Search terms, without dates already supplied as filters. May be empty for a live listing with a date bound or sortBy newest or oldest.',
+            'Search terms, without dates already supplied as filters. May be empty for a live listing with a date bound or sortBy newest or oldest where supported; Notion requires search terms.',
           type: 'string',
           maxLength: 2000,
         },

apps/sim/lib/mothership/tools/server/knowledge/workspace-search.test.ts

@@ -90,6 +90,25 @@ describe('Assistant retrieval tools', () => {
       next: null,
     })
   })
+  it.each([{ startDate: '2026-09-01T00:00:00Z' }, { sortBy: 'newest' }, { sortBy: 'oldest' }])(
+    'returns actionable validation for empty Notion native queries with %j',
+    async (bound) => {
+      setEnvFlags({ isLiveEnterpriseSearchEnabled: true })
+      const result = await searchWorkspaceServerTool.execute(
+        {
+          ...bound,
+          query: 'fallback terms',
+          nativeQueries: [{ provider: 'notion', query: ' \t ' }],
+        },
+        { ...context, assistantSearch: undefined }
+      )
+      expect(result).toMatchObject({
+        success: false,
+        message: 'Notion requires search terms. Add keywords or a concise question.',
+      })
+      expect(result).not.toHaveProperty('data')
+    }
+  )
   it('returns a safe permanent configuration failure instead of empty results or opaque error', async () => {
     mocks.search.mockRejectedValue(new EmbeddingConfigurationError())
     const result = await searchWorkspaceServerTool.execute({ query: 'policy' }, context)

apps/sim/lib/mothership/tools/server/knowledge/workspace-search.ts

@@ -4,6 +4,7 @@ import {
   readDocumentInputSchema,
   searchWorkspaceInputSchema,
 } from '@/lib/api/contracts/mothership-assistant-tools'
+import { getValidationErrorMessage } from '@/lib/api/server/validation'
 import { getBaseUrl } from '@/lib/core/utils/urls'
 import { EmbeddingConfigurationError } from '@/lib/embeddings/configuration-error'
 import { sourceAuthor } from '@/lib/knowledge/search/author'
@@ -211,7 +212,7 @@ export const searchWorkspaceServerTool: BaseServerTool = {
               error instanceof SearchDeadlineError
                 ? error.message
                 : error instanceof z.ZodError
-                  ? 'Invalid search arguments'
+                  ? getValidationErrorMessage(error, 'Invalid search arguments')
                   : messageForCopilotKnowledgeError(error),
           }
         }

apps/sim/lib/sim-search/live/application.test.ts

@@ -124,6 +124,33 @@ describe('authorized live retrieval', () => {
     })
     mocks.adminVerify.mockResolvedValue(true)
   })
+  it.each([
+    { startDate: '2026-08-01T00:00:00Z' },
+    { source: ' notion ', startDate: '2026-08-01T00:00:00Z' },
+    { sortBy: 'newest' as const },
+    { sortBy: 'oldest' as const },
+  ])('rejects a Notion-only live listing with %j', async (bound) => {
+    await expect(
+      searchLiveKnowledge.execute({
+        principal,
+        input: { ...input, query: ' \t ', filters: { source: 'notion', ...bound } },
+      })
+    ).rejects.toMatchObject({
+      code: 'validation',
+      message: 'Notion requires search terms. Add keywords or a concise question.',
+    })
+  })
+  it.each([undefined, 'google_drive'])(
+    'preserves date-only live results with source %s',
+    async (source) => {
+      const result = await searchLiveKnowledge.execute({
+        principal,
+        input: { ...input, query: '', filters: { source, startDate: '2026-08-01T00:00:00Z' } },
+      })
+      expect(result.results.map((row) => decodeLiveReference(row.documentId).id)).toEqual(['doc'])
+      expect(result.live?.accounts[0]?.status).toBe('ok')
+    }
+  )
   it('filters admin-token GitLab results through the reader ACL before projection', async () => {
     const gitlab = {
       ...account,

apps/sim/lib/sim-search/live/application.ts

@@ -10,6 +10,7 @@ import {
   type LiveSearchAccountStatus,
   liveSearchProviderSchema,
   type NativeSearchQuery,
+  NOTION_SEARCH_TERMS_REQUIRED,
   nativeSearchQueriesSchema,
   workspaceSearchFiltersSchema,
 } from '@/lib/api/contracts/mothership-assistant-tools'
@@ -312,6 +313,8 @@ export const searchLiveKnowledge = defineAuthorizedKnowledgeUseCase({
     if (input.filters)
       input = { ...input, filters: workspaceSearchFiltersSchema.parse(input.filters) }
     const filters = input.filters
+    if (!queries && filters?.source === 'notion' && !input.query.trim())
+      throw new OrchestrationError('validation', NOTION_SEARCH_TERMS_REQUIRED)
     if (
       filters?.startDate &&
       filters.endDate &&

apps/sim/lib/sim-search/live/providers.ts

@@ -216,7 +216,7 @@ export const LIVE_SEARCH_PROVIDERS = {
     transport: 'managed_mcp',
     guide: {
       syntax:
-        'Natural-language or plain keyword content search through Notion MCP. Availability depends on the connected account and plan; results are restricted to Notion pages, excluding connected apps.',
+        'Natural-language or plain keyword content search through Notion MCP. Search terms are required even with dates or sorting. Availability depends on the connected account and plan; results are restricted to Notion pages, excluding connected apps.',
       scope:
         'project optionally takes a known Notion page URL when the advertised tool supports page scoping. Dates use explicit last-edited timestamps; results without those timestamps cannot satisfy date filters. Read a result for page content.',
       example: 'deployment rollback checklist',