-
- {documents.length === 1 ? '1 document' : `${documents.length} documents`}
-
- {' · searched as you'}
+ {incomplete && documents.length === 0 ? (
+ 'Search is incomplete.'
+ ) : (
+ <>
+
+ {documents.length === 1 ? '1 document' : `${documents.length} documents`}
+
+ {' · searched as you'}
+ {incomplete && Some results may be missing.}
+
+ )}
{indexingNote && {indexingNote}}
+ {incomplete && (
+ void refetchSearch()}>
+ {isFetching ? 'Retrying…' : 'Try again'}
+
+ )}
{showFilters && (
@@ -245,11 +257,13 @@ export function KnowledgeSearchResults({
)}
{documents.length === 0 ? (
-
- {filtersActive
- ? 'No documents match these filters.'
- : `No documents you can read match “${query}”.`}
-
+ !incomplete && (
+
+ {filtersActive
+ ? 'No documents match these filters.'
+ : `No documents you can read match “${query}”.`}
+
+ )
) : (
{documents.map((result) => {
diff --git a/apps/sim/hooks/queries/kb/knowledge.test.ts b/apps/sim/hooks/queries/kb/knowledge.test.ts
index 2d408adf15c..71729ec1bde 100644
--- a/apps/sim/hooks/queries/kb/knowledge.test.ts
+++ b/apps/sim/hooks/queries/kb/knowledge.test.ts
@@ -132,12 +132,17 @@ describe('knowledge query placeholder scope', () => {
})
it('forwards search cancellation and leaves provider retries to the server', async () => {
- mocks.requestJson.mockResolvedValueOnce({ data: { results: [] } })
+ const data = {
+ query: 'query',
+ results: [],
+ retrieval: { status: 'partial', timedOutLegs: ['vector'] },
+ }
+ mocks.requestJson.mockResolvedValueOnce({ data })
const query = captureQuery(() =>
useWorkspaceKnowledgeSearch('workspace-1', ' query ', { source: 'slack' })
)
const controller = new AbortController()
- await query.queryFn({ signal: controller.signal })
+ await expect(query.queryFn({ signal: controller.signal })).resolves.toEqual(data)
expect(query.retry).toBe(false)
expect(mocks.requestJson).toHaveBeenLastCalledWith(
diff --git a/apps/sim/hooks/queries/kb/knowledge.ts b/apps/sim/hooks/queries/kb/knowledge.ts
index 5870972170e..bb0a023bcda 100644
--- a/apps/sim/hooks/queries/kb/knowledge.ts
+++ b/apps/sim/hooks/queries/kb/knowledge.ts
@@ -52,7 +52,7 @@ import {
updateKnowledgeDocumentContract,
updateKnowledgeDocumentTagsContract,
type WorkspaceKnowledgeSearchBody,
- type WorkspaceKnowledgeSearchResult,
+ type WorkspaceKnowledgeSearchData,
} from '@/lib/api/contracts/knowledge'
import type { WorkspaceSearchFilters } from '@/lib/api/contracts/knowledge/search'
import type { ChunkingStrategy, StrategyOptions } from '@/lib/chunkers/types'
@@ -1197,9 +1197,9 @@ export function useBulkDeleteKnowledgeBases(workspaceId: string) {
async function searchWorkspaceKnowledge(
body: WorkspaceKnowledgeSearchBody,
signal?: AbortSignal
-): Promise {
+): Promise {
const data = await requestJson(searchWorkspaceKnowledgeContract, { body, signal })
- return data.data.results
+ return data.data
}
/** Searches the canonical index under the signed-in person's ACLs. */
diff --git a/apps/sim/lib/api/contracts/knowledge/search.ts b/apps/sim/lib/api/contracts/knowledge/search.ts
index 63c86baeaf6..912ca4690aa 100644
--- a/apps/sim/lib/api/contracts/knowledge/search.ts
+++ b/apps/sim/lib/api/contracts/knowledge/search.ts
@@ -186,6 +186,16 @@ export const workspaceKnowledgeSearchBodySchema = resourceOwnerSchema.safeExtend
})
export type WorkspaceKnowledgeSearchBody = z.input
+export const workspaceKnowledgeSearchDataSchema = z.object({
+ query: z.string(),
+ results: z.array(workspaceKnowledgeSearchResultSchema),
+ retrieval: z.object({
+ status: z.enum(['complete', 'partial']),
+ timedOutLegs: z.array(z.enum(['vector', 'keyword', 'tags'])).max(3),
+ }),
+})
+export type WorkspaceKnowledgeSearchData = z.output
+
/**
* The search a signed-in person runs from the composer: what their own
* account may read across the workspace's knowledge bases, presented as
@@ -199,10 +209,7 @@ export const searchWorkspaceKnowledgeContract = defineRouteContract({
mode: 'json',
schema: z.object({
success: z.literal(true),
- data: z.object({
- query: z.string(),
- results: z.array(workspaceKnowledgeSearchResultSchema),
- }),
+ data: workspaceKnowledgeSearchDataSchema,
}),
},
})
diff --git a/apps/sim/lib/knowledge/__integration__/search-latency.integration.ts b/apps/sim/lib/knowledge/__integration__/search-latency.integration.ts
index 7f62c21dae8..919cce00ac4 100644
--- a/apps/sim/lib/knowledge/__integration__/search-latency.integration.ts
+++ b/apps/sim/lib/knowledge/__integration__/search-latency.integration.ts
@@ -17,8 +17,11 @@ import {
import { createLogger, Logger } from '@sim/logger'
import { generateId } from '@sim/utils/id'
import { and, eq, inArray, sql } from 'drizzle-orm'
+import { NextRequest } from 'next/server'
import { afterAll, beforeAll, describe, expect, it, type MockInstance, vi } from 'vitest'
import { z } from 'zod'
+import { workspaceKnowledgeSearchDataSchema } from '@/lib/api/contracts/knowledge/search'
+import { internalSessionAuth } from '@/lib/api/server/routes'
import type {
MothershipStreamV1CheckpointPausePayload,
MothershipStreamV1ToolCallDescriptor,
@@ -41,6 +44,7 @@ import {
} from '@/lib/knowledge/search/budget'
import type { SearchStage } from '@/lib/knowledge/search/diagnostics'
import type { WorkspaceSearchFilters } from '@/lib/knowledge/search/filters'
+import { POST as searchRoute } from '@/app/api/knowledge/search/route'
import { ResolvedSecretTraceRegistry } from '@/executor/utils/resolved-secret-trace-registry'
/** Initialize controlled provider configuration before the real application modules load. */
@@ -532,6 +536,60 @@ describe.skipIf(!enabled)('Assistant search latency on a realistic indexed corpu
30_000
)
+ it.each(['vector', 'both'] as const)(
+ 'returns incomplete dashboard coverage when %s SQL branches exceed their deadline',
+ async (delayedLegs) => {
+ const authenticate = vi.spyOn(internalSessionAuth, 'authenticate').mockResolvedValue({
+ kind: 'session',
+ userId: ids.aliceId,
+ sessionId: 'fixture-dashboard',
+ })
+ const query = SearchBudget.prototype.query
+ const delayed = vi.spyOn(SearchBudget.prototype, 'query').mockImplementation(function (
+ this: SearchBudget,
+ stage: SearchStage,
+ run: (executor: SearchExecutor) => PromiseLike
+ ): Promise {
+ return query.call(this, stage, async (tx) => {
+ if (delayedLegs === 'both' || this.leg === 'vector')
+ await tx.execute(sql`SELECT pg_sleep(9)`)
+ return run(tx)
+ }) as Promise
+ })
+ try {
+ const response = await searchRoute(
+ new NextRequest('http://localhost/api/knowledge/search', {
+ method: 'POST',
+ headers: { 'content-type': 'application/json' },
+ body: JSON.stringify({
+ workspaceId: ids.workspaceId,
+ query: 'Orion deployment',
+ topK: 15,
+ }),
+ })
+ )
+ expect(response.status).toBe(200)
+ const data = workspaceKnowledgeSearchDataSchema.parse((await response.json()).data)
+ expect(data.retrieval).toEqual({
+ status: 'partial',
+ timedOutLegs: delayedLegs === 'both' ? ['vector', 'keyword'] : ['vector'],
+ })
+ if (delayedLegs === 'both') expect(data.results).toEqual([])
+ else {
+ expect(data.results.length).toBeGreaterThan(0)
+ expect(
+ data.results.every((result) => result.knowledgeBaseId === ids.knowledgeBaseId)
+ ).toBe(true)
+ }
+ report[`dashboard.deadline.${delayedLegs}`] = { resultCount: data.results.length }
+ } finally {
+ delayed.mockRestore()
+ authenticate.mockRestore()
+ }
+ },
+ 30_000
+ )
+
it('records first and repeated application searches with the actual SQL plans', async () => {
const before = embeddingCalls
for (let iteration = 0; iteration < 2; iteration++) {
diff --git a/apps/sim/lib/knowledge/application/search.ts b/apps/sim/lib/knowledge/application/search.ts
index 8d2e1dc2057..cdf61bfc5fd 100644
--- a/apps/sim/lib/knowledge/application/search.ts
+++ b/apps/sim/lib/knowledge/application/search.ts
@@ -423,12 +423,12 @@ const searchKnowledgeUseCase = defineAuthorizedKnowledgeUseCase({
})
)
- if (retrieved.retrieval.status === 'partial' && !input.allowPartialResults)
- throw new SearchDeadlineError()
annotateSearchDiagnostics({
retrievalStatus: retrieved.retrieval.status,
timedOutLegs: retrieved.retrieval.timedOutLegs,
})
+ if (retrieved.retrieval.status === 'partial' && !input.allowPartialResults)
+ throw new SearchDeadlineError()
let rows = retrieved.rows
input.signal?.throwIfAborted()
/** Public callers have no input envelope, but persisted reranker inputs still need provenance. */
diff --git a/apps/sim/lib/knowledge/search/diagnostics.ts b/apps/sim/lib/knowledge/search/diagnostics.ts
index f4c506ca629..4265db0c886 100644
--- a/apps/sim/lib/knowledge/search/diagnostics.ts
+++ b/apps/sim/lib/knowledge/search/diagnostics.ts
@@ -70,6 +70,7 @@ export interface SearchDiagnosticMetadata {
boostRecency?: boolean
embeddingDimensions?: number
vectorRanking?: 'exact' | 'binary-rerank'
+ vectorCandidateLimit?: number
vectorCandidateCount?: number
resultCount?: number
/** Tool output before the executor's final egress projection; counts only, never content. */
diff --git a/apps/sim/lib/knowledge/search/queries.test.ts b/apps/sim/lib/knowledge/search/queries.test.ts
index 168c64609ed..f73f148ff63 100644
--- a/apps/sim/lib/knowledge/search/queries.test.ts
+++ b/apps/sim/lib/knowledge/search/queries.test.ts
@@ -553,7 +553,10 @@ describe('live repository authorization follows ranked candidates', () => {
expect(dbChainMockFns.limit.mock.invocationCallOrder[1]).toBeLessThan(
dbChainMockFns.select.mock.invocationCallOrder[2]
)
- expect(JSON.stringify(dbChainMockFns.where.mock.calls[1][0])).toContain('OFFSET 0')
+ const visibilityJoin = render(dbChainMockFns.innerJoin.mock.calls[1][0])
+ expect(visibilityJoin.sql).toContain('LATERAL')
+ expect(visibilityJoin.sql).toContain('OFFSET 0')
+ expect(JSON.stringify(dbChainMockFns.innerJoin.mock.calls[1][0])).toContain('required_clause')
expect(getForConnectors).toHaveBeenCalledExactlyOnceWith(['allowed-source'], undefined)
expect(JSON.stringify(dbChainMockFns.where.mock.calls[3][0])).toContain('github_read_grant')
})
diff --git a/apps/sim/lib/knowledge/search/queries.ts b/apps/sim/lib/knowledge/search/queries.ts
index 2e349830dea..af35de09073 100644
--- a/apps/sim/lib/knowledge/search/queries.ts
+++ b/apps/sim/lib/knowledge/search/queries.ts
@@ -793,10 +793,10 @@ export async function handleVectorOnlySearch(params: SearchParams): Promise candidate.id))
}
if (useExactRanking) return exactPage()
+ annotateSearchDiagnostics({
+ vectorRanking: 'binary-rerank',
+ vectorCandidateLimit: candidateLimit,
+ })
const identities = await withVectorScanSettings(
(executor) =>
executor
.select({ id: embedding.id })
.from(embedding)
- .where(
- and(
- inArray(embedding.knowledgeBaseId, params.knowledgeBaseIds),
- sql`EXISTS (
- SELECT 1 FROM ${document}
- WHERE ${and(eq(document.id, embedding.documentId), ...visibility)}
- OFFSET 0
- )`
- )
+ .innerJoin(
+ sql`LATERAL (
+ SELECT 1 FROM ${document}
+ WHERE ${and(eq(document.id, embedding.documentId), ...visibility)}
+ OFFSET 0
+ ) AS visible_document`,
+ sql`true`
)
+ .where(inArray(embedding.knowledgeBaseId, params.knowledgeBaseIds))
.orderBy(candidateDistance)
.limit(candidateLimit),
params.budget,
'binary'
)
annotateSearchDiagnostics({
- vectorRanking: 'binary-rerank',
vectorCandidateCount: identities.length,
})
if (!identities.length) return exactPage()