fix(tests): keep test sources out of file tabs, refresh after runs, reopen tests

- File-edit tool results mark a non-tab file `fileTab: false`, and the browser
  skips promoting it.
- Idle test pages poll every 15s so runs started elsewhere appear; a Mothership
  run returns its tests as resource changes.
- open_resource accepts test resources through an authorized read.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01EHGBgHrtePpi7KMrEfy41R

Author: TheodoreSpeaks

apps/sim/app/workspace/[workspaceId]/home/hooks/stream/handle-tool-event.ts

@@ -1,5 +1,6 @@
 import { isCurrentBrowserToolName } from '@sim/browser-protocol'
 import { isTerminalToolName } from '@sim/terminal-protocol'
+import { toRecordOrNull } from '@sim/utils/object'
 import { isDesktopApp } from '@/lib/desktop'
 import {
   MothershipStreamV1ToolPhase,
@@ -97,12 +98,12 @@ function runToolResultSideEffects(ctx: StreamLoopContext, node: ToolNode, replay
     invalidateResourceQueries(deps.queryClient, deps.workspaceId, resource.type, resource.id)
   }
 
-  if (!replay && (name === ApplyFileEdit.id || name === PrepareFileEdit.id) && isSuccess) {
-    const out = output as Record<string, unknown> | undefined
-    const editData =
-      out && typeof out.data === 'object' && out.data !== null
-        ? (out.data as Record<string, unknown>)
-        : undefined
+  const isFileEdit = name === ApplyFileEdit.id || name === PrepareFileEdit.id
+  const editData = isFileEdit ? toRecordOrNull(toRecordOrNull(output)?.data) : null
+  /** A file another resource owns (a test's source) never opens as a file tab. */
+  const editedFileIsTab = editData?.fileTab !== false
+
+  if (!replay && isFileEdit && isSuccess && editedFileIsTab) {
     const editedFileId =
       (typeof editData?.id === 'string' ? editData.id : undefined) ??
       deps.previewSessionRef.current?.fileId
@@ -139,7 +140,9 @@ function runToolResultSideEffects(ctx: StreamLoopContext, node: ToolNode, replay
     if (name === PrepareFileEdit.id) {
       deps.removePreviewSessionImmediate(node.id)
     }
-    const fileResource = extractedResources.find((r) => r.type === 'file')
+    const fileResource = editedFileIsTab
+      ? extractedResources.find((r) => r.type === 'file')
+      : undefined
     if (fileResource) {
       deps.promoteFileResource(fileResource.id, fileResource.title)
       deps.onResourceEventRef.current?.(

apps/sim/hooks/queries/workflow-tests.ts

@@ -11,8 +11,10 @@ import {
 export type WorkflowTestVersion = 'draft' | 'deployed'
 
 const WORKFLOW_TEST_STALE_TIME = 15_000
-/** While a run is in flight; 20 reads a minute stays inside the routes' per-user limit of 30. */
+/** While a run is in flight, so its progress and result appear as they land. */
 const WORKFLOW_TEST_RUNNING_REFETCH_INTERVAL = 3_000
+/** Otherwise, so a run started elsewhere (Sim, another person, the API) shows up. */
+const WORKFLOW_TEST_IDLE_REFETCH_INTERVAL = 15_000
 
 export const workflowTestKeys = {
   all: ['workflow-tests'] as const,
@@ -43,7 +45,7 @@ export function useWorkflowTests(workspaceId: string, options?: { enabled?: bool
     refetchInterval: (query) =>
       query.state.data?.tests.some((test) => test.status === 'running')
         ? WORKFLOW_TEST_RUNNING_REFETCH_INTERVAL
-        : false,
+        : WORKFLOW_TEST_IDLE_REFETCH_INTERVAL,
   })
 }
 
@@ -62,7 +64,7 @@ export function useWorkflowTest(workspaceId: string, name: string) {
     refetchInterval: (query) =>
       query.state.data?.latestRun?.status === 'running'
         ? WORKFLOW_TEST_RUNNING_REFETCH_INTERVAL
-        : false,
+        : WORKFLOW_TEST_IDLE_REFETCH_INTERVAL,
   })
 }
 

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

@@ -10,7 +10,7 @@ export const openResourceInputSchema = z.strictObject({
   resources: z
     .array(
       z.strictObject({
-        type: z.enum(['workflow', 'table', 'knowledgebase', 'file', 'dashboard', 'log']),
+        type: z.enum(['workflow', 'table', 'knowledgebase', 'file', 'dashboard', 'log', 'test']),
         id: z.string().trim().min(1),
         viewId: z
           .string()
@@ -26,7 +26,7 @@ export const openResourceInputSchema = z.strictObject({
 export const openResourceOutputSchema = z.object({
   resources: z.array(
     z.object({
-      type: z.enum(['workflow', 'table', 'knowledgebase', 'file', 'dashboard', 'log']),
+      type: z.enum(['workflow', 'table', 'knowledgebase', 'file', 'dashboard', 'log', 'test']),
       id: z.string(),
       title: z.string(),
       viewId: z.string().optional(),

apps/sim/lib/mothership/resources/file-tabs.ts

@@ -1,6 +1,8 @@
 import { db } from '@sim/db'
 import { workspaceFiles } from '@sim/db/schema'
+import { toStringOrNull } from '@sim/utils/coerce'
 import { and, inArray, notInArray } from 'drizzle-orm'
+import type { BaseServerTool } from '@/lib/mothership/tools/server/base-tool'
 
 /** Row contexts that open as a file tab: workspace files and chat uploads. */
 const FILE_TAB_CONTEXTS = ['workspace', 'mothership']
@@ -23,3 +25,21 @@ export async function findNonTabFileIds(fileIds: string[]): Promise<Set<string>>
     )
   return new Set(rows.map((row) => row.id))
 }
+
+/**
+ * Marks a file-editing tool's result `fileTab: false` when the file it touched is not a file tab,
+ * so the browser, which promotes edited files to tabs from the result, leaves it alone.
+ */
+export function withFileTabFlag<TArgs, TResult extends { data?: Record<string, unknown> }>(
+  tool: BaseServerTool<TArgs, TResult>
+): BaseServerTool<TArgs, TResult> {
+  return {
+    ...tool,
+    async execute(args, context) {
+      const result = await tool.execute(args, context)
+      const id = toStringOrNull(result.data?.id)
+      if (id === null || !(await findNonTabFileIds([id])).has(id)) return result
+      return { ...result, data: { ...result.data, fileTab: false } }
+    },
+  }
+}

apps/sim/lib/mothership/tools/server/open-resource.ts

@@ -15,6 +15,7 @@ import { executeDashboardUseCase } from '@/lib/mothership/application/execute-da
 import { executeCopilotFileUseCase } from '@/lib/mothership/application/execute-file-use-case'
 import { executeCopilotKnowledgeUseCase } from '@/lib/mothership/application/execute-knowledge-use-case'
 import { executeCopilotTableUseCase } from '@/lib/mothership/application/execute-table-use-case'
+import { executeWorkflowTestUseCase } from '@/lib/mothership/application/execute-workflow-test-use-case'
 import { executeCopilotWorkflowUseCase } from '@/lib/mothership/application/execute-workflow-use-case'
 import {
   COPILOT_APPLICATION_DELEGATION_TTL_MS,
@@ -26,6 +27,7 @@ import {
 } from '@/lib/mothership/tools/server/base-tool'
 import { readTableDefinitionUseCase } from '@/lib/table/application/tables'
 import { readTableViewUseCase } from '@/lib/table/application/views'
+import { getWorkflowTestDetail } from '@/lib/workflow-tests/application/tests'
 import { readWorkflowMetadata } from '@/lib/workflows/application/read-workflow'
 import { readWorkspaceFileMetadata } from '@/lib/workspace-files/application/read-workspace-file-metadata'
 
@@ -96,6 +98,14 @@ export const openResourceServerTool: BaseServerTool<OpenResourceInput, OpenResou
           resources.push({ ...base, title: dashboard.name })
           break
         }
+        case 'test': {
+          const { test } = await executeWorkflowTestUseCase(context, getWorkflowTestDetail, {
+            workspaceId,
+            name: resource.id,
+          })
+          resources.push({ ...base, title: test.title })
+          break
+        }
         case 'file': {
           const { file } = await executeCopilotFileUseCase(
             context,

apps/sim/lib/mothership/tools/server/router.ts

@@ -9,6 +9,7 @@ import {
   GenerateImage,
   GenerateVideo,
 } from '@/lib/mothership/generated/tool-catalog-v1'
+import { withFileTabFlag } from '@/lib/mothership/resources/file-tabs'
 import { copilotToolCanWrite } from '@/lib/mothership/tools/permissions'
 import {
   assertServerToolNotAborted,
@@ -74,8 +75,8 @@ const baseServerToolRegistry: Record<string, BaseServerTool> = {
   [readDocumentServerTool.name]: readDocumentServerTool,
   // The streamed file-writing pair: prepare opens the write (live preview),
   // apply continues it. The preview machinery keys off these exact names.
-  [workspaceFileServerTool.name]: workspaceFileServerTool,
-  [editContentServerTool.name]: editContentServerTool,
+  [workspaceFileServerTool.name]: withFileTabFlag(workspaceFileServerTool),
+  [editContentServerTool.name]: withFileTabFlag(editContentServerTool),
   [generateImageServerTool.name]: generateImageServerTool,
   [generateVideoServerTool.name]: generateVideoServerTool,
   [generateAudioServerTool.name]: generateAudioServerTool,

apps/sim/lib/mothership/tools/server/tests.ts

@@ -143,7 +143,12 @@ export const testsServerTool: BaseServerTool = {
             }
           )
           if (current.every((run) => run.status !== 'running')) {
+            const resources: ResourceChange[] = current.map((run) => ({
+              op: 'upsert',
+              resource: { type: 'test', workspaceId, id: run.name, title: run.title },
+            }))
             return {
+              resources,
               version: input.version,
               results: current.map((run) => ({
                 name: run.name,

apps/sim/lib/workflow-tests/application/tests.ts

@@ -390,6 +390,7 @@ export const readWorkflowTestRuns = defineAuthorizedWorkspaceUseCase({
       runs: rows.map((row) => ({
         ...presentRun(row, row.testSourceHash, facts),
         name: row.testName,
+        title: row.testTitle,
         report: row.report as TestReport | null,
       })),
     }

apps/sim/lib/workflow-tests/repository.ts

@@ -437,16 +437,24 @@ export async function getWorkflowTestRunById(
 export async function listWorkflowTestRunsById(
   runIds: string[],
   workspaceId: string
-): Promise<Array<WorkflowTestRunRow & { testName: string; testSourceHash: string }>> {
+): Promise<
+  Array<WorkflowTestRunRow & { testName: string; testTitle: string; testSourceHash: string }>
+> {
   if (runIds.length === 0) return []
   const rows = await db
     .select({
       run: workflowTestRun,
       testName: workflowTest.name,
+      testTitle: workflowTest.title,
       testSourceHash: workflowTest.sourceHash,
     })
     .from(workflowTestRun)
     .innerJoin(workflowTest, eq(workflowTest.id, workflowTestRun.testId))
     .where(and(inArray(workflowTestRun.id, runIds), eq(workflowTestRun.workspaceId, workspaceId)))
-  return rows.map(({ run, testName, testSourceHash }) => ({ ...run, testName, testSourceHash }))
+  return rows.map(({ run, testName, testTitle, testSourceHash }) => ({
+    ...run,
+    testName,
+    testTitle,
+    testSourceHash,
+  }))
 }