fix: resolve AI chat undo feature not working

[2witstudios]

Summary

• Fixed AI chat undo feature that was always showing "Failed to undo changes"
• The root cause was incorrect usage of the post() helper from auth-fetch
• Updated activity logging to properly store previous values for rollback support

Details

The post() function from auth-fetch.ts:

• Returns parsed JSON on success (not a Response object)
• Throws an error on non-2xx responses

Components were incorrectly:

1. Using <Response> type annotation
2. Checking res.ok which is undefined on parsed JSON → !res.ok was always true
3. Calling res.json() which fails because res is already a plain object

This caused the error branch to always trigger, even when undo succeeded.

Changes

Dialog/Component Fixes

• UndoAiChangesDialog.tsx - Fixed to properly await post() and catch errors
• ActivityDashboard.tsx - Same pattern fix
• SidebarActivityTab.tsx - Same pattern fix
• VersionHistoryPanel.tsx - Same pattern fix

Activity Logging (for proper rollback)

• Updated logPageActivityAsync helper to accept previousValues, newValues, and updatedFields
• replace_lines - Now stores original page.content for rollback
• rename_page - Now stores original page.title for rollback
• trash/restore - Now stores isTrashed state for rollback
• move_page - Now stores original parentId and position for rollback
• edit_sheet_cells - Now stores original page.content for rollback

Test plan

• TypeScript compilation passes
• Page-write-tools unit tests pass (23/23)
• Manual test: Create AI chat, have AI make changes, click undo → changes should revert
• Manual test: Verify rollback works for page content, title, move, trash operations

🤖 Generated with Claude Code

Summary by CodeRabbit

Release Notes

• New Features

  • Create operations are now rollbackable

• Bug Fixes

  • Unified error handling for undo and rollback operations with standardized error messages
  • Enhanced activity logging with rollback metadata (previous values, new values, and updated fields)

• Refactor

  • Simplified API response handling by removing manual parsing in components

_{✏️ Tip: You can customize this high-level summary in your review settings.}

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.
To continue using code reviews, you can upgrade your account or add credits to your account and enable them for code reviews in your settings.

[coderabbitai]

📝 Walkthrough

Walkthrough

This PR consolidates error handling across undo and rollback operations by simplifying API response processing in multiple components, updating the undo endpoint to uniformly return 500 on failure, expanding activity logging with rollback metadata, and adjusting permission rules to classify 'create' operations as rollbackable.

Changes

Cohort / File(s)	Summary
Undo API Endpoint apps/web/src/app/api/ai/chat/messages/[messageId]/undo/route.ts, apps/web/src/app/api/ai/chat/messages/[messageId]/undo/__tests__/route.test.ts	Simplified failure handling to always return 500 with uniform message "Undo failed. No changes were applied." instead of treating partial success scenarios separately. Test failure scenarios updated to expect 500 status and standardized error messaging.
Component Rollback/Undo Calls apps/web/src/components/activity/ActivityDashboard.tsx, apps/web/src/components/ai/shared/chat/UndoAiChangesDialog.tsx, apps/web/src/components/layout/right-sidebar/ai-assistant/SidebarActivityTab.tsx, apps/web/src/components/version-history/VersionHistoryPanel.tsx	Removed explicit HTTP response validation (response.ok checks and JSON parsing). Now rely on POST helper to throw on errors, with error handling via outer catch block and toast notifications.
Hook Test Updates apps/web/src/hooks/__tests__/useBreadcrumbs.test.ts, apps/web/src/hooks/__tests__/usePermissions.test.ts	Updated null pageId test expectation from isLoading: true to isLoading: false. Enhanced usePermissions mock to trigger onSuccess callback and simulate initial load state for accurate isPaused assertions.
Activity Logging Infrastructure apps/web/src/lib/ai/tools/page-write-tools.ts, apps/web/src/services/api/ai-undo-service.ts	Expanded logPageActivityAsync signature to accept options with metadata, previousValues, newValues, and updatedFields for rollback support. Updated all call sites across page-write tools (create_page, replace_lines, rename_page, etc.). Zeroed error-path counters in undo service.
Permission Rules packages/lib/src/permissions/__tests__/rollback-permissions.test.ts	Reclassified 'create' operation as rollbackable; adjusted test expectations and non-rollbackable operation list accordingly.
Layout Component apps/web/src/components/layout/middle-content/CenterPanel.tsx	Removed useParams hook invocation from OptimizedViewHeader memoized component, reducing hook dependencies.

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~25 minutes

Possibly related PRs

• test: improve test quality with observable outcomes #82: Modifies the same useBreadcrumbs test file with changes to isLoading expectations and observable behavior assertions.
• test: add unit tests for lib modules #78: Updates unit tests for page-write-tools.ts module, directly complementing the expanded logPageActivityAsync signature and call sites in this PR.
• feat: Version History & Rollback #118: Modifies the same undo flow endpoint to standardize failure handling with uniform 500 responses and consistent error messaging.

Poem

🐰 With a hop and a bound, we simplify the sound,
No more checking responses, just let errors be found,
Rollback tracks the changes, previousValues so bright,
Create's now rollbackable—everything's just right! 🌟

Pre-merge checks and finishing touches

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 66.67% which is insufficient. The required threshold is 80.00%.	You can run @coderabbitai generate docstrings to improve docstring coverage.

✅ Passed checks (2 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title accurately describes the main fix: resolving an AI chat undo feature that was not working, which aligns with the core changes across multiple components and services.

✨ Finishing touches

• 📝 Generate docstrings

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Post copyable unit tests in a comment
• Commit unit tests in branch feature/version-history-rollback

---

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[coderabbitai]

Actionable comments posted: 0

🧹 Nitpick comments (3)

apps/web/src/hooks/__tests__/usePermissions.test.ts (2)

41-54: Type assertion for onSuccess config parameter could be improved.

Line 45 uses {} as never for the third parameter of the onSuccess callback. While this works because the actual onSuccess handler in usePermissions doesn't use this parameter, the type assertion is semantically incorrect. The SWR onSuccess signature expects a Readonly<PublicConfiguration<...>> type for the third parameter.

Consider using a more appropriate type assertion for better type safety:

🔎 Suggested improvements

       // Mock SWR and trigger onSuccess to set hasLoadedRef.current = true
       vi.mocked(useSWR).mockImplementation((key, fetcher, config) => {
         // Trigger onSuccess to simulate initial load completed
         if (config?.onSuccess) {
-          config.onSuccess({ canView: true, canEdit: true, canShare: true, canDelete: true }, key as string, {} as never);
+          config.onSuccess({ canView: true, canEdit: true, canShare: true, canDelete: true }, key as string, {} as any);
         }
         return {
           data: { canView: true, canEdit: true, canShare: true, canDelete: true },
           error: undefined,
           isLoading: false,
           mutate: vi.fn(),
           isValidating: false,
         } as SWRResponse;
       });

This change should be applied to all three test cases (lines 45, 77, and 107).

---

42-54: Consider extracting common mock implementation into a helper.

The same mock implementation pattern is repeated in all three tests. While the current approach is clear and readable, extracting this into a helper function could reduce duplication and improve maintainability.

🔎 Suggested refactor

// Add helper function before the describe block
function mockUseSWRWithOnSuccess(permissions = { canView: true, canEdit: true, canShare: true, canDelete: true }) {
  vi.mocked(useSWR).mockImplementation((key, fetcher, config) => {
    // Trigger onSuccess to simulate initial load completed
    if (config?.onSuccess) {
      config.onSuccess(permissions, key as string, {} as any);
    }
    return {
      data: permissions,
      error: undefined,
      isLoading: false,
      mutate: vi.fn(),
      isValidating: false,
    } as SWRResponse;
  });
}

// Then use in tests:
it('given user is editing a document, should pause permission revalidation', () => {
  vi.mocked(useEditingStore).mockReturnValue(true);
  mockUseSWRWithOnSuccess();
  renderHook(() => usePermissions('page-123'));
  // ... assertions
});

Also applies to: 74-86, 104-116

apps/web/src/lib/ai/tools/page-write-tools.ts (1)

613-616: Consider adding newValues for completeness.

The trash and restore operations correctly log previousValues for rollback support, but unlike other operations (replace_lines, rename_page, move_page, edit_sheet_cells), they don't log newValues.

While the current implementation is sufficient for rollback functionality, adding newValues would improve:

• Consistency with other tools that log both before/after states
• Verification that the operation succeeded as expected
• Audit trail completeness

💡 Suggested enhancement

For trash operation (line 613-616):

 logPageActivityAsync(userId, 'trash', {
   id: page.id,
   title: page.title,
   driveId: page.driveId,
 }, context as ToolExecutionContext, {
   metadata: { withChildren, childrenCount },
   previousValues: { isTrashed: false },
+  newValues: { isTrashed: true },
+  updatedFields: ['isTrashed'],
 });

For restore operation (line 687-689):

 logPageActivityAsync(userId, 'restore', {
   id: page.id,
   title: page.title,
   driveId: page.driveId,
 }, context as ToolExecutionContext, {
   previousValues: { isTrashed: true },
+  newValues: { isTrashed: false },
+  updatedFields: ['isTrashed'],
 });

Also applies to: 687-689

📜 Review details

Configuration used: defaults

Review profile: CHILL

Plan: Pro

📥 Commits

Reviewing files that changed from the base of the PR and between 1e490ae and 1666f81.

📒 Files selected for processing (12)

• apps/web/src/app/api/ai/chat/messages/[messageId]/undo/__tests__/route.test.ts
• apps/web/src/app/api/ai/chat/messages/[messageId]/undo/route.ts
• apps/web/src/components/activity/ActivityDashboard.tsx
• apps/web/src/components/ai/shared/chat/UndoAiChangesDialog.tsx
• apps/web/src/components/layout/middle-content/CenterPanel.tsx
• apps/web/src/components/layout/right-sidebar/ai-assistant/SidebarActivityTab.tsx
• apps/web/src/components/version-history/VersionHistoryPanel.tsx
• apps/web/src/hooks/__tests__/useBreadcrumbs.test.ts
• apps/web/src/hooks/__tests__/usePermissions.test.ts
• apps/web/src/lib/ai/tools/page-write-tools.ts
• apps/web/src/services/api/ai-undo-service.ts
• packages/lib/src/permissions/__tests__/rollback-permissions.test.ts

💤 Files with no reviewable changes (1)

• apps/web/src/components/layout/middle-content/CenterPanel.tsx

🧰 Additional context used

📓 Path-based instructions (9)

apps/web/src/app/**/route.{ts,tsx}

📄 CodeRabbit inference engine (AGENTS.md)

apps/web/src/app/**/route.{ts,tsx}: In Next.js 15 route handlers, params in dynamic routes are Promise objects and MUST be awaited before destructuring
Use Response.json() or NextResponse.json() for returning JSON from route handlers
Get request body using const body = await request.json();
Get search parameters using const { searchParams } = new URL(request.url);

apps/web/src/app/**/route.{ts,tsx}: In Next.js 15 dynamic routes, params are Promise objects and MUST be awaited before destructuring: const { id } = await context.params;
In Route Handlers, get request body with const body = await request.json();
In Route Handlers, get search parameters with const { searchParams } = new URL(request.url);
In Route Handlers, return JSON using Response.json(data) or NextResponse.json(data)
For permission logic, use centralized functions from @pagespace/lib/permissions: getUserAccessLevel(), canUserEditPage()

Files:

• apps/web/src/app/api/ai/chat/messages/[messageId]/undo/route.ts

**/*.{ts,tsx}

📄 CodeRabbit inference engine (AGENTS.md)

**/*.{ts,tsx}: Never use any types - always use proper TypeScript types
Use camelCase for variable and function names
Use UPPER_SNAKE_CASE for constants
Use PascalCase for type and enum names
Use kebab-case for filenames, except React hooks (camelCase with use prefix), Zustand stores (camelCase with use prefix), and React components (PascalCase)
Lint with Next/ESLint as configured in apps/web/eslint.config.mjs
Message content should always use the message parts structure with { parts: [{ type: 'text', text: '...' }] }
Use centralized permission functions from @pagespace/lib/permissions (e.g., getUserAccessLevel, canUserEditPage) instead of implementing permission logic locally
Always use Drizzle client from @pagespace/db package for database access
Use ESM modules throughout the codebase

**/*.{ts,tsx}: Never use any types - always use proper TypeScript types
Write code that is explicit over implicit and self-documenting

Files:

• apps/web/src/app/api/ai/chat/messages/[messageId]/undo/route.ts
• apps/web/src/hooks/__tests__/usePermissions.test.ts
• apps/web/src/components/ai/shared/chat/UndoAiChangesDialog.tsx
• packages/lib/src/permissions/__tests__/rollback-permissions.test.ts
• apps/web/src/components/version-history/VersionHistoryPanel.tsx
• apps/web/src/app/api/ai/chat/messages/[messageId]/undo/__tests__/route.test.ts
• apps/web/src/services/api/ai-undo-service.ts
• apps/web/src/components/layout/right-sidebar/ai-assistant/SidebarActivityTab.tsx
• apps/web/src/lib/ai/tools/page-write-tools.ts
• apps/web/src/components/activity/ActivityDashboard.tsx
• apps/web/src/hooks/__tests__/useBreadcrumbs.test.ts

**/*.ts

📄 CodeRabbit inference engine (AGENTS.md)

**/*.ts: React hook files should use camelCase matching the exported hook name (e.g., useAuth.ts)
Zustand store files should use camelCase with use prefix (e.g., useAuthStore.ts)

Files:

• apps/web/src/app/api/ai/chat/messages/[messageId]/undo/route.ts
• apps/web/src/hooks/__tests__/usePermissions.test.ts
• packages/lib/src/permissions/__tests__/rollback-permissions.test.ts
• apps/web/src/app/api/ai/chat/messages/[messageId]/undo/__tests__/route.test.ts
• apps/web/src/services/api/ai-undo-service.ts
• apps/web/src/lib/ai/tools/page-write-tools.ts
• apps/web/src/hooks/__tests__/useBreadcrumbs.test.ts

**/*.{ts,tsx,js,jsx,json}

📄 CodeRabbit inference engine (AGENTS.md)

Format code with Prettier

Files:

• apps/web/src/app/api/ai/chat/messages/[messageId]/undo/route.ts
• apps/web/src/hooks/__tests__/usePermissions.test.ts
• apps/web/src/components/ai/shared/chat/UndoAiChangesDialog.tsx
• packages/lib/src/permissions/__tests__/rollback-permissions.test.ts
• apps/web/src/components/version-history/VersionHistoryPanel.tsx
• apps/web/src/app/api/ai/chat/messages/[messageId]/undo/__tests__/route.test.ts
• apps/web/src/services/api/ai-undo-service.ts
• apps/web/src/components/layout/right-sidebar/ai-assistant/SidebarActivityTab.tsx
• apps/web/src/lib/ai/tools/page-write-tools.ts
• apps/web/src/components/activity/ActivityDashboard.tsx
• apps/web/src/hooks/__tests__/useBreadcrumbs.test.ts

apps/web/src/**/*.{ts,tsx}

📄 CodeRabbit inference engine (CLAUDE.md)

apps/web/src/**/*.{ts,tsx}: Use message parts structure for message content: { parts: [{ type: 'text', text: '...' }] }
For database access, always use Drizzle client from @pagespace/db: import { db, pages } from '@pagespace/db';
Use centralized Drizzle ORM with PostgreSQL for all database operations - no direct SQL or other ORMs
Use Socket.IO for real-time collaboration features - imported from the realtime service at port 3001
Use Vercel AI SDK with async/await for all AI operations and streaming
Use Next.js 15 App Router and TypeScript for all routes and components

Files:

• apps/web/src/app/api/ai/chat/messages/[messageId]/undo/route.ts
• apps/web/src/hooks/__tests__/usePermissions.test.ts
• apps/web/src/components/ai/shared/chat/UndoAiChangesDialog.tsx
• apps/web/src/components/version-history/VersionHistoryPanel.tsx
• apps/web/src/app/api/ai/chat/messages/[messageId]/undo/__tests__/route.test.ts
• apps/web/src/services/api/ai-undo-service.ts
• apps/web/src/components/layout/right-sidebar/ai-assistant/SidebarActivityTab.tsx
• apps/web/src/lib/ai/tools/page-write-tools.ts
• apps/web/src/components/activity/ActivityDashboard.tsx
• apps/web/src/hooks/__tests__/useBreadcrumbs.test.ts

**/*.tsx

📄 CodeRabbit inference engine (AGENTS.md)

**/*.tsx: React component files should use PascalCase (e.g., UserProfile.tsx)
Use @dnd-kit for drag-and-drop functionality
Use Zustand for client state management
Use SWR for server state management and caching

Files:

• apps/web/src/components/ai/shared/chat/UndoAiChangesDialog.tsx
• apps/web/src/components/version-history/VersionHistoryPanel.tsx
• apps/web/src/components/layout/right-sidebar/ai-assistant/SidebarActivityTab.tsx
• apps/web/src/components/activity/ActivityDashboard.tsx

**/*.{tsx,css}

📄 CodeRabbit inference engine (AGENTS.md)

Use Tailwind CSS and shadcn/ui components for styling and UI

Files:

• apps/web/src/components/ai/shared/chat/UndoAiChangesDialog.tsx
• apps/web/src/components/version-history/VersionHistoryPanel.tsx
• apps/web/src/components/layout/right-sidebar/ai-assistant/SidebarActivityTab.tsx
• apps/web/src/components/activity/ActivityDashboard.tsx

apps/web/src/components/**/*.{ts,tsx}

📄 CodeRabbit inference engine (CLAUDE.md)

apps/web/src/components/**/*.{ts,tsx}: When document editing, register editing state with useEditingStore.getState().startEditing() to prevent UI refreshes, and clean up in return statement
When AI is streaming, register streaming state with useEditingStore.getState().startStreaming() to prevent UI refreshes, and clean up in return statement
For SWR data fetching with editing protection, use isPaused: () => hasLoadedRef.current && isEditingActive() to allow initial fetch and only pause after, with revalidateOnFocus: false
Use Zustand for client-side state management as the primary state solution
Use SWR for server state and caching with proper configuration including revalidateOnFocus: false for editing protection
Use TipTap rich text editor with markdown support for document editing
Use Monaco Editor for code editing features
Use @dnd-kit for drag-and-drop functionality instead of other libraries
Use Tailwind CSS with shadcn/ui components for all UI styling and components

Files:

• apps/web/src/components/ai/shared/chat/UndoAiChangesDialog.tsx
• apps/web/src/components/version-history/VersionHistoryPanel.tsx
• apps/web/src/components/layout/right-sidebar/ai-assistant/SidebarActivityTab.tsx
• apps/web/src/components/activity/ActivityDashboard.tsx

**/*ai*.{ts,tsx}

📄 CodeRabbit inference engine (AGENTS.md)

Use Vercel AI SDK for AI integrations

Files:

• apps/web/src/services/api/ai-undo-service.ts

🧠 Learnings (7)

📚 Learning: 2025-12-23T18:49:41.955Z

Learnt from: CR
Repo: 2witstudios/PageSpace PR: 0
File: CLAUDE.md:0-0
Timestamp: 2025-12-23T18:49:41.955Z
Learning: Applies to apps/web/src/components/**/*.{ts,tsx} : For SWR data fetching with editing protection, use `isPaused: () => hasLoadedRef.current && isEditingActive()` to allow initial fetch and only pause after, with `revalidateOnFocus: false`

Applied to files:

• apps/web/src/hooks/__tests__/usePermissions.test.ts

📚 Learning: 2025-12-23T18:49:41.955Z

Learnt from: CR
Repo: 2witstudios/PageSpace PR: 0
File: CLAUDE.md:0-0
Timestamp: 2025-12-23T18:49:41.955Z
Learning: Applies to apps/web/src/components/**/*.{ts,tsx} : Use SWR for server state and caching with proper configuration including `revalidateOnFocus: false` for editing protection

Applied to files:

• apps/web/src/hooks/__tests__/usePermissions.test.ts

📚 Learning: 2025-12-22T20:04:40.892Z

Learnt from: CR
Repo: 2witstudios/PageSpace PR: 0
File: AGENTS.md:0-0
Timestamp: 2025-12-22T20:04:40.892Z
Learning: Applies to **/*.{ts,tsx} : Use centralized permission functions from `pagespace/lib/permissions` (e.g., `getUserAccessLevel`, `canUserEditPage`) instead of implementing permission logic locally

Applied to files:

• apps/web/src/hooks/__tests__/usePermissions.test.ts

📚 Learning: 2025-12-16T19:03:59.870Z

Learnt from: 2witstudios
Repo: 2witstudios/PageSpace PR: 91
File: apps/web/src/components/ai/shared/chat/tool-calls/CompactToolCallRenderer.tsx:253-277
Timestamp: 2025-12-16T19:03:59.870Z
Learning: In apps/web/src/components/ai/shared/chat/tool-calls/CompactToolCallRenderer.tsx (TypeScript/React), use the `getLanguageFromPath` utility from `formatters.ts` to infer syntax highlighting language from file paths instead of hardcoding language values in DocumentRenderer calls.

Applied to files:

• apps/web/src/components/ai/shared/chat/UndoAiChangesDialog.tsx

📚 Learning: 2025-12-23T18:49:41.955Z

Learnt from: CR
Repo: 2witstudios/PageSpace PR: 0
File: CLAUDE.md:0-0
Timestamp: 2025-12-23T18:49:41.955Z
Learning: Applies to apps/web/src/**/*.{ts,tsx} : Use Vercel AI SDK with async/await for all AI operations and streaming

Applied to files:

• apps/web/src/components/ai/shared/chat/UndoAiChangesDialog.tsx

📚 Learning: 2025-12-22T20:04:40.892Z

Learnt from: CR
Repo: 2witstudios/PageSpace PR: 0
File: AGENTS.md:0-0
Timestamp: 2025-12-22T20:04:40.892Z
Learning: Applies to **/*ai*.{ts,tsx} : Use Vercel AI SDK for AI integrations

Applied to files:

• apps/web/src/components/ai/shared/chat/UndoAiChangesDialog.tsx

📚 Learning: 2025-12-18T05:22:42.263Z

Learnt from: 2witstudios
Repo: 2witstudios/PageSpace PR: 96
File: apps/web/src/components/layout/right-sidebar/ai-assistant/SidebarChatTab.tsx:641-657
Timestamp: 2025-12-18T05:22:42.263Z
Learning: In apps/web/src/components/layout/right-sidebar/ai-assistant/SidebarChatTab.tsx, the provider/model selector buttons are intentionally non-functional placeholders in the compact sidebar view. The `hideModelSelector={true}` prop is passed to ChatInput to hide the full ProviderModelSelector. Users are expected to use the full GlobalAssistantView for model selection. A settings link may be added in a future iteration.

Applied to files:

• apps/web/src/components/layout/right-sidebar/ai-assistant/SidebarActivityTab.tsx

🧬 Code graph analysis (4)

apps/web/src/hooks/__tests__/usePermissions.test.ts (1)

apps/web/src/hooks/usePermissions.ts (1)

• usePermissions (33-98)

apps/web/src/app/api/ai/chat/messages/[messageId]/undo/__tests__/route.test.ts (2)

apps/web/src/services/api/ai-undo-service.ts (1)

• executeAiUndo (254-368)

apps/web/src/app/api/ai/chat/messages/[messageId]/undo/route.ts (1)

• POST (112-186)

apps/web/src/lib/ai/tools/page-write-tools.ts (1)

apps/web/src/lib/ai/core/types.ts (1)

• ToolExecutionContext (8-37)

apps/web/src/hooks/__tests__/useBreadcrumbs.test.ts (1)

apps/web/src/hooks/useBreadcrumbs.ts (1)

• useBreadcrumbs (23-50)

⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (1)

• GitHub Check: Unit Tests

🔇 Additional comments (19)

apps/web/src/hooks/__tests__/useBreadcrumbs.test.ts (1)

175-182: LGTM! Test correctly fixed to match hook implementation.

The change correctly fixes the test assertion. The useBreadcrumbs hook computes isLoading as !error && !data && !!pageId, which evaluates to false when pageId is null (since !!null is false). The updated test description, comment, and assertion all accurately reflect this behavior.

apps/web/src/components/version-history/VersionHistoryPanel.tsx (1)

132-150: LGTM! Correct usage of post() helper.

The refactored code correctly uses the post() helper, which returns parsed JSON on success and throws on non-2xx responses. Error handling is properly delegated to the catch block, and the success path remains unchanged.

apps/web/src/services/api/ai-undo-service.ts (1)

354-367: LGTM! Correct handling of transactional failure.

Returning zeroed counters on failure correctly reflects the all-or-nothing transaction semantics. Since the entire transaction is rolled back on any error (line 281), no changes are actually applied, so returning 0 for both counters is accurate and prevents misleading partial-success indicators.

apps/web/src/components/ai/shared/chat/UndoAiChangesDialog.tsx (1)

74-92: LGTM! Correct usage with helpful documentation.

The refactored code correctly uses the post() helper and includes helpful comments explaining the behavior. Error handling is properly delegated to the catch block, and the success flow remains intact.

apps/web/src/app/api/ai/chat/messages/[messageId]/undo/route.ts (1)

159-168: LGTM! Uniform failure handling.

The simplified error handling correctly implements all-or-nothing transaction semantics with a clear, uniform failure message. The 500 status code is appropriate, and including the result object provides debugging information while maintaining a consistent user-facing message.

apps/web/src/components/layout/right-sidebar/ai-assistant/SidebarActivityTab.tsx (1)

250-272: LGTM! Consistent implementation.

The refactored rollback handler correctly uses the post() helper without manual response handling. Error handling is properly delegated to the catch block, and the success path maintains the expected behavior (toast notification and activity list refresh).

apps/web/src/components/activity/ActivityDashboard.tsx (1)

212-225: LGTM! Correct implementation.

The refactored rollback handler correctly uses the post() helper. The useCallback dependencies are accurate, error handling is properly delegated to the catch block, and the success path maintains the expected behavior.

apps/web/src/app/api/ai/chat/messages/[messageId]/undo/__tests__/route.test.ts (1)

349-381: LGTM! Tests correctly validate the new failure contract.

The updated failure tests correctly validate the uniform 500 error response with the message "Undo failed. No changes were applied." The mock data properly reflects transactional all-or-nothing semantics with zeroed counters, and the tests cover both scenarios (with and without specific error messages).

packages/lib/src/permissions/__tests__/rollback-permissions.test.ts (3)

159-167: Correct update of test case to reflect 'create' reclassification.

The test correctly switches from 'create' to 'signup' to validate non-rollbackable operation behavior, since 'create' is now classified as rollbackable. This maintains test validity and consistency.

---

195-209: Parametrized test correctly updated to exclude 'create'.

The test array appropriately excludes 'create' and uses representative non-rollbackable operations (signup, login, logout) to validate rejection logic. The subset approach is efficient while maintaining test coverage.

---

55-73: Implementation correctly marks 'create' as rollbackable.

The implementation in rollback-permissions.ts is consistent with the test changes. The isRollbackableOperation() function includes 'create' in its rollbackable operations list (line 169), and the nonRollbackableOperations correctly defines only ['signup', 'login', 'logout']. The permission checks across all contexts (page, drive, ai_tool, user_dashboard) are properly configured to allow rollback of create operations for authorized users.

apps/web/src/hooks/__tests__/usePermissions.test.ts (1)

36-128: Well-structured tests that properly simulate initial load completion.

The changes successfully implement the testing strategy for SWR editing protection by:

• Triggering the onSuccess callback to simulate initial load completion (setting hasLoadedRef.current = true)
• Verifying isPaused behavior in three scenarios: editing active, AI streaming active, and no activity
• Updating comments to clarify the "after initial load" behavior

The test logic correctly matches the implementation in usePermissions.ts where isPaused: () => hasLoadedRef.current && isAnyActive().

Based on learnings, the isPaused implementation tested here follows the documented pattern for SWR data fetching with editing protection.

apps/web/src/lib/ai/tools/page-write-tools.ts (7)

29-78: Excellent implementation of rollback-enabled activity logging!

The extension of logPageActivityAsync to accept previousValues, newValues, and updatedFields is well-designed:

• Properly typed optional parameter maintains backward compatibility
• Metadata merging correctly preserves agent chain context
• Both success and error branches consistently propagate rollback fields
• Clean separation of concerns between metadata and rollback-specific data

This provides a solid foundation for the undo/rollback feature.

---

340-354: Perfect rollback logging for content updates!

The replace_lines tool correctly captures all necessary data for rollback:

• previousValues stores original content before modification
• newValues stores new content after changes
• updatedFields array accurately identifies the modified field
• Metadata provides useful context (linesChanged, changeType)

The timing is correct—original page.content is captured before the database update at line 330.

---

552-556: Well-implemented rollback support for page renaming.

The logging correctly captures the title change with proper before/after values:

• previousValues preserves the original title for rollback
• newValues records the new title for verification
• State capture timing is accurate relative to the DB update at line 537

---

795-800: Comprehensive rollback data for page moves.

The move_page tool properly logs all location changes:

• Both parentId and position are captured in previousValues/newValues
• updatedFields correctly lists both modified fields
• The metadata provides quick reference to the new location

This enables complete restoration of the page's original position during rollback.

---

898-909: Solid rollback implementation for sheet edits.

Similar to replace_lines, this correctly implements full rollback support for sheet content:

• Original content captured before database update
• Both previousValues and newValues properly populated
• Helpful comment documents rollback usage
• Metadata provides useful context about the scope of changes

---

463-465: Appropriate logging for create operations.

The create_page tool logs only metadata (pageType, parentId) without previousValues or newValues, which is correct for create operations:

• There's no prior state to capture for a newly created page
• Rollback of a create operation means deleting the created page
• The activity log already contains the page ID and operation type, which is sufficient for rollback logic

This aligns with the PR objective to make create operations rollbackable.

---

1-952: Excellent implementation of rollback-enabled activity logging across all tools!

This file successfully implements the PR objectives by enhancing activity logging to support the undo/rollback feature. Key strengths:

✅ Consistent rollback data capture across all content-modifying operations
✅ Proper timing of state capture (before DB updates)
✅ Clean, backward-compatible API design
✅ Strong type safety throughout
✅ Follows coding guidelines (centralized permissions, repository pattern, no any types)
✅ Well-documented with helpful comments explaining rollback usage

The implementation provides all necessary data for the rollback endpoints to restore pages to their previous states.