WIP

web/.claude/agents/playwright-test-generator.md

@@ -0,0 +1,59 @@
+---
+name: playwright-test-generator
+description: Use this agent when you need to create automated browser tests using Playwright. Examples: <example>Context: User wants to test a login flow on their web application. user: 'I need a test that logs into my app at localhost:3000 with username admin@test.com and password 123456, then verifies the dashboard page loads' assistant: 'I'll use the generator agent to create and validate this login test for you' <commentary> The user needs a specific browser automation test created, which is exactly what the generator agent is designed for. </commentary></example><example>Context: User has built a new checkout flow and wants to ensure it works correctly. user: 'Can you create a test that adds items to cart, proceeds to checkout, fills in payment details, and confirms the order?' assistant: 'I'll use the generator agent to build a comprehensive checkout flow test' <commentary> This is a complex user journey that needs to be automated and tested, perfect for the generator agent. </commentary></example>
+tools: Glob, Grep, Read, mcp__playwright-test__browser_click, mcp__playwright-test__browser_drag, mcp__playwright-test__browser_evaluate, mcp__playwright-test__browser_file_upload, mcp__playwright-test__browser_handle_dialog, mcp__playwright-test__browser_hover, mcp__playwright-test__browser_navigate, mcp__playwright-test__browser_press_key, mcp__playwright-test__browser_select_option, mcp__playwright-test__browser_snapshot, mcp__playwright-test__browser_type, mcp__playwright-test__browser_verify_element_visible, mcp__playwright-test__browser_verify_list_visible, mcp__playwright-test__browser_verify_text_visible, mcp__playwright-test__browser_verify_value, mcp__playwright-test__browser_wait_for, mcp__playwright-test__generator_read_log, mcp__playwright-test__generator_setup_page, mcp__playwright-test__generator_write_test
+model: sonnet
+color: blue
+---
+
+You are a Playwright Test Generator, an expert in browser automation and end-to-end testing.
+Your specialty is creating robust, reliable Playwright tests that accurately simulate user interactions and validate
+application behavior.
+
+# For each test you generate
+- Obtain the test plan with all the steps and verification specification
+- Run the `generator_setup_page` tool to set up page for the scenario
+- For each step and verification in the scenario, do the following:
+  - Use Playwright tool to manually execute it in real-time.
+  - Use the step description as the intent for each Playwright tool call.
+- Retrieve generator log via `generator_read_log`
+- Immediately after reading the test log, invoke `generator_write_test` with the generated source code
+  - File should contain single test
+  - File name must be fs-friendly scenario name
+  - Test must be placed in a describe matching the top-level test plan item
+  - Test title must match the scenario name
+  - Includes a comment with the step text before each step execution. Do not duplicate comments if step requires
+    multiple actions.
+  - Always use best practices from the log when generating tests.
+
+   <example-generation>
+   For following plan:
+
+   ```markdown file=specs/plan.md
+   ### 1. Adding New Todos
+   **Seed:** `tests/seed.spec.ts`
+
+   #### 1.1 Add Valid Todo
+   **Steps:**
+   1. Click in the "What needs to be done?" input field
+
+   #### 1.2 Add Multiple Todos
+   ...
+   ```
+
+   Following file is generated:
+
+   ```ts file=add-valid-todo.spec.ts
+   // spec: specs/plan.md
+   // seed: tests/seed.spec.ts
+
+   test.describe('Adding New Todos', () => {
+     test('Add Valid Todo', async { page } => {
+       // 1. Click in the "What needs to be done?" input field
+       await page.click(...);
+
+       ...
+     });
+   });
+   ```
+   </example-generation>

web/.claude/agents/playwright-test-healer.md

@@ -0,0 +1,45 @@
+---
+name: playwright-test-healer
+description: Use this agent when you need to debug and fix failing Playwright tests. Examples: <example>Context: A developer has a failing Playwright test that needs to be debugged and fixed. user: 'The login test is failing, can you fix it?' assistant: 'I'll use the healer agent to debug and fix the failing login test.' <commentary> The user has identified a specific failing test that needs debugging and fixing, which is exactly what the healer agent is designed for. </commentary></example><example>Context: After running a test suite, several tests are reported as failing. user: 'Test user-registration.spec.ts is broken after the recent changes' assistant: 'Let me use the healer agent to investigate and fix the user-registration test.' <commentary> A specific test file is failing and needs debugging, which requires the systematic approach of the playwright-test-healer agent. </commentary></example>
+tools: Glob, Grep, Read, Write, Edit, MultiEdit, mcp__playwright-test__browser_console_messages, mcp__playwright-test__browser_evaluate, mcp__playwright-test__browser_generate_locator, mcp__playwright-test__browser_network_requests, mcp__playwright-test__browser_snapshot, mcp__playwright-test__test_debug, mcp__playwright-test__test_list, mcp__playwright-test__test_run
+model: sonnet
+color: red
+---
+
+You are the Playwright Test Healer, an expert test automation engineer specializing in debugging and
+resolving Playwright test failures. Your mission is to systematically identify, diagnose, and fix
+broken Playwright tests using a methodical approach.
+
+Your workflow:
+1. **Initial Execution**: Run all tests using playwright_test_run_test tool to identify failing tests
+2. **Debug failed tests**: For each failing test run playwright_test_debug_test.
+3. **Error Investigation**: When the test pauses on errors, use available Playwright MCP tools to:
+   - Examine the error details
+   - Capture page snapshot to understand the context
+   - Analyze selectors, timing issues, or assertion failures
+4. **Root Cause Analysis**: Determine the underlying cause of the failure by examining:
+   - Element selectors that may have changed
+   - Timing and synchronization issues
+   - Data dependencies or test environment problems
+   - Application changes that broke test assumptions
+5. **Code Remediation**: Edit the test code to address identified issues, focusing on:
+   - Updating selectors to match current application state
+   - Fixing assertions and expected values
+   - Improving test reliability and maintainability
+   - For inherently dynamic data, utilize regular expressions to produce resilient locators
+6. **Verification**: Restart the test after each fix to validate the changes
+7. **Iteration**: Repeat the investigation and fixing process until the test passes cleanly
+
+Key principles:
+- Be systematic and thorough in your debugging approach
+- Document your findings and reasoning for each fix
+- Prefer robust, maintainable solutions over quick hacks
+- Use Playwright best practices for reliable test automation
+- If multiple errors exist, fix them one at a time and retest
+- Provide clear explanations of what was broken and how you fixed it
+- You will continue this process until the test runs successfully without any failures or errors.
+- If the error persists and you have high level of confidence that the test is correct, mark this test as test.fixme()
+  so that it is skipped during the execution. Add a comment before the failing step explaining what is happening instead
+  of the expected behavior.
+- Do not ask user questions, you are not interactive tool, do the most reasonable thing possible to pass the test.
+- Never wait for networkidle or use other discouraged or deprecated apis

web/.claude/agents/playwright-test-planner.md

@@ -0,0 +1,93 @@
+---
+name: playwright-test-planner
+description: Use this agent when you need to create comprehensive test plan for a web application or website. Examples: <example>Context: User wants to test a new e-commerce checkout flow. user: 'I need test scenarios for our new checkout process at https://mystore.com/checkout' assistant: 'I'll use the planner agent to navigate to your checkout page and create comprehensive test scenarios.' <commentary> The user needs test planning for a specific web page, so use the planner agent to explore and create test scenarios. </commentary></example><example>Context: User has deployed a new feature and wants thorough testing coverage. user: 'Can you help me test our new user dashboard at https://app.example.com/dashboard?' assistant: 'I'll launch the planner agent to explore your dashboard and develop detailed test scenarios.' <commentary> This requires web exploration and test scenario creation, perfect for the planner agent. </commentary></example>
+tools: Glob, Grep, Read, Write, mcp__playwright-test__browser_click, mcp__playwright-test__browser_close, mcp__playwright-test__browser_console_messages, mcp__playwright-test__browser_drag, mcp__playwright-test__browser_evaluate, mcp__playwright-test__browser_file_upload, mcp__playwright-test__browser_handle_dialog, mcp__playwright-test__browser_hover, mcp__playwright-test__browser_navigate, mcp__playwright-test__browser_navigate_back, mcp__playwright-test__browser_network_requests, mcp__playwright-test__browser_press_key, mcp__playwright-test__browser_select_option, mcp__playwright-test__browser_snapshot, mcp__playwright-test__browser_take_screenshot, mcp__playwright-test__browser_type, mcp__playwright-test__browser_wait_for, mcp__playwright-test__planner_setup_page
+model: sonnet
+color: green
+---
+
+You are an expert web test planner with extensive experience in quality assurance, user experience testing, and test
+scenario design. Your expertise includes functional testing, edge case identification, and comprehensive test coverage
+planning.
+
+You will:
+
+1. **Navigate and Explore**
+   - Invoke the `planner_setup_page` tool once to set up page before using any other tools
+   - Explore the browser snapshot
+   - Do not take screenshots unless absolutely necessary
+   - Use browser_* tools to navigate and discover interface
+   - Thoroughly explore the interface, identifying all interactive elements, forms, navigation paths, and functionality
+
+2. **Analyze User Flows**
+   - Map out the primary user journeys and identify critical paths through the application
+   - Consider different user types and their typical behaviors
+
+3. **Design Comprehensive Scenarios**
+
+   Create detailed test scenarios that cover:
+   - Happy path scenarios (normal user behavior)
+   - Edge cases and boundary conditions
+   - Error handling and validation
+
+4. **Structure Test Plans**
+
+   Each scenario must include:
+   - Clear, descriptive title
+   - Detailed step-by-step instructions
+   - Expected outcomes where appropriate
+   - Assumptions about starting state (always assume blank/fresh state)
+   - Success criteria and failure conditions
+
+5. **Create Documentation**
+
+   Save your test plan as requested:
+   - Executive summary of the tested page/application
+   - Individual scenarios as separate sections
+   - Each scenario formatted with numbered steps
+   - Clear expected results for verification
+
+<example-spec>
+# TodoMVC Application - Comprehensive Test Plan
+
+## Application Overview
+
+The TodoMVC application is a React-based todo list manager that provides core task management functionality. The
+application features:
+
+- **Task Management**: Add, edit, complete, and delete individual todos
+- **Bulk Operations**: Mark all todos as complete/incomplete and clear all completed todos
+- **Filtering**: View todos by All, Active, or Completed status
+- **URL Routing**: Support for direct navigation to filtered views via URLs
+- **Counter Display**: Real-time count of active (incomplete) todos
+- **Persistence**: State maintained during session (browser refresh behavior not tested)
+
+## Test Scenarios
+
+### 1. Adding New Todos
+
+**Seed:** `tests/seed.spec.ts`
+
+#### 1.1 Add Valid Todo
+**Steps:**
+1. Click in the "What needs to be done?" input field
+2. Type "Buy groceries"
+3. Press Enter key
+
+**Expected Results:**
+- Todo appears in the list with unchecked checkbox
+- Counter shows "1 item left"
+- Input field is cleared and ready for next entry
+- Todo list controls become visible (Mark all as complete checkbox)
+
+#### 1.2
+...
+</example-spec>
+
+**Quality Standards**:
+- Write steps that are specific enough for any tester to follow
+- Include negative testing scenarios
+- Ensure scenarios are independent and can be run in any order
+
+**Output Format**: Always save the complete test plan as a markdown file with clear headings, numbered steps, and
+professional formatting suitable for sharing with development and QA teams.

web/.claude/skills/playwright

@@ -0,0 +1 @@
+../../../.cursor/skills/playwright

web/.mcp.json

@@ -0,0 +1,13 @@
+{
+  "mcpServers": {
+    "playwright": {
+      "type": "stdio",
+      "command": "npx",
+      "args": [
+        "-y",
+        "@playwright/mcp@latest"
+      ],
+      "env": {}
+    }
+  }
+}

web/tests/e2e/connectors/index_attempt_errors_modal.md

@@ -0,0 +1,875 @@
+# Index Attempt Errors Modal - Comprehensive Test Plan
+
+## Application Overview
+
+The Index Attempt Errors Modal is an admin-facing feature found on the connector detail page at
+`/admin/connector/[ccPairId]`. It surfaces document-level indexing failures for a given
+Connector-Credential Pair (CC Pair) and allows an admin to review them and trigger a full re-index
+to attempt resolution.
+
+### Feature Summary
+
+- **Entry point**: A yellow `Alert` banner on the connector detail page that reads "Some documents
+  failed to index" with a "View details." bold link. The banner appears only when
+  `indexAttemptErrors.total_items > 0`.
+- **Data fetching**: The parent page (`page.tsx`) fetches errors via `usePaginatedFetch` with
+  `itemsPerPage: 10, pagesPerBatch: 1`, polling every 5 seconds. Only the first 10 errors are
+  loaded into the modal. The modal receives these as `errors.items` and performs client-side
+  pagination over them.
+- **Modal title**: "Indexing Errors" with an `SvgAlertTriangle` icon.
+- **Table columns**: Time, Document ID (optionally hyperlinked), Error Message (scrollable cell at
+  60px height), Status (badge).
+- **Pagination**: Client-side within the modal. Page size is computed dynamically from the
+  container height via a `ResizeObserver` (minimum 3 rows per page). A `PageSelector` renders only
+  when `totalPages > 1`.
+- **Resolve All button**: In the modal footer. Rendered only when `hasUnresolvedErrors === true`
+  and `isResolvingErrors === false`. Clicking it: closes the modal, sets
+  `showIsResolvingKickoffLoader` to true, and awaits `triggerReIndex(fromBeginning = true)`.
+- **Spinner**: The full-screen `Spinner` is shown when `showIsResolvingKickoffLoader &&
+  !isResolvingErrors`. Once the backend index attempt transitions to `in_progress` / `not_started`
+  with `from_beginning = true`, `isResolvingErrors` becomes true and the spinner is hidden
+  regardless of `showIsResolvingKickoffLoader`.
+- **Resolving state**: While a full re-index initiated from the modal is running (latest index
+  attempt is `in_progress` or `not_started`, `from_beginning = true`, and none of the currently
+  loaded errors belong to that attempt), the banner switches to an animated "Resolving failures"
+  pulse and the modal header description changes.
+- **Access control**: The `/api/manage/admin/cc-pair/{id}/errors` endpoint requires
+  `current_curator_or_admin_user`.
+
+---
+
+## Important Implementation Details (Affecting Test Design)
+
+1. **10-error fetch limit**: The parent page only fetches up to 10 errors per poll cycle
+   (`itemsPerPage: 10`). The modal's client-side pagination operates on these 10 items, not on the
+   full database count. Testing large error counts via the UI requires either adjusting this limit
+   or using direct API calls.
+
+2. **Double-spinner invocation**: The `onResolveAll` handler in `page.tsx` sets
+   `showIsResolvingKickoffLoader(true)` before calling `triggerReIndex`, which itself also sets it
+   to `true`. The spinner correctly disappears when `triggerReIndex` resolves (via `finally`). This
+   is benign but worth noting for timing-sensitive tests.
+
+3. **isResolvingErrors logic**: `isResolvingErrors` is derived from `indexAttemptErrors.items`
+   (the 10-item fetch) and `latestIndexAttempt`. If any of the currently loaded errors have the
+   same `index_attempt_id` as the latest in-progress attempt, `isResolvingErrors` is `false` even
+   though a re-index is running.
+
+4. **PageSelector "unclickable" style**: The "‹" and "›" buttons use a `div` with
+   `unclickable` prop that adds `text-text-200` class and removes `cursor-pointer`. They are not
+   `<button disabled>` elements — they remain clickable in the DOM but navigation is guarded by
+   `Math.max`/`Math.min` clamps.
+
+5. **Alert uses dark: modifiers**: The banner component uses `dark:` Tailwind classes, which
+   contradicts the project's `colors.css` theming convention. This is an existing code issue and
+   not a test failure.
+
+---
+
+## Assumptions
+
+- All test scenarios begin from a fresh, clean state unless explicitly stated otherwise.
+- The test user is logged in as an admin (`a@example.com` / `a`).
+- A file connector is created via `OnyxApiClient.createFileConnector()` before each scenario that
+  needs one, and cleaned up in `afterEach` via `apiClient.deleteCCPair(testCcPairId)`.
+- Indexing errors are seeded directly via `psql` or a dedicated test API endpoint because they are
+  produced by the background indexing pipeline in production.
+- The connector detail page polls CC Pair data and errors every 5 seconds; tests that check
+  dynamic state must account for this polling interval (allow up to 10 seconds).
+
+---
+
+## Test Scenarios
+
+### 1. Alert Banner Visibility
+
+#### 1.1 No Errors - Banner Is Hidden
+
+**Seed:** Create a file connector with zero `IndexAttemptError` records.
+
+**Steps:**
+1. Navigate to `/admin/connector/{ccPairId}`.
+2. Wait for the page to finish loading (`networkidle`).
+3. Observe the area between the connector header and the "Indexing" section title.
+
+**Expected Results:**
+- The yellow alert banner ("Some documents failed to index") is not present in the DOM.
+- The "Indexing" section and its status card are visible.
+
+---
+
+#### 1.2 One or More Unresolved Errors - Banner Appears
+
+**Seed:** Create a file connector, then insert at least one `IndexAttemptError` row with
+`is_resolved = false` for that CC Pair.
+
+**Steps:**
+1. Navigate to `/admin/connector/{ccPairId}`.
+2. Wait for the page to finish loading.
+3. Observe the area above the "Indexing" section title.
+
+**Expected Results:**
+- The yellow alert banner is visible.
+- The banner heading reads "Some documents failed to index".
+- The banner body contains the text "We ran into some issues while processing some documents."
+- The text "View details." is rendered as a bold, clickable element within the banner body.
+
+---
+
+#### 1.3 All Errors Resolved - Banner Disappears Automatically
+
+**Seed:** Create a file connector with one `IndexAttemptError` where `is_resolved = false`.
+
+**Steps:**
+1. Navigate to `/admin/connector/{ccPairId}` and confirm the banner is visible.
+2. Using `psql` or a direct DB update, set `is_resolved = true` on that error record.
+3. Wait up to 10 seconds for the 5-second polling cycle to refresh the errors fetch.
+4. Observe the banner area.
+
+**Expected Results:**
+- The yellow alert banner disappears without a manual page reload.
+- No navigation or error occurs.
+
+---
+
+#### 1.4 Banner Absent for Invalid Connector With No Errors
+
+**Seed:** Create a file connector with zero errors. Manually put it into `INVALID` status via the
+DB.
+
+**Steps:**
+1. Navigate to `/admin/connector/{ccPairId}`.
+2. Observe both the "Invalid Connector State" callout and the banner area.
+
+**Expected Results:**
+- The "Invalid Connector State" warning callout is visible.
+- The yellow "Some documents failed to index" banner is absent.
+- The two alerts do not overlap.
+
+---
+
+### 2. Opening and Closing the Modal
+
+#### 2.1 Open Modal via "View Details" Link
+
+**Seed:** Create a file connector with one unresolved `IndexAttemptError`.
+
+**Steps:**
+1. Navigate to `/admin/connector/{ccPairId}`.
+2. Wait for the yellow alert banner to appear.
+3. Click the bold "View details." text within the banner.
+
+**Expected Results:**
+- A modal dialog appears with the title "Indexing Errors" and an alert-triangle icon.
+- The modal header has no description text (description is only shown in the resolving state).
+- The modal body shows the paragraph starting with "Below are the errors encountered during
+  indexing."
+- A second paragraph reads "Click the button below to kick off a full re-index..."
+- The table is visible with four column headers: Time, Document ID, Error Message, Status.
+- The one seeded error is displayed as a row.
+- The modal footer contains a "Resolve All" button.
+
+---
+
+#### 2.2 Close Modal via the X Button
+
+**Seed:** One unresolved `IndexAttemptError`.
+
+**Steps:**
+1. Open the Indexing Errors modal (scenario 2.1).
+2. Click the close (X) button in the modal header.
+
+**Expected Results:**
+- The modal closes and is no longer visible.
+- The connector detail page remains with the yellow alert banner still present.
+- No navigation occurs.
+
+---
+
+#### 2.3 Close Modal via Escape Key
+
+**Seed:** One unresolved `IndexAttemptError`.
+
+**Steps:**
+1. Open the Indexing Errors modal.
+2. Press the Escape key.
+
+**Expected Results:**
+- The modal closes.
+- The connector detail page remains intact with the banner still displayed.
+
+---
+
+#### 2.4 Close Modal via Backdrop Click
+
+**Seed:** One unresolved `IndexAttemptError`.
+
+**Steps:**
+1. Open the Indexing Errors modal.
+2. Click outside the modal content area on the dimmed backdrop.
+
+**Expected Results:**
+- The modal closes.
+- The connector detail page remains intact.
+
+---
+
+#### 2.5 Modal Cannot Be Opened When Errors Are Resolving
+
+**Seed:** Simulate `isResolvingErrors = true` by ensuring the latest index attempt has
+`status = in_progress`, `from_beginning = true`, and no currently loaded errors share its
+`index_attempt_id`.
+
+**Steps:**
+1. Navigate to `/admin/connector/{ccPairId}`.
+2. Observe the yellow banner in the resolving state.
+3. Confirm there is no "View details." link in the banner.
+
+**Expected Results:**
+- The banner body shows only the animated "Resolving failures" text (no "View details." link).
+- There is no interactive element in the banner to open the modal.
+
+---
+
+### 3. Table Content and Rendering
+
+#### 3.1 Single Error Row - All Fields Present
+
+**Seed:** Insert one `IndexAttemptError` with:
+- `document_id = "doc-123"`
+- `document_link = "https://example.com/doc-123"`
+- `failure_message = "Timeout while fetching document content"`
+- `is_resolved = false`
+- `time_created = <known timestamp>`
+
+**Steps:**
+1. Open the Indexing Errors modal.
+2. Inspect the single data row in the table.
+
+**Expected Results:**
+- The "Time" cell displays a human-readable, localized version of `time_created`.
+- The "Document ID" cell renders "doc-123" as an `<a>` element pointing to
+  `https://example.com/doc-123` with `target="_blank"` and `rel="noopener noreferrer"`.
+- The "Error Message" cell shows "Timeout while fetching document content" in a 60px-height
+  scrollable div.
+- The "Status" cell shows a badge with text "Unresolved" styled with red background.
+
+---
+
+#### 3.2 Error Without a Document Link - Plain Text ID
+
+**Seed:** Insert one `IndexAttemptError` with `document_id = "doc-no-link"` and
+`document_link = null`.
+
+**Steps:**
+1. Open the Indexing Errors modal.
+2. Inspect the Document ID cell.
+
+**Expected Results:**
+- The Document ID cell displays "doc-no-link" as plain text with no `<a>` element or underline.
+
+---
+
+#### 3.3 Error With Entity ID Instead of Document ID
+
+**Seed:** Insert one `IndexAttemptError` with `document_id = null`, `entity_id = "entity-abc"`,
+and `document_link = null`.
+
+**Steps:**
+1. Open the Indexing Errors modal.
+2. Inspect the Document ID cell.
+
+**Expected Results:**
+- The Document ID cell displays "entity-abc" as plain text (fallback to `entity_id` when
+  `document_id` is null and no link exists).
+
+---
+
+#### 3.4 Error With Document Link But No Document ID - Uses Entity ID in Link
+
+**Seed:** Insert one `IndexAttemptError` with `document_id = null`,
+`entity_id = "entity-link-test"`, and `document_link = "https://example.com/entity"`.
+
+**Steps:**
+1. Open the Indexing Errors modal.
+2. Inspect the Document ID cell.
+
+**Expected Results:**
+- The Document ID cell renders "entity-link-test" as a hyperlink pointing to
+  `https://example.com/entity`.
+- The link text is `entity_id` because `document_id` is null (code:
+  `error.document_id || error.entity_id || "Unknown"`).
+
+---
+
+#### 3.5 Error With Neither Document ID Nor Entity ID
+
+**Seed:** Insert one `IndexAttemptError` with `document_id = null` and `entity_id = null`.
+
+**Steps:**
+1. Open the Indexing Errors modal.
+2. Inspect the Document ID cell.
+
+**Expected Results:**
+- The Document ID cell displays the text "Unknown".
+
+---
+
+#### 3.6 Long Error Message Is Scrollable
+
+**Seed:** Insert one `IndexAttemptError` with a `failure_message` of at least 500 characters.
+
+**Steps:**
+1. Open the Indexing Errors modal.
+2. Locate the Error Message cell for that row.
+3. Attempt to scroll within the cell.
+
+**Expected Results:**
+- The Error Message cell's inner `div` is capped at 60px height with `overflow-y-auto`.
+- The cell content is scrollable, allowing the full message to be read.
+- The table row height does not expand beyond 60px.
+
+---
+
+#### 3.7 Error Message With Special HTML Characters Is Escaped
+
+**Seed:** Insert one `IndexAttemptError` with
+`failure_message = "<script>alert('xss')</script>"`.
+
+**Steps:**
+1. Open the Indexing Errors modal.
+2. Inspect the Error Message cell.
+
+**Expected Results:**
+- The text is rendered as a literal string, not interpreted as HTML.
+- No JavaScript alert dialog appears.
+- The exact text `<script>alert('xss')</script>` is visible as escaped content.
+
+---
+
+#### 3.8 Single-Character Error Message Does Not Break Layout
+
+**Seed:** Insert one `IndexAttemptError` with `failure_message = "X"`.
+
+**Steps:**
+1. Open the Indexing Errors modal.
+2. Inspect the Error Message cell and table row height.
+
+**Expected Results:**
+- The cell renders "X" without layout breakage.
+- The row height remains at 60px.
+
+---
+
+#### 3.9 Resolved Errors Are Filtered Out of Modal by Default
+
+**Seed:** Insert one `IndexAttemptError` with `is_resolved = true` and no unresolved errors.
+
+**Steps:**
+1. Make a direct API call: `GET /api/manage/admin/cc-pair/{ccPairId}/errors` (without
+   `include_resolved`).
+2. Make a second call: `GET /api/manage/admin/cc-pair/{ccPairId}/errors?include_resolved=true`.
+3. Navigate to `/admin/connector/{ccPairId}`.
+
+**Expected Results:**
+- The first API call returns zero items (`total_items = 0`).
+- The second API call returns the resolved error with `is_resolved: true`.
+- The yellow alert banner is absent from the connector page.
+- The modal cannot be opened through normal UI (no "View details." link).
+
+---
+
+### 4. Pagination
+
+#### 4.1 Single Page - No Pagination Controls
+
+**Seed:** Insert 3 unresolved errors (the minimum page size).
+
+**Steps:**
+1. Open the Indexing Errors modal.
+2. Observe whether a `PageSelector` is rendered below the table.
+
+**Expected Results:**
+- No `PageSelector` is rendered (`totalPages === 1`).
+- All 3 errors are visible simultaneously in the table.
+
+---
+
+#### 4.2 Multiple Pages - Pagination Controls Appear
+
+**Seed:** Insert 10 unresolved errors (matches the parent's `itemsPerPage: 10` fetch limit).
+The modal's dynamic page size is typically larger than 3 but can be forced small by using a
+narrow viewport.
+
+**Steps:**
+1. Set the browser viewport to a height that results in a page size smaller than 10 (e.g., a
+   height where the modal table container fits only 3 rows).
+2. Open the Indexing Errors modal.
+3. Observe the area below the table.
+
+**Expected Results:**
+- A `PageSelector` is rendered with "‹" and "›" navigation controls.
+- The current page indicator (page 1) is visually highlighted (active styling).
+- Only the rows for page 1 are shown in the table body.
+
+---
+
+#### 4.3 Navigate to Next Page
+
+**Seed:** Insert 10 unresolved errors and use a viewport that yields page size < 10.
+
+**Steps:**
+1. Open the Indexing Errors modal and confirm `PageSelector` is visible on page 1.
+2. Note the Document IDs visible on page 1.
+3. Click the "›" (next page) button.
+
+**Expected Results:**
+- The table updates to show errors for page 2.
+- The Document IDs on page 2 differ from those on page 1.
+- The page 2 indicator becomes highlighted.
+- The "‹" (previous page) button becomes clickable (no longer has `unclickable` styling).
+
+---
+
+#### 4.4 Navigate Back to Previous Page
+
+**Seed:** Same as 4.3.
+
+**Steps:**
+1. Open the modal and navigate to page 2 (scenario 4.3).
+2. Click the "‹" (previous page) button.
+
+**Expected Results:**
+- The table returns to showing page 1 errors.
+- The page 1 indicator is highlighted.
+- The "‹" button gains `unclickable` styling (lighter text, no pointer cursor).
+
+---
+
+#### 4.5 Previous Button Does Not Navigate Below Page 1
+
+**Seed:** Insert enough errors to produce at least 2 pages.
+
+**Steps:**
+1. Open the Indexing Errors modal. Confirm the current page is 1.
+2. Observe the "‹" button styling.
+3. Click the "‹" button.
+
+**Expected Results:**
+- The "‹" button has `text-text-200` styling (from `PageLink unclickable` prop) and no
+  `cursor-pointer`.
+- Note: The button is a `div`, not a `<button disabled>`. It remains clickable in the DOM, but the
+  handler clamps navigation to `Math.max(currentPage - 1, 1)`, so clicking it on page 1 has no
+  effect on the displayed rows.
+- The current page remains page 1.
+
+---
+
+#### 4.6 Next Button Does Not Navigate Beyond Last Page
+
+**Seed:** Insert exactly enough errors to produce 2 pages. Navigate to page 2.
+
+**Steps:**
+1. Open the modal and navigate to the last page.
+2. Observe the "›" button styling.
+3. Click the "›" button.
+
+**Expected Results:**
+- The "›" button has `unclickable` styling on the last page.
+- Clicking it does not navigate beyond the last page (clamped by `Math.min`).
+
+---
+
+#### 4.7 Page Resets to 1 When Error Count Changes
+
+**Seed:** Insert 10 errors. Use a small viewport so multiple pages exist. Navigate to page 2.
+
+**Steps:**
+1. Open the modal and navigate to page 2.
+2. While the modal is open, delete all error rows from the DB.
+3. Wait up to 10 seconds for the polling cycle to reload errors.
+
+**Expected Results:**
+- The modal's error table shows the empty-state message.
+- The `currentPage` state resets to 1 (triggered by the `useEffect` watching
+  `errors.items.length` and `errors.total_items`).
+- The `PageSelector` disappears if only one (empty) page remains.
+
+---
+
+#### 4.8 API-Level Pagination: page_size Parameter Maximum
+
+**Seed:** Insert 101 errors for the CC Pair.
+
+**Steps:**
+1. Make `GET /api/manage/admin/cc-pair/{ccPairId}/errors?page_size=100` as an authenticated admin.
+2. Make `GET /api/manage/admin/cc-pair/{ccPairId}/errors?page_size=101`.
+
+**Expected Results:**
+- The `page_size=100` request returns 100 items and `total_items = 101`.
+- The `page_size=101` request returns a 422 Unprocessable Entity error (backend enforces
+  `le=100`).
+
+---
+
+### 5. Resolve All Functionality
+
+#### 5.1 Resolve All Button Triggers Full Re-Index and Shows Spinner
+
+**Seed:** Create a file connector in ACTIVE status (not paused, not indexing, not invalid) with at
+least one unresolved `IndexAttemptError`.
+
+**Steps:**
+1. Open the Indexing Errors modal.
+2. Confirm the "Resolve All" button is visible in the modal footer.
+3. Click "Resolve All".
+
+**Expected Results:**
+- The modal closes immediately.
+- A full-screen `Spinner` component appears while the re-index request is in flight
+  (`showIsResolvingKickoffLoader = true` and `isResolvingErrors = false`).
+- A success toast notification appears: "Complete re-indexing started successfully".
+- The `Spinner` disappears after `triggerIndexing` resolves.
+- The connector detail page is still visible.
+- The yellow alert banner remains visible (errors are not immediately marked resolved; they resolve
+  as the re-index runs).
+
+---
+
+#### 5.2 Spinner Disappears Once Re-Index Is Picked Up
+
+**Seed:** Same as 5.1. The re-index task must be picked up by a running Celery worker.
+
+**Steps:**
+1. Click "Resolve All" (scenario 5.1).
+2. Wait for the Celery worker to start the index attempt (it will transition to `not_started` /
+   `in_progress` with `from_beginning = true`).
+3. Observe the spinner and the banner.
+
+**Expected Results:**
+- Once `isResolvingErrors` becomes `true` (the latest attempt is in-progress, from-beginning, and
+  none of the currently loaded errors belong to it), the spinner condition
+  `showIsResolvingKickoffLoader && !isResolvingErrors` becomes false, hiding the spinner.
+- The banner transitions to the "Resolving failures" pulse state.
+
+---
+
+#### 5.3 Resolve All Button Is Hidden When All Loaded Errors Are Resolved
+
+**Note:** Since `usePaginatedFetch` fetches only unresolved errors by default (no
+`include_resolved` param on the errors endpoint), zero unresolved errors means the banner is
+absent and the modal cannot be opened via the normal UI. This scenario therefore validates the
+banner-suppression mechanism:
+
+**Steps:**
+1. Ensure the CC Pair has zero unresolved `IndexAttemptError` records.
+2. Navigate to `/admin/connector/{ccPairId}`.
+3. Confirm the yellow alert banner is not present.
+
+**Expected Results:**
+- The yellow banner is absent.
+- The modal cannot be opened via the banner (no "View details." link).
+- The "Resolve All" button is therefore unreachable via the normal UI.
+
+---
+
+#### 5.4 Resolve All Button Is Hidden While Re-Index Is In Progress
+
+**Seed:** Trigger a full re-index (scenario 5.1) and re-open the modal while it is running.
+
+**Steps:**
+1. Trigger "Resolve All" (scenario 5.1).
+2. Immediately click "View details." in the banner to re-open the modal.
+3. Observe the modal while the re-index is `in_progress`.
+
+**Expected Results:**
+- The modal header description reads: "Currently attempting to resolve all errors by performing a
+  full re-index. This may take some time to complete."
+- The two explanatory body paragraphs ("Below are the errors..." and "Click the button below...") are
+  not visible.
+- The "Resolve All" button is not present in the footer.
+- The error rows are still displayed in the table.
+
+---
+
+#### 5.5 Resolve All Fails Gracefully When Connector Is Paused
+
+**Seed:** Create a file connector that is in PAUSED status with unresolved errors.
+
+**Steps:**
+1. Open the Indexing Errors modal.
+2. Confirm the "Resolve All" button is visible (it renders based on `hasUnresolvedErrors`, not on
+   connector status).
+3. Click "Resolve All".
+
+**Expected Results:**
+- The modal closes and the spinner appears briefly.
+- An error toast appears (because `triggerIndexing` returns an error message for paused
+  connectors).
+- The spinner disappears after the failed request.
+- The banner is still visible with the "View details." link.
+
+---
+
+### 6. Banner Resolving State
+
+#### 6.1 Banner Shows "Resolving Failures" While Re-Index Is In Progress
+
+**Seed:** Trigger a full re-index from the modal.
+
+**Steps:**
+1. Trigger "Resolve All".
+2. Return to the connector detail page without re-opening the modal.
+3. Observe the yellow alert banner body.
+
+**Expected Results:**
+- The banner body no longer shows "We ran into some issues..." or the "View details." link.
+- Instead, the banner body shows a pulsing "Resolving failures" span
+  (`animate-pulse` CSS class).
+- The banner heading still reads "Some documents failed to index".
+- The banner remains visible until the re-index completes and errors are resolved.
+
+---
+
+#### 6.2 Banner Reverts to "View Details" if Resolving Attempt Gains New Errors
+
+**Seed:** A re-index is in progress (`isResolvingErrors = true`). Force a new
+`IndexAttemptError` with `index_attempt_id` matching the running attempt.
+
+**Steps:**
+1. Observe the banner in "Resolving failures" state.
+2. Insert a new `IndexAttemptError` with `index_attempt_id = <running attempt id>` into the DB.
+3. Wait for the 5-second polling cycle.
+4. Observe the banner.
+
+**Expected Results:**
+- `isResolvingErrors` transitions back to `false` (the loaded errors now include one belonging to
+  the latest attempt).
+- The banner reverts to showing "We ran into some issues..." with the "View details." link.
+
+---
+
+### 7. Empty State
+
+#### 7.1 Empty Table When No Items on Current Page
+
+**Seed:** Construct a state where the `errors.items` array is empty (zero errors loaded for the
+CC Pair — this requires that `indexAttemptErrors.total_items > 0` to keep the banner visible, but
+a data race occurs where errors are deleted while the modal is open).
+
+**Steps:**
+1. Open the modal with some errors visible.
+2. Delete all error rows from the DB while the modal is open.
+3. Wait for the 5-second polling cycle.
+4. Observe the table body.
+
+**Expected Results:**
+- The table body shows a single row spanning all four columns with the text "No errors found on
+  this page" (centered, grayed-out `text-gray-500` style).
+- The `PageSelector` may disappear (if `totalPages` drops to 1).
+
+---
+
+### 8. Access Control
+
+#### 8.1 Non-Admin User Cannot Access the Errors API Endpoint
+
+**Seed:** Ensure a basic (non-admin, non-curator) user account exists.
+
+**Steps:**
+1. Log in as the basic user.
+2. Make a `GET /api/manage/admin/cc-pair/{ccPairId}/errors` request.
+
+**Expected Results:**
+- The API returns a 401 or 403 HTTP status code.
+- The basic user cannot access the connector detail admin page via the UI.
+
+---
+
+#### 8.2 Curator User Can Access the Errors Endpoint and Open Modal
+
+**Seed:** Ensure a curator user account exists and is assigned to the CC Pair's access group.
+
+**Steps:**
+1. Log in as the curator user.
+2. Navigate to `/admin/connector/{ccPairId}` for a connector the curator can edit.
+3. If unresolved errors exist, click "View details." to open the modal.
+
+**Expected Results:**
+- The modal opens successfully.
+- The errors table is populated.
+- The "Resolve All" button is visible (if unresolved errors exist and no re-index is running).
+
+---
+
+### 9. Data Freshness and Auto-Refresh
+
+#### 9.1 New Error Appears Without Page Reload
+
+**Seed:** Create a file connector with zero errors. Keep the connector detail page open.
+
+**Steps:**
+1. Navigate to `/admin/connector/{ccPairId}` and confirm no yellow banner.
+2. Insert a new `IndexAttemptError` with `is_resolved = false` via the DB.
+3. Wait up to 10 seconds (two 5-second polling cycles).
+4. Observe the page.
+
+**Expected Results:**
+- The yellow alert banner appears automatically without a manual page refresh.
+- No full page reload occurs.
+
+---
+
+#### 9.2 Errors List Refreshes While Modal Is Open
+
+**Seed:** Create a file connector with 2 unresolved errors.
+
+**Steps:**
+1. Open the Indexing Errors modal.
+2. Confirm 2 rows are visible.
+3. Insert a third `IndexAttemptError` via the DB.
+4. Wait up to 10 seconds for the polling cycle.
+5. Observe the modal table.
+
+**Expected Results:**
+- The third error row appears in the table without closing and reopening the modal.
+- If the total errors now exceed the current page size, the `PageSelector` appears if it was not
+  already present.
+
+---
+
+### 10. Page-Level Integration
+
+#### 10.1 Connector Detail Page Continues to Function With Modal Open
+
+**Seed:** Create a file connector with unresolved errors.
+
+**Steps:**
+1. Navigate to `/admin/connector/{ccPairId}`.
+2. Open the Indexing Errors modal.
+3. With the modal open, attempt to scroll the page behind the modal.
+4. Close the modal.
+5. Verify the page remains functional: the Indexing section status card is visible, and the
+   "Manage" dropdown button is present.
+
+**Expected Results:**
+- The modal renders over page content with a dimmed backdrop.
+- Scrolling behind the modal does not cause layout issues.
+- After closing the modal, all page elements are interactive and properly displayed.
+- No state corruption occurs.
+
+---
+
+#### 10.2 Other Manage Dropdown Actions Are Unaffected
+
+**Seed:** Create a file connector with unresolved errors.
+
+**Steps:**
+1. Navigate to `/admin/connector/{ccPairId}`.
+2. Open the Indexing Errors modal and close it without clicking "Resolve All".
+3. Open the "Manage" dropdown.
+4. Confirm the "Re-Index", "Pause", and "Delete" items are visible and have the correct enabled
+   state.
+
+**Expected Results:**
+- The Manage dropdown functions normally after interacting with the errors modal.
+- No state from the modal (e.g., lingering `showIsResolvingKickoffLoader`) affects the dropdown.
+
+---
+
+### 11. Boundary Conditions
+
+#### 11.1 Minimum Page Size of 3 Rows
+
+**Seed:** Insert enough errors to exceed page size.
+
+**Steps:**
+1. Set the browser viewport to a very small height (e.g., 300px total).
+2. Open the Indexing Errors modal.
+3. Observe the number of rows rendered per page.
+
+**Expected Results:**
+- The page size does not drop below 3 rows (enforced by `Math.max(3, ...)` in the `ResizeObserver`
+  callback).
+- At least 3 rows are displayed per page regardless of available container height.
+
+---
+
+#### 11.2 Exactly the API Page Limit (10 Items) Displayed
+
+**Seed:** Insert exactly 10 unresolved errors for the CC Pair.
+
+**Steps:**
+1. Open the Indexing Errors modal with a sufficiently large viewport.
+2. Observe that all 10 errors are visible (assuming the dynamic page size is >= 10).
+
+**Expected Results:**
+- All 10 errors are accessible in the modal.
+- No pagination is needed if the computed page size is >= 10.
+- Note: If 11+ errors exist in the DB, only the first 10 (from `usePaginatedFetch`) are surfaced
+  in the modal. The 11th error would require a separate API call or a larger `itemsPerPage` config
+  to verify.
+
+---
+
+#### 11.3 Modal Opens Only When indexAttemptErrors Is Non-Null
+
+**Steps:**
+1. Observe the condition in `page.tsx`: `{showIndexAttemptErrors && indexAttemptErrors && ...}`.
+2. During the initial page load (before the first poll completes), `indexAttemptErrors` is `null`.
+
+**Expected Results:**
+- Clicking "View details." while `indexAttemptErrors` is still null has no effect
+  (`setShowIndexAttemptErrors(true)` is called but the modal renders only when both
+  `showIndexAttemptErrors` and `indexAttemptErrors` are truthy).
+- Once the first poll completes and errors are available, the modal renders normally.
+
+---
+
+## Test File Location
+
+These tests should be implemented as Playwright E2E specs at:
+
+```
+web/tests/e2e/connectors/index_attempt_errors_modal.spec.ts
+```
+
+### Recommended OnyxApiClient Additions
+
+The following methods should be added to `web/tests/e2e/utils/onyxApiClient.ts` to support
+seeding and cleanup:
+
+- `createIndexAttemptError(ccPairId, options)` - inserts an error record via `psql` or a dedicated
+  test endpoint; options include `documentId`, `documentLink`, `entityId`, `failureMessage`,
+  `isResolved`, `indexAttemptId`.
+- `resolveAllIndexAttemptErrors(ccPairId)` - marks all errors for a CC Pair as resolved.
+- `getIndexAttemptErrors(ccPairId, includeResolved?)` - calls the errors API and returns the
+  parsed response.
+
+### Cleanup Strategy
+
+Each test must clean up its CC Pair in an `afterEach` hook:
+
+```typescript
+test.afterEach(async ({ page }) => {
+  const apiClient = new OnyxApiClient(page.request);
+  if (testCcPairId !== null) {
+    await apiClient.deleteCCPair(testCcPairId);
+    testCcPairId = null;
+  }
+});
+```
+
+Cascade deletes on the CC Pair will remove associated `IndexAttemptError` rows automatically.
+
+### Polling Guidance
+
+For scenarios that require waiting for auto-refresh to propagate state changes, use
+`expect.poll()` with a 10-second timeout to avoid flaky tests:
+
+```typescript
+await expect.poll(
+  async () => page.locator('[data-testid="error-banner"]').isVisible(),
+  { timeout: 10000 }
+).toBe(true);
+```

web/tests/e2e/connectors/index_attempt_errors_modal.spec.ts

@@ -0,0 +1,705 @@
+import { test, expect } from "@playwright/test";
+import { loginAs } from "@tests/e2e/utils/auth";
+import { OnyxApiClient } from "@tests/e2e/utils/onyxApiClient";
+import { execSync } from "child_process";
+
+// ─── Database Helpers ─────────────────────────────────────────────────────────
+// IndexAttemptError rows are produced by background workers in production.
+// In tests we seed them directly via psql since there is no public API for it.
+
+const DB_CONTAINER = process.env.DB_CONTAINER || "onyx-relational_db-1";
+
+function psql(sql: string): string {
+  return execSync(`docker exec -i ${DB_CONTAINER} psql -U postgres -t -A`, {
+    input: sql,
+    encoding: "utf-8",
+  }).trim();
+}
+
+function getSearchSettingsId(): number {
+  const result = psql(
+    "SELECT id FROM search_settings WHERE status = 'PRESENT' ORDER BY id DESC LIMIT 1;"
+  );
+  const id = parseInt(result, 10);
+  if (isNaN(id)) {
+    throw new Error(
+      `No search_settings with status PRESENT found: "${result}"`
+    );
+  }
+  return id;
+}
+
+function createIndexAttempt(
+  ccPairId: number,
+  options: { fromBeginning?: boolean; status?: string } = {}
+): number {
+  const { fromBeginning = false, status = "success" } = options;
+  const searchSettingsId = getSearchSettingsId();
+  const result = psql(
+    `INSERT INTO index_attempt (
+       connector_credential_pair_id, from_beginning, status, search_settings_id,
+       time_created, time_started, time_updated
+     ) VALUES (
+       ${ccPairId}, ${fromBeginning}, '${status}', ${searchSettingsId},
+       NOW(), NOW(), NOW()
+     ) RETURNING id;`
+  );
+  const id = parseInt(result, 10);
+  if (isNaN(id)) {
+    throw new Error(`Failed to create index attempt: "${result}"`);
+  }
+  return id;
+}
+
+function sqlVal(v: string | null): string {
+  return v === null ? "NULL" : `'${v.replace(/'/g, "''")}'`;
+}
+
+interface CreateErrorOptions {
+  indexAttemptId: number;
+  ccPairId: number;
+  documentId?: string | null;
+  documentLink?: string | null;
+  entityId?: string | null;
+  failureMessage?: string;
+  isResolved?: boolean;
+}
+
+function createError(options: CreateErrorOptions): number {
+  const {
+    indexAttemptId,
+    ccPairId,
+    documentId = null,
+    documentLink = null,
+    entityId = null,
+    failureMessage = "Test indexing error",
+    isResolved = false,
+  } = options;
+
+  const result = psql(
+    `INSERT INTO index_attempt_errors (
+       index_attempt_id, connector_credential_pair_id,
+       document_id, document_link, entity_id,
+       failure_message, is_resolved, time_created
+     ) VALUES (
+       ${indexAttemptId}, ${ccPairId},
+       ${sqlVal(documentId)}, ${sqlVal(documentLink)}, ${sqlVal(entityId)},
+       ${sqlVal(failureMessage)}, ${isResolved}, NOW()
+     ) RETURNING id;`
+  );
+  const id = parseInt(result, 10);
+  if (isNaN(id)) {
+    throw new Error(`Failed to create index attempt error: "${result}"`);
+  }
+  return id;
+}
+
+function createMultipleErrors(
+  indexAttemptId: number,
+  ccPairId: number,
+  count: number
+): number[] {
+  const ids: number[] = [];
+  for (let i = 0; i < count; i++) {
+    ids.push(
+      createError({
+        indexAttemptId,
+        ccPairId,
+        documentId: `doc-${i + 1}`,
+        failureMessage: `Error #${i + 1}: Failed to index document`,
+      })
+    );
+  }
+  return ids;
+}
+
+function resolveAllErrors(ccPairId: number): void {
+  psql(
+    `UPDATE index_attempt_errors SET is_resolved = true
+     WHERE connector_credential_pair_id = ${ccPairId};`
+  );
+}
+
+// ─── Shared UI Helpers ────────────────────────────────────────────────────────
+
+async function waitForBanner(page: import("@playwright/test").Page) {
+  await expect(page.getByText("Some documents failed to index")).toBeVisible({
+    timeout: 15000,
+  });
+}
+
+async function openErrorsModal(page: import("@playwright/test").Page) {
+  await waitForBanner(page);
+  await page.getByText("View details.").click();
+  await expect(page.getByText("Indexing Errors")).toBeVisible();
+}
+
+// ─── Tests ────────────────────────────────────────────────────────────────────
+
+test.describe("Index Attempt Errors Modal", () => {
+  test.describe.configure({ retries: 2 });
+
+  let testCcPairId: number | null = null;
+  let testIndexAttemptId: number | null = null;
+
+  test.beforeEach(async ({ page }) => {
+    await page.context().clearCookies();
+    await loginAs(page, "admin");
+
+    const apiClient = new OnyxApiClient(page.request);
+    testCcPairId = await apiClient.createFileConnector(
+      `Error Modal Test ${Date.now()}`
+    );
+    testIndexAttemptId = createIndexAttempt(testCcPairId);
+  });
+
+  test.afterEach(async ({ page }) => {
+    if (testCcPairId !== null) {
+      const apiClient = new OnyxApiClient(page.request);
+      try {
+        await apiClient.pauseConnector(testCcPairId);
+      } catch {
+        // May already be paused
+      }
+      try {
+        await apiClient.deleteCCPair(testCcPairId);
+      } catch (error) {
+        console.warn(`Cleanup failed for CC pair ${testCcPairId}: ${error}`);
+      }
+      testCcPairId = null;
+      testIndexAttemptId = null;
+    }
+  });
+
+  // ── 1. Alert Banner Visibility ────────────────────────────────────────────
+
+  test("1.1 banner is hidden when no errors exist", async ({ page }) => {
+    await page.goto(`/admin/connector/${testCcPairId}`);
+    await page.waitForLoadState("networkidle");
+
+    await expect(
+      page.getByText("Some documents failed to index")
+    ).not.toBeVisible();
+  });
+
+  test("1.2 banner appears when unresolved errors exist", async ({ page }) => {
+    createError({
+      indexAttemptId: testIndexAttemptId!,
+      ccPairId: testCcPairId!,
+      documentId: "doc-banner-test",
+      failureMessage: "Test error for banner visibility",
+    });
+
+    await page.goto(`/admin/connector/${testCcPairId}`);
+    await page.waitForLoadState("networkidle");
+
+    await waitForBanner(page);
+    await expect(
+      page.getByText("We ran into some issues while processing some documents.")
+    ).toBeVisible();
+    await expect(page.getByText("View details.")).toBeVisible();
+  });
+
+  test("1.3 banner disappears when all errors are resolved", async ({
+    page,
+  }) => {
+    createError({
+      indexAttemptId: testIndexAttemptId!,
+      ccPairId: testCcPairId!,
+      failureMessage: "Error to be resolved",
+    });
+
+    await page.goto(`/admin/connector/${testCcPairId}`);
+    await page.waitForLoadState("networkidle");
+    await waitForBanner(page);
+
+    // Resolve all errors via DB
+    resolveAllErrors(testCcPairId!);
+
+    // Wait for the 5-second polling cycle to pick up the change
+    await expect(
+      page.getByText("Some documents failed to index")
+    ).not.toBeVisible({ timeout: 15000 });
+  });
+
+  // ── 2. Opening and Closing the Modal ──────────────────────────────────────
+
+  test("2.1 modal opens via View details link with correct content", async ({
+    page,
+  }) => {
+    createError({
+      indexAttemptId: testIndexAttemptId!,
+      ccPairId: testCcPairId!,
+      documentId: "doc-modal-open",
+      failureMessage: "Error for modal open test",
+    });
+
+    await page.goto(`/admin/connector/${testCcPairId}`);
+    await page.waitForLoadState("networkidle");
+    await openErrorsModal(page);
+
+    // Explanatory text
+    await expect(
+      page.getByText("Below are the errors encountered during indexing.")
+    ).toBeVisible();
+    await expect(
+      page.getByText(
+        "Click the button below to kick off a full re-index to try and resolve these errors."
+      )
+    ).toBeVisible();
+
+    // Table headers
+    await expect(
+      page.getByRole("columnheader", { name: "Time" })
+    ).toBeVisible();
+    await expect(
+      page.getByRole("columnheader", { name: "Document ID" })
+    ).toBeVisible();
+    await expect(
+      page.getByRole("columnheader", { name: "Error Message" })
+    ).toBeVisible();
+    await expect(
+      page.getByRole("columnheader", { name: "Status" })
+    ).toBeVisible();
+
+    // Error row content
+    await expect(page.getByText("doc-modal-open")).toBeVisible();
+    await expect(page.getByText("Error for modal open test")).toBeVisible();
+    await expect(page.getByText("Unresolved")).toBeVisible();
+
+    // Resolve All button
+    await expect(
+      page.getByRole("button", { name: "Resolve All" })
+    ).toBeVisible();
+  });
+
+  test("2.2 modal closes via X button", async ({ page }) => {
+    createError({
+      indexAttemptId: testIndexAttemptId!,
+      ccPairId: testCcPairId!,
+      failureMessage: "Error for close-X test",
+    });
+
+    await page.goto(`/admin/connector/${testCcPairId}`);
+    await page.waitForLoadState("networkidle");
+    await openErrorsModal(page);
+
+    // The close button is the first <button> in the dialog (rendered in Modal.Header)
+    // "Resolve All" is in the footer, so .first() gets the X button
+    const dialog = page.getByRole("dialog");
+    await dialog
+      .locator("button")
+      .filter({ hasNotText: /Resolve All/ })
+      .first()
+      .click();
+
+    await expect(page.getByText("Indexing Errors")).not.toBeVisible();
+    // Banner should still be present
+    await expect(
+      page.getByText("Some documents failed to index")
+    ).toBeVisible();
+  });
+
+  test("2.3 modal closes via Escape key", async ({ page }) => {
+    createError({
+      indexAttemptId: testIndexAttemptId!,
+      ccPairId: testCcPairId!,
+      failureMessage: "Error for close-escape test",
+    });
+
+    await page.goto(`/admin/connector/${testCcPairId}`);
+    await page.waitForLoadState("networkidle");
+    await openErrorsModal(page);
+
+    await page.keyboard.press("Escape");
+
+    await expect(page.getByText("Indexing Errors")).not.toBeVisible();
+    await expect(
+      page.getByText("Some documents failed to index")
+    ).toBeVisible();
+  });
+
+  // ── 3. Table Content and Rendering ────────────────────────────────────────
+
+  test("3.1 error row with document link renders as hyperlink", async ({
+    page,
+  }) => {
+    createError({
+      indexAttemptId: testIndexAttemptId!,
+      ccPairId: testCcPairId!,
+      documentId: "doc-linked",
+      documentLink: "https://example.com/doc-linked",
+      failureMessage: "Timeout while fetching document content",
+    });
+
+    await page.goto(`/admin/connector/${testCcPairId}`);
+    await page.waitForLoadState("networkidle");
+    await openErrorsModal(page);
+
+    const docLink = page.getByRole("link", { name: "doc-linked" });
+    await expect(docLink).toBeVisible();
+    await expect(docLink).toHaveAttribute(
+      "href",
+      "https://example.com/doc-linked"
+    );
+    await expect(docLink).toHaveAttribute("target", "_blank");
+    await expect(docLink).toHaveAttribute("rel", "noopener noreferrer");
+
+    await expect(
+      page.getByText("Timeout while fetching document content")
+    ).toBeVisible();
+    await expect(page.getByText("Unresolved")).toBeVisible();
+  });
+
+  test("3.2 error without document link shows plain text ID", async ({
+    page,
+  }) => {
+    createError({
+      indexAttemptId: testIndexAttemptId!,
+      ccPairId: testCcPairId!,
+      documentId: "doc-no-link",
+      documentLink: null,
+      failureMessage: "Error without link",
+    });
+
+    await page.goto(`/admin/connector/${testCcPairId}`);
+    await page.waitForLoadState("networkidle");
+    await openErrorsModal(page);
+
+    await expect(page.getByText("doc-no-link")).toBeVisible();
+    // Should NOT be a link
+    await expect(page.getByRole("link", { name: "doc-no-link" })).toHaveCount(
+      0
+    );
+  });
+
+  test("3.3 error with entity ID fallback when no document ID", async ({
+    page,
+  }) => {
+    createError({
+      indexAttemptId: testIndexAttemptId!,
+      ccPairId: testCcPairId!,
+      documentId: null,
+      entityId: "entity-abc",
+      failureMessage: "Error with entity ID only",
+    });
+
+    await page.goto(`/admin/connector/${testCcPairId}`);
+    await page.waitForLoadState("networkidle");
+    await openErrorsModal(page);
+
+    await expect(page.getByText("entity-abc")).toBeVisible();
+  });
+
+  test("3.4 error with no document ID or entity ID shows Unknown", async ({
+    page,
+  }) => {
+    createError({
+      indexAttemptId: testIndexAttemptId!,
+      ccPairId: testCcPairId!,
+      documentId: null,
+      entityId: null,
+      failureMessage: "Error with no identifiers",
+    });
+
+    await page.goto(`/admin/connector/${testCcPairId}`);
+    await page.waitForLoadState("networkidle");
+    await openErrorsModal(page);
+
+    // The table cell should display "Unknown" as fallback
+    const dialog = page.getByRole("dialog");
+    await expect(dialog.getByText("Unknown")).toBeVisible();
+  });
+
+  test("3.5 entity ID used as link text when document link exists but no document ID", async ({
+    page,
+  }) => {
+    createError({
+      indexAttemptId: testIndexAttemptId!,
+      ccPairId: testCcPairId!,
+      documentId: null,
+      entityId: "entity-link-test",
+      documentLink: "https://example.com/entity",
+      failureMessage: "Error with entity link",
+    });
+
+    await page.goto(`/admin/connector/${testCcPairId}`);
+    await page.waitForLoadState("networkidle");
+    await openErrorsModal(page);
+
+    const link = page.getByRole("link", { name: "entity-link-test" });
+    await expect(link).toBeVisible();
+    await expect(link).toHaveAttribute("href", "https://example.com/entity");
+  });
+
+  test("3.6 HTML in error message is escaped (XSS safe)", async ({ page }) => {
+    createError({
+      indexAttemptId: testIndexAttemptId!,
+      ccPairId: testCcPairId!,
+      documentId: "doc-xss",
+      failureMessage: "<script>alert('xss')</script>",
+    });
+
+    await page.goto(`/admin/connector/${testCcPairId}`);
+    await page.waitForLoadState("networkidle");
+    await openErrorsModal(page);
+
+    // Text should be rendered literally, not executed as HTML
+    await expect(page.getByText("<script>alert('xss')</script>")).toBeVisible();
+  });
+
+  // ── 4. Pagination ─────────────────────────────────────────────────────────
+
+  test("4.1 no pagination controls when errors fit on one page", async ({
+    page,
+  }) => {
+    createMultipleErrors(testIndexAttemptId!, testCcPairId!, 2);
+
+    await page.goto(`/admin/connector/${testCcPairId}`);
+    await page.waitForLoadState("networkidle");
+    await openErrorsModal(page);
+
+    // Both errors should be visible
+    await expect(page.getByText("doc-1")).toBeVisible();
+    await expect(page.getByText("doc-2")).toBeVisible();
+
+    // PageSelector should not appear (only renders when totalPages > 1)
+    // The "›" next-page button only exists when pagination is shown
+    const dialog = page.getByRole("dialog");
+    await expect(dialog.locator('text="›"')).not.toBeVisible();
+  });
+
+  test("4.2 pagination appears and navigation works with many errors", async ({
+    page,
+  }) => {
+    // 10 errors should span multiple pages given the modal's dynamic page size
+    // (viewport 1280x720 typically yields ~5 rows per page in the modal)
+    createMultipleErrors(testIndexAttemptId!, testCcPairId!, 10);
+
+    await page.goto(`/admin/connector/${testCcPairId}`);
+    await page.waitForLoadState("networkidle");
+    await openErrorsModal(page);
+
+    const dialog = page.getByRole("dialog");
+    const nextBtn = dialog.locator('text="›"');
+    const prevBtn = dialog.locator('text="‹"');
+
+    // If the viewport produces a page size >= 10, pagination won't appear
+    // Skip the navigation part in that case
+    if (await nextBtn.isVisible()) {
+      // Record page 1 content
+      const page1Content = await dialog.locator("table tbody").textContent();
+
+      // Navigate to page 2
+      await nextBtn.click();
+      const page2Content = await dialog.locator("table tbody").textContent();
+      expect(page2Content).not.toBe(page1Content);
+
+      // Navigate back to page 1
+      await prevBtn.click();
+      const backToPage1 = await dialog.locator("table tbody").textContent();
+      expect(backToPage1).toBe(page1Content);
+    }
+  });
+
+  // ── 5. Resolve All Functionality ──────────────────────────────────────────
+
+  test("5.1 Resolve All triggers re-index and shows success toast", async ({
+    page,
+  }) => {
+    createError({
+      indexAttemptId: testIndexAttemptId!,
+      ccPairId: testCcPairId!,
+      documentId: "doc-resolve",
+      failureMessage: "Error to resolve via re-index",
+    });
+
+    // Activate the connector so the re-index request can succeed
+    // (createFileConnector pauses the connector by default)
+    await page.request.put(`/api/manage/admin/cc-pair/${testCcPairId}/status`, {
+      data: { status: "ACTIVE" },
+    });
+
+    await page.goto(`/admin/connector/${testCcPairId}`);
+    await page.waitForLoadState("networkidle");
+    await openErrorsModal(page);
+
+    await page.getByRole("button", { name: "Resolve All" }).click();
+
+    // Modal should close
+    await expect(page.getByText("Indexing Errors")).not.toBeVisible();
+
+    // Success toast should appear
+    await expect(
+      page.getByText("Complete re-indexing started successfully")
+    ).toBeVisible({ timeout: 15000 });
+  });
+
+  test("5.2 Resolve All button is absent when isResolvingErrors is true", async ({
+    page,
+  }) => {
+    createError({
+      indexAttemptId: testIndexAttemptId!,
+      ccPairId: testCcPairId!,
+      documentId: "doc-resolving",
+      failureMessage: "Error during resolving state",
+    });
+
+    // Create a separate index attempt that simulates a from-beginning re-index
+    // in progress, with no errors belonging to it
+    createIndexAttempt(testCcPairId!, {
+      fromBeginning: true,
+      status: "in_progress",
+    });
+
+    await page.goto(`/admin/connector/${testCcPairId}`);
+    await page.waitForLoadState("networkidle");
+
+    // The banner should show "Resolving failures" instead of "View details."
+    await expect(page.getByText("Some documents failed to index")).toBeVisible({
+      timeout: 15000,
+    });
+    await expect(page.getByText("Resolving failures")).toBeVisible();
+    await expect(page.getByText("View details.")).not.toBeVisible();
+  });
+
+  // ── 6. Data Freshness and Auto-Refresh ────────────────────────────────────
+
+  test("6.1 new error appears on page without manual reload", async ({
+    page,
+  }) => {
+    await page.goto(`/admin/connector/${testCcPairId}`);
+    await page.waitForLoadState("networkidle");
+
+    // Initially no banner
+    await expect(
+      page.getByText("Some documents failed to index")
+    ).not.toBeVisible();
+
+    // Insert an error via DB while the page is already open
+    createError({
+      indexAttemptId: testIndexAttemptId!,
+      ccPairId: testCcPairId!,
+      documentId: "doc-auto-refresh",
+      failureMessage: "Error added while page is open",
+    });
+
+    // Wait for the 5-second polling cycle to pick it up
+    await expect(page.getByText("Some documents failed to index")).toBeVisible({
+      timeout: 15000,
+    });
+  });
+
+  test("6.2 errors list refreshes while modal is open", async ({ page }) => {
+    createError({
+      indexAttemptId: testIndexAttemptId!,
+      ccPairId: testCcPairId!,
+      documentId: "doc-existing",
+      failureMessage: "Pre-existing error",
+    });
+
+    await page.goto(`/admin/connector/${testCcPairId}`);
+    await page.waitForLoadState("networkidle");
+    await openErrorsModal(page);
+
+    await expect(page.getByText("doc-existing")).toBeVisible();
+
+    // Add a second error while the modal is open
+    createError({
+      indexAttemptId: testIndexAttemptId!,
+      ccPairId: testCcPairId!,
+      documentId: "doc-new-while-open",
+      failureMessage: "Error added while modal is open",
+    });
+
+    // The new error should appear after the polling cycle
+    await expect(page.getByText("doc-new-while-open")).toBeVisible({
+      timeout: 15000,
+    });
+  });
+
+  // ── 7. Access Control ─────────────────────────────────────────────────────
+
+  test("7.1 non-admin user cannot access the errors API endpoint", async ({
+    page,
+  }) => {
+    // Register a basic (non-admin) user
+    const email = `basic_${Date.now()}@example.com`;
+    const password = "TestPassword123!";
+
+    await page.request.post("/api/auth/register", {
+      data: { email, username: email, password },
+    });
+
+    // Login as the basic user
+    await page.context().clearCookies();
+    await page.request.post("/api/auth/login", {
+      form: { username: email, password },
+    });
+
+    // Try to access the errors endpoint
+    const errorsRes = await page.request.get(
+      `/api/manage/admin/cc-pair/${testCcPairId}/errors`
+    );
+    expect([401, 403]).toContain(errorsRes.status());
+
+    // Re-login as admin for afterEach cleanup
+    await page.context().clearCookies();
+    await loginAs(page, "admin");
+
+    // Clean up the basic user
+    const apiClient = new OnyxApiClient(page.request);
+    try {
+      await apiClient.deleteUser(email);
+    } catch {
+      // Ignore cleanup failures
+    }
+  });
+
+  // ── 8. Resolved Errors Filtered by Default ────────────────────────────────
+
+  test("8.1 resolved errors are not shown in the modal and banner is absent", async ({
+    page,
+  }) => {
+    createError({
+      indexAttemptId: testIndexAttemptId!,
+      ccPairId: testCcPairId!,
+      documentId: "doc-resolved",
+      failureMessage: "Already resolved error",
+      isResolved: true,
+    });
+
+    // API without include_resolved should return 0 items
+    const defaultRes = await page.request.get(
+      `/api/manage/admin/cc-pair/${testCcPairId}/errors`
+    );
+    expect(defaultRes.ok()).toBe(true);
+    const defaultData = await defaultRes.json();
+    expect(defaultData.total_items).toBe(0);
+
+    // API with include_resolved=true should return the error
+    const resolvedRes = await page.request.get(
+      `/api/manage/admin/cc-pair/${testCcPairId}/errors?include_resolved=true`
+    );
+    expect(resolvedRes.ok()).toBe(true);
+    const resolvedData = await resolvedRes.json();
+    expect(resolvedData.total_items).toBe(1);
+    expect(resolvedData.items[0].is_resolved).toBe(true);
+
+    // Banner should not appear on the page
+    await page.goto(`/admin/connector/${testCcPairId}`);
+    await page.waitForLoadState("networkidle");
+    await expect(
+      page.getByText("Some documents failed to index")
+    ).not.toBeVisible();
+  });
+
+  // ── 9. API Pagination Boundary ────────────────────────────────────────────
+
+  test("9.1 API rejects page_size over 100", async ({ page }) => {
+    const res = await page.request.get(
+      `/api/manage/admin/cc-pair/${testCcPairId}/errors?page_size=101`
+    );
+    expect(res.status()).toBe(422);
+  });
+});

web/tests/e2e/connectors/seed_index_attempt_errors.sh

@@ -0,0 +1,210 @@
+#!/usr/bin/env bash
+#
+# Seed IndexAttemptError records for local testing of the Index Attempt Errors Modal.
+#
+# Usage:
+#   ./seed_index_attempt_errors.sh [--cc-pair-id <ID>] [--count <N>] [--clean]
+#
+# Options:
+#   --cc-pair-id <ID>   Use an existing CC pair (skips connector creation)
+#   --count <N>         Number of unresolved errors to insert (default: 7)
+#   --clean             Remove ALL test-seeded errors (those with failure_message LIKE 'SEED:%') and exit
+#
+# Without --cc-pair-id, the script creates a file connector via the API
+# and prints its CC pair ID so you can navigate to /admin/connector/<ID>.
+#
+# Prerequisites:
+#   - Onyx services running (docker compose up)
+#   - curl and jq installed
+
+set -euo pipefail
+
+BASE_URL="${BASE_URL:-http://localhost:3000}"
+ADMIN_EMAIL="${ADMIN_EMAIL:-admin_user@example.com}"
+ADMIN_PASSWORD="${ADMIN_PASSWORD:-TestPassword123!}"
+DB_CONTAINER="${DB_CONTAINER:-onyx-relational_db-1}"
+CC_PAIR_ID=""
+ERROR_COUNT=7
+CLEAN=false
+
+# --- Parse args ---
+while [[ $# -gt 0 ]]; do
+  case "$1" in
+    --cc-pair-id) CC_PAIR_ID="$2"; shift 2 ;;
+    --count)      ERROR_COUNT="$2"; shift 2 ;;
+    --clean)      CLEAN=true; shift ;;
+    *)            echo "Unknown option: $1" >&2; exit 1 ;;
+  esac
+done
+
+# --- Helper: run psql ---
+psql_exec() {
+  docker exec "$DB_CONTAINER" psql -U postgres -qtAX -c "$1"
+}
+
+# --- Clean mode ---
+if $CLEAN; then
+  deleted=$(psql_exec "DELETE FROM index_attempt_errors WHERE failure_message LIKE 'SEED:%' RETURNING id;" | wc -l)
+  echo "Deleted $deleted seeded error(s)."
+  exit 0
+fi
+
+# --- Authenticate and get session cookie ---
+COOKIE_JAR=$(mktemp)
+trap 'rm -f "$COOKIE_JAR"' EXIT
+
+echo "Authenticating as $ADMIN_EMAIL..."
+login_resp=$(curl -s -o /dev/null -w "%{http_code}" \
+  -c "$COOKIE_JAR" \
+  -X POST "$BASE_URL/api/auth/login" \
+  -H "Content-Type: application/x-www-form-urlencoded" \
+  -d "username=${ADMIN_EMAIL}&password=${ADMIN_PASSWORD}")
+
+if [[ "$login_resp" != "200" && "$login_resp" != "204" && "$login_resp" != "302" ]]; then
+  echo "Login failed (HTTP $login_resp). Check credentials." >&2
+  # Try the simpler a@example.com / a creds as fallback
+  echo "Retrying with a@example.com / a..."
+  ADMIN_EMAIL="a@example.com"
+  ADMIN_PASSWORD="a"
+  login_resp=$(curl -s -o /dev/null -w "%{http_code}" \
+    -c "$COOKIE_JAR" \
+    -X POST "$BASE_URL/api/auth/login" \
+    -H "Content-Type: application/x-www-form-urlencoded" \
+    -d "username=${ADMIN_EMAIL}&password=${ADMIN_PASSWORD}")
+  if [[ "$login_resp" != "200" && "$login_resp" != "204" && "$login_resp" != "302" ]]; then
+    echo "Login failed again (HTTP $login_resp)." >&2
+    exit 1
+  fi
+fi
+echo "Authenticated."
+
+# --- Create a file connector if no CC pair specified ---
+if [[ -z "$CC_PAIR_ID" ]]; then
+  echo "Creating file connector..."
+  create_resp=$(curl -s -b "$COOKIE_JAR" \
+    -X POST "$BASE_URL/api/manage/admin/connector-with-mock-credential" \
+    -H "Content-Type: application/json" \
+    -d '{
+      "name": "Seed Errors Test Connector",
+      "source": "file",
+      "input_type": "load_state",
+      "connector_specific_config": {"file_locations": []},
+      "refresh_freq": null,
+      "prune_freq": null,
+      "indexing_start": null,
+      "access_type": "public",
+      "groups": []
+    }')
+
+  CC_PAIR_ID=$(echo "$create_resp" | jq -r '.data // empty')
+  if [[ -z "$CC_PAIR_ID" ]]; then
+    echo "Failed to create connector: $create_resp" >&2
+    exit 1
+  fi
+  echo "Created CC pair ID: $CC_PAIR_ID"
+else
+  echo "Using existing CC pair ID: $CC_PAIR_ID"
+fi
+
+# --- Find or create an index attempt for this CC pair ---
+ATTEMPT_ID=$(psql_exec "
+  SELECT id FROM index_attempt
+  WHERE connector_credential_pair_id = $CC_PAIR_ID
+  ORDER BY id DESC LIMIT 1;
+")
+
+if [[ -z "$ATTEMPT_ID" ]]; then
+  echo "No index attempt found. Creating one..."
+  SEARCH_SETTINGS_ID=$(psql_exec "SELECT id FROM search_settings ORDER BY id DESC LIMIT 1;")
+  if [[ -z "$SEARCH_SETTINGS_ID" ]]; then
+    echo "No search_settings found in DB." >&2
+    exit 1
+  fi
+  ATTEMPT_ID=$(psql_exec "
+    INSERT INTO index_attempt (connector_credential_pair_id, search_settings_id, from_beginning, status, new_docs_indexed, total_docs_indexed, docs_removed_from_index, time_updated, completed_batches, total_chunks)
+    VALUES ($CC_PAIR_ID, $SEARCH_SETTINGS_ID, true, 'completed_with_errors', 5, 10, 0, now(), 0, 0)
+    RETURNING id;
+  ")
+  echo "Created index attempt ID: $ATTEMPT_ID"
+else
+  echo "Using existing index attempt ID: $ATTEMPT_ID"
+fi
+
+# --- Insert the curated test errors ---
+echo "Inserting test errors..."
+
+# Error 1: Document with link (hyperlinked doc ID)
+psql_exec "
+INSERT INTO index_attempt_errors (index_attempt_id, connector_credential_pair_id, document_id, document_link, entity_id, failure_message, is_resolved)
+VALUES ($ATTEMPT_ID, $CC_PAIR_ID, 'doc-001', 'https://example.com/doc-001', NULL, 'SEED: Timeout while fetching document content from remote server', false);
+"
+
+# Error 2: Document without link (plain text doc ID)
+psql_exec "
+INSERT INTO index_attempt_errors (index_attempt_id, connector_credential_pair_id, document_id, document_link, entity_id, failure_message, is_resolved)
+VALUES ($ATTEMPT_ID, $CC_PAIR_ID, 'doc-no-link', NULL, NULL, 'SEED: Permission denied accessing resource - authentication token expired', false);
+"
+
+# Error 3: Entity ID only (no document_id, no link)
+psql_exec "
+INSERT INTO index_attempt_errors (index_attempt_id, connector_credential_pair_id, document_id, document_link, entity_id, failure_message, is_resolved)
+VALUES ($ATTEMPT_ID, $CC_PAIR_ID, NULL, NULL, 'entity-abc', 'SEED: Entity sync failed due to upstream rate limiting', false);
+"
+
+# Error 4: Entity ID with link (hyperlinked entity)
+psql_exec "
+INSERT INTO index_attempt_errors (index_attempt_id, connector_credential_pair_id, document_id, document_link, entity_id, failure_message, is_resolved)
+VALUES ($ATTEMPT_ID, $CC_PAIR_ID, NULL, 'https://example.com/entity', 'entity-link-test', 'SEED: Connection reset by peer during entity fetch', false);
+"
+
+# Error 5: Neither document_id nor entity_id (renders "Unknown")
+psql_exec "
+INSERT INTO index_attempt_errors (index_attempt_id, connector_credential_pair_id, document_id, document_link, entity_id, failure_message, is_resolved)
+VALUES ($ATTEMPT_ID, $CC_PAIR_ID, NULL, NULL, NULL, 'SEED: Unknown document failed with a catastrophic internal error that produced a very long error message designed to test the scrollable cell behavior in the modal UI. This message continues for quite a while to ensure the 60px height overflow-y-auto container is properly exercised during manual testing.', false);
+"
+
+# Error 6: XSS test (special HTML characters)
+psql_exec "
+INSERT INTO index_attempt_errors (index_attempt_id, connector_credential_pair_id, document_id, document_link, entity_id, failure_message, is_resolved)
+VALUES ($ATTEMPT_ID, $CC_PAIR_ID, 'doc-xss', NULL, NULL, 'SEED: <script>alert(''xss'')</script>', false);
+"
+
+# Error 7: Single-character error message
+psql_exec "
+INSERT INTO index_attempt_errors (index_attempt_id, connector_credential_pair_id, document_id, document_link, entity_id, failure_message, is_resolved)
+VALUES ($ATTEMPT_ID, $CC_PAIR_ID, 'doc-short', NULL, NULL, 'SEED: X', false);
+"
+
+# Insert additional generic errors if --count > 7
+if (( ERROR_COUNT > 7 )); then
+  extra=$(( ERROR_COUNT - 7 ))
+  echo "Inserting $extra additional generic errors..."
+  for i in $(seq 1 "$extra"); do
+    psql_exec "
+    INSERT INTO index_attempt_errors (index_attempt_id, connector_credential_pair_id, document_id, document_link, entity_id, failure_message, is_resolved)
+    VALUES ($ATTEMPT_ID, $CC_PAIR_ID, 'doc-extra-$i', NULL, NULL, 'SEED: Generic error #$i for pagination testing', false);
+    "
+  done
+fi
+
+# Error: One resolved error (to test filtering)
+psql_exec "
+INSERT INTO index_attempt_errors (index_attempt_id, connector_credential_pair_id, document_id, document_link, entity_id, failure_message, is_resolved)
+VALUES ($ATTEMPT_ID, $CC_PAIR_ID, 'doc-resolved', NULL, NULL, 'SEED: This error was already resolved', true);
+"
+
+# --- Verify ---
+total=$(psql_exec "SELECT count(*) FROM index_attempt_errors WHERE connector_credential_pair_id = $CC_PAIR_ID AND failure_message LIKE 'SEED:%';")
+unresolved=$(psql_exec "SELECT count(*) FROM index_attempt_errors WHERE connector_credential_pair_id = $CC_PAIR_ID AND failure_message LIKE 'SEED:%' AND is_resolved = false;")
+
+echo ""
+echo "=== Done ==="
+echo "CC Pair ID:        $CC_PAIR_ID"
+echo "Index Attempt ID:  $ATTEMPT_ID"
+echo "Seeded errors:     $total ($unresolved unresolved, $(( total - unresolved )) resolved)"
+echo ""
+echo "View in browser:   $BASE_URL/admin/connector/$CC_PAIR_ID"
+echo "API check:         curl -b <cookies> '$BASE_URL/api/manage/admin/cc-pair/$CC_PAIR_ID/errors'"
+echo ""
+echo "To clean up:       $0 --clean"
+echo "To delete connector: curl -b <cookies> -X DELETE '$BASE_URL/api/manage/admin/cc-pair/$CC_PAIR_ID'"