Fix: Repair setup script, dependencies, and test suite

backend/requirements.txt

@@ -11,4 +11,5 @@ requests==2.31.0
 pyyaml==6.0.1
 websockets==12.0
 sse-starlette==1.8.2
-aiohttp==3.9.1
+aiohttp==3.9.1
+psutil

expected_user_behavior.md

@@ -0,0 +1,41 @@
+# Expected User Behavior: Core Research Workflow
+
+This document outlines the expected behavior for a primary user workflow in the DataKiln application: performing a basic research query. This can be used as a manual test plan to identify non-functional parts of the system.
+
+---
+
+### Prerequisites
+
+1.  **Action:** The user opens a new browser tab and manually logs into `gemini.google.com`.
+    *   **Expected Result:** The user is successfully logged into their Google account on the Gemini website. This ensures the browser automation has an authenticated session to work with.
+
+### Workflow Steps
+
+1.  **Action:** The user navigates to the DataKiln frontend URL (e.g., `http://localhost:3000`).
+    *   **Expected Result:** The application loads, and the main Dashboard page is displayed without errors.
+
+2.  **Action:** On the Dashboard, the user clicks the "New Research" button.
+    *   **Expected Result:** The UI transitions to a view where the user can configure a new research query.
+
+3.  **Action:** The user selects a research mode. For this test, they click on "Balanced".
+    *   **Expected Result:** The "Balanced" mode is visually selected or highlighted, indicating it's the active choice.
+
+4.  **Action:** The user types a simple query, such as `"What is the Playwright framework?"`, into the main input field.
+    *   **Expected Result:** The text is correctly entered and visible in the input field.
+
+5.  **Action:** The user clicks the "Start Research" button.
+    *   **Expected Result:**
+        *   The UI provides immediate feedback that the task has started (e.g., a loading indicator appears, the "Start" button is disabled).
+        *   A new entry appears in a "Jobs" or "Runs" panel, showing the status as "In Progress" or "Running".
+
+6.  **Action:** The user waits for the research task to complete.
+    *   **Expected Result:**
+        *   The backend should trigger a Playwright instance to perform the research on Gemini's website.
+        *   The UI should display real-time progress updates (e.g., in a log panel or through a progress bar).
+        *   After a few minutes, the status of the job in the UI should change to "Complete" or "Success".
+
+7.  **Action:** The user clicks on the completed job to view the results.
+    *   **Expected Result:** The UI displays the formatted research report, including a summary, sources, and the main body of text.
+
+8.  **Action:** The user checks the designated output directory.
+    *   **Expected Result:** A new Markdown file (`.md`) containing the research report has been saved to the local filesystem (e.g., in a default `output` folder or a configured Obsidian vault path). The file's content should match the results shown in the UI.

frontend/e2e/comprehensive-e2e-test.js

@@ -102,8 +102,8 @@ class E2ETestSuite {
 
     for (const item of navItems) {
       try {
-        // Look for sidebar navigation links
-        const selector = `a[href="${item.path}"]`;
+        // Look for sidebar navigation links, accounting for HashRouter
+        const selector = `a[href="#${item.path}"]`;
         await this.page.waitForSelector(selector, { timeout: 5000 });
         console.log(`Navigation item '${item.label}' found`);
       } catch (error) {

frontend/e2e/e2e-test-report.json

@@ -1,29 +1,29 @@
 {
-  "timestamp": "2025-09-23T04:45:05.317Z",
+  "timestamp": "2025-09-23T06:21:54.364Z",
   "results": {
     "passed": 9,
     "failed": 0,
     "errors": [],
     "screenshots": [
-      "screenshot-application-launch-2025-09-23T04-44-27-703Z.png",
-      "screenshot-navigation-2025-09-23T04-44-57-786Z.png",
-      "screenshot-dashboard-2025-09-23T04-45-02-852Z.png",
-      "screenshot-deep-research-dialog-2025-09-23T04-45-02-922Z.png",
-      "screenshot-youtube-transcript-2025-09-23T04-45-04-991Z.png",
-      "screenshot-workflow-management-2025-09-23T04-45-05-058Z.png",
-      "screenshot-results-browser-2025-09-23T04-45-05-133Z.png",
-      "screenshot-selector-lab-2025-09-23T04-45-05-201Z.png"
+      "screenshot-application-launch-2025-09-23T06-21-45-566Z.png",
+      "screenshot-navigation-2025-09-23T06-21-45-773Z.png",
+      "screenshot-dashboard-2025-09-23T06-21-50-947Z.png",
+      "screenshot-deep-research-dialog-2025-09-23T06-21-51-148Z.png",
+      "screenshot-youtube-transcript-2025-09-23T06-21-53-409Z.png",
+      "screenshot-workflow-management-2025-09-23T06-21-53-595Z.png",
+      "screenshot-results-browser-2025-09-23T06-21-53-878Z.png",
+      "screenshot-selector-lab-2025-09-23T06-21-54-085Z.png"
     ],
     "timings": {
-      "Application Launch": 1079,
-      "Navigation": 30059,
-      "Dashboard Functionality": 5077,
-      "Deep Research Feature": 58,
-      "YouTube Transcript Analysis": 2071,
-      "Workflow Management": 63,
-      "Results Browser": 83,
-      "Selector Lab": 54,
-      "API Integration": 16
+      "Application Launch": 2654,
+      "Navigation": 182,
+      "Dashboard Functionality": 5202,
+      "Deep Research Feature": 232,
+      "YouTube Transcript Analysis": 2193,
+      "Workflow Management": 229,
+      "Results Browser": 252,
+      "Selector Lab": 209,
+      "API Integration": 46
     }
   },
   "environment": {

setup.sh

@@ -76,7 +76,7 @@ setup_backend() {
 
     # Install Playwright browsers
     echo "Installing Playwright browsers..."
-    playwright install --with-deps chromium
+    playwright install chromium
 
     # Go back to root directory
     cd ..