Pull Downloaded LLMs, Fixed and added tests

cucumber.json

@@ -1,7 +1,16 @@
 {
   "default": {
     "require": ["features/step_definitions/**/*.js"],
-    "paths": ["features/**/*.feature"],
+    "paths": [
+      "features/chat.feature",
+      "features/history.feature",
+      "features/temporary_chat.feature",
+      "features/settings.feature",
+      "features/model_selection.feature",
+      "features/account_management.feature",
+      "features/clipboard_copy.feature",
+      "features/chat_streaming.feature"
+    ],
     "format": ["progress"]
   }
 }

features/account_management.feature

@@ -0,0 +1,9 @@
+Feature: Account management
+
+  Scenario: Access account settings, attempt password change, and attempt account deletion
+    Given I have an authenticated session
+    When I navigate to the account page
+    And I open account from the navigation
+    And I attempt to change my password
+    And I attempt to delete my account
+    Then the account management actions should be recorded

features/chat.feature

@@ -0,0 +1,17 @@
+Feature: Core chat functionality
+
+  Scenario: Navigate to chat, switch models, and receive math and weather answers
+    Given I have an authenticated session
+    And local models are available: qwen3:0.5b, gemma3:1b, phi4-mini
+    And cloud models are available: gpt-4o, gemini-1.5-flash, claude-3-haiku
+    When I navigate to the chat page
+    And I open chat from the navigation
+    And I choose the local model "qwen3:0.5b"
+    And I ask "What is 144 divided by 12?"
+    Then I should see a numeric chat answer
+    And the active model label should be "qwen3:0.5b"
+    When I choose the cloud model "gemini-1.5-flash"
+    And I store cloud credentials for "gemini-1.5-flash"
+    And I ask "What is the weather like in Seattle?"
+    Then I should see a descriptive weather reply
+    And the active model label should be "gemini-1.5-flash"

features/chat_streaming.feature

@@ -1,7 +1,8 @@
-Feature: Real-time Message Streaming
-  Scenario: User sees incremental response
-    Given I am logged in
-    And I have sent a prompt to the LLM
-    When the server begins streaming the response chunks
-    Then I should see the message text appearing character-by-character
-    And I should see a "typing" or "processing" indicator until the stream ends
+Feature: Streaming chat responses
+
+  Scenario: Show typing feedback and character-by-character streaming
+    Given I have an authenticated session
+    When I navigate to the chat page
+    And I stream a response for "What is 8 times 8?"
+    Then I should observe a typing indicator while the model is thinking
+    And the response should be streamed character by character

features/clipboard_copy.feature

@@ -0,0 +1,11 @@
+Feature: Clipboard copy actions
+
+  Scenario: Copy user and assistant text and code blocks
+    Given I have an authenticated session
+    And I have a conversation with text and code messages
+    When I navigate to the clipboard icon targets
+    And I copy the "user message text"
+    And I copy the "LLM message text"
+    And I copy the "user code block"
+    And I copy the "LLM code block"
+    Then the clipboard history should include all copied message variants

features/history.feature

@@ -0,0 +1,9 @@
+Feature: Chat history
+
+  Scenario: Navigate to history and delete a conversation thread
+    Given I have an authenticated session
+    And I have a saved conversation thread
+    When I navigate to the history page
+    And I open history from the navigation
+    And I delete the saved conversation thread
+    Then the conversation thread should be removed from history

features/model_selection.feature

@@ -0,0 +1,19 @@
+Feature: Model selection
+
+  Scenario: Retrieve model lists and validate local and cloud selection behavior
+    Given I have an authenticated session
+    And local models are available: qwen3:0.5b, gemma3:1b, phi4-mini
+    And cloud models are available: gpt-4o, gemini-1.5-flash, claude-3-haiku
+    When I retrieve the available model list
+    Then the API model list should include local models: qwen3:0.5b, gemma3:1b, phi4-mini
+    And the API model list should include cloud models: gpt-4o, gemini-1.5-flash, claude-3-haiku
+    When I choose the local model "phi4-mini"
+    And I ask "What is 21 divided by 3?"
+    Then the local model request should use "phi4-mini"
+    When I choose the cloud model "claude-3-haiku"
+    And I store cloud credentials for "claude-3-haiku"
+    And I ask "What is the weather like in Seattle?"
+    Then the cloud model request should use "claude-3-haiku"
+    When I clear cloud credentials for "claude-3-haiku"
+    And I ask "What is the weather like in Seattle?"
+    Then I should see a missing credentials error

features/settings.feature

@@ -0,0 +1,13 @@
+Feature: Settings page
+
+  Scenario: Save a default model preference across sessions
+    Given I have an authenticated session
+    And local models are available: qwen3:0.5b, gemma3:1b, phi4-mini
+    And cloud models are available: gpt-4o, gemini-1.5-flash, claude-3-haiku
+    When I navigate to the settings page
+    And I open settings from the navigation
+    And I switch the model category to "cloud"
+    And I choose the cloud model "gpt-4o"
+    And I save the selected model preference
+    And I start a new session
+    Then the saved model preference should be "gpt-4o"