supervertaler 1.9.194__tar.gz → 1.9.196__tar.gz

{supervertaler-1.9.194 → supervertaler-1.9.196}/CHANGELOG.md

@@ -2,10 +2,36 @@
 
 All notable changes to Supervertaler are documented in this file.
 
-**Current Version:** v1.9.194 (February 1, 2026)
+**Current Version:** v1.9.196 (February 2, 2026)
 
 
 
+## v1.9.196 - February 2, 2026
+
+### 🐛 Bug Fixes
+
+- **AI Assistant "Analyze Project" Fixed**: Fixed a bug where clicking "Analyze Project & Generate Prompts" in the AI Assistant would fail with "AI responded but no actions were found". The issue was that the LLM wasn't receiving a system prompt explaining the ACTION format. Now all providers (OpenAI, Claude, Gemini, Ollama) receive proper instructions on how to format their responses.
+
+### 🎨 UI Improvements
+
+- **Supervoice Settings Two-Column Layout**: Reorganized the Supervoice settings tab into a two-column layout. Voice commands table now displays on the right side, allowing more commands to be visible at once. Settings (Always-On Listening, Speech Recognition Model, AutoHotkey) remain on the left.
+
+- **Green Checkmark Checkbox**: The "Enable voice commands" checkbox in Supervoice settings now uses the standard green checkmark style (CheckmarkCheckBox) for visual consistency with the rest of the application.
+
+### 🏗️ Code Architecture
+
+- **LLM Client System Prompt Support**: Added `system_prompt` parameter to the LLM client's `translate()` method and all provider-specific methods. This enables passing behavioral instructions to AI models for features like the AI Assistant.
+
+---
+
+## v1.9.195 - February 2, 2026
+
+### 🏗️ Code Architecture
+
+- **Module Rename**: Renamed `mt_quick_popup.py` → `quicktrans.py` to align with the QuickTrans branding. This prepares the module for future standalone distribution as a separate Windows executable.
+
+---
+
 ## v1.9.194 - February 1, 2026
 
 ### 🎨 Branding & Naming

{supervertaler-1.9.194 → supervertaler-1.9.196}/FAQ.md

@@ -672,6 +672,34 @@ ollama pull qwen2.5:3b
 - Multiple API keys (GPT-4o, Claude, Gemini)
 - Sufficient API credits for testing runs
 
+### Why is my antivirus flagging Supervertaler as malware?
+
+This is a **known false positive** that affects applications built with PyInstaller (the tool used to create the Windows EXE).
+
+**Why it happens:**
+- PyInstaller bundles Python + all libraries into a single package
+- This binary pattern triggers heuristic detection in some AV software (especially Norton 360)
+- Generic detections like "Win64:Malware-gen" are pattern matches, not actual malware signatures
+- The executable is not code-signed (certificates cost ~$300/year)
+
+**Solutions:**
+
+1. **Use pip install (Recommended)**
+   - Install: `pip install supervertaler`
+   - Run: `supervertaler`
+   - No AV warnings because it runs from Python source
+
+2. **Whitelist the application**
+   - Add an exception in your antivirus settings
+   - The application is safe - you can verify by checking the [source code on GitHub](https://github.com/michaelbeijer/Supervertaler)
+
+3. **Report as false positive**
+   - Norton: https://submit.norton.com/
+   - Windows Defender: Click "More info" → "Run anyway"
+   - Other AV: Check vendor's false positive submission page
+
+**This is common for PyInstaller apps** - it affects tools like Calibre, many game mods, and scientific software. See the [PyInstaller FAQ](https://pyinstaller.org/en/stable/faq.html) for more details.
+
 ---
 
 ## Workflow & Integration

{supervertaler-1.9.194 → supervertaler-1.9.196}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: supervertaler
-Version: 1.9.194
+Version: 1.9.196
 Summary: Professional AI-enhanced translation workbench with multi-LLM support, glossary system, TM, spellcheck, voice commands, and PyQt6 interface. Batteries included (core).
 Home-page: https://supervertaler.com
 Author: Michael Beijer

{supervertaler-1.9.194 → supervertaler-1.9.196}/Supervertaler.py

@@ -32,9 +32,9 @@ License: MIT
 """
 
 # Version Information.
-__version__ = "1.9.194"
+__version__ = "1.9.196"
 __phase__ = "0.9"
-__release_date__ = "2026-02-01"
+__release_date__ = "2026-02-02"
 __edition__ = "Qt"
 
 import sys
@@ -7447,7 +7447,7 @@ class SupervertalerQt(QMainWindow):
                 return
 
             # Import and create the popup
-            from modules.mt_quick_popup import MTQuickPopup
+            from modules.quicktrans import MTQuickPopup
 
             # Create popup
             popup = MTQuickPopup(
@@ -18307,14 +18307,15 @@ class SupervertalerQt(QMainWindow):
 
     def _create_voice_dictation_settings_tab(self):
         """Create Supervoice Settings tab content with Voice Commands"""
-        from PyQt6.QtWidgets import (QGroupBox, QPushButton, QComboBox, QSpinBox, 
+        from PyQt6.QtWidgets import (QGroupBox, QPushButton, QComboBox, QSpinBox,
                                      QTableWidget, QTableWidgetItem, QHeaderView,
-                                     QAbstractItemView, QCheckBox)
+                                     QAbstractItemView, QCheckBox, QSplitter)
+        from modules.keyboard_shortcuts_widget import CheckmarkCheckBox
 
         tab = QWidget()
-        layout = QVBoxLayout(tab)
-        layout.setContentsMargins(20, 20, 20, 20)
-        layout.setSpacing(15)
+        main_layout = QVBoxLayout(tab)
+        main_layout.setContentsMargins(20, 20, 20, 20)
+        main_layout.setSpacing(15)
 
         # Load current dictation settings
         dictation_settings = self.load_dictation_settings()
@@ -18327,19 +18328,33 @@ class SupervertalerQt(QMainWindow):
         header_info.setTextFormat(Qt.TextFormat.RichText)
         header_info.setStyleSheet("font-size: 9pt; color: #444; padding: 10px; background-color: #E3F2FD; border-radius: 4px;")
         header_info.setWordWrap(True)
-        layout.addWidget(header_info)
+        main_layout.addWidget(header_info)
 
-        # ===== Voice Commands Section =====
-        commands_group = QGroupBox("🗣️ Voice Commands (Talon-style)")
-        commands_layout = QVBoxLayout()
+        # ===== Two-column layout: Left = Settings, Right = Voice Commands Table =====
+        columns_layout = QHBoxLayout()
+        columns_layout.setSpacing(15)
 
-        # Enable voice commands checkbox
-        voice_cmd_enabled = QCheckBox("Enable voice commands (spoken phrases trigger actions)")
+        # --- LEFT COLUMN: Settings ---
+        left_column = QVBoxLayout()
+        left_column.setSpacing(15)
+
+        # Enable voice commands checkbox (green checkmark style)
+        voice_cmd_enabled = CheckmarkCheckBox("Enable voice commands (spoken phrases trigger actions)")
         voice_cmd_enabled.setChecked(dictation_settings.get('voice_commands_enabled', True))
         voice_cmd_enabled.setToolTip("When enabled, spoken phrases like 'confirm' or 'next segment' will execute commands instead of being inserted as text")
-        commands_layout.addWidget(voice_cmd_enabled)
+        left_column.addWidget(voice_cmd_enabled)
         self.voice_commands_enabled_checkbox = voice_cmd_enabled
 
+        # Create a layout variable for settings sections to be added below
+        layout = left_column
+
+        # --- RIGHT COLUMN: Voice Commands Table ---
+        right_column = QVBoxLayout()
+        right_column.setSpacing(10)
+
+        commands_group = QGroupBox("🗣️ Voice Commands (Talon-style)")
+        commands_layout = QVBoxLayout()
+
         commands_info = QLabel(
             "Voice commands let you control Supervertaler by voice. Say a phrase to execute an action.\n"
             "If no command matches, the spoken text is inserted as dictation."
@@ -18358,38 +18373,47 @@ class SupervertalerQt(QMainWindow):
         self.voice_commands_table.horizontalHeader().setSectionResizeMode(3, QHeaderView.ResizeMode.ResizeToContents)
         self.voice_commands_table.setSelectionBehavior(QAbstractItemView.SelectionBehavior.SelectRows)
         self.voice_commands_table.setEditTriggers(QAbstractItemView.EditTrigger.NoEditTriggers)
-        self.voice_commands_table.setMinimumHeight(200)
-        self.voice_commands_table.setMaximumHeight(300)
-        
+        self.voice_commands_table.setMinimumHeight(350)
+
         # Populate table with current commands
         self._populate_voice_commands_table()
         commands_layout.addWidget(self.voice_commands_table)
 
         # Command buttons
         cmd_btn_layout = QHBoxLayout()
-        
-        add_cmd_btn = QPushButton("➕ Add Command")
+
+        add_cmd_btn = QPushButton("➕ Add")
         add_cmd_btn.clicked.connect(self._add_voice_command)
         cmd_btn_layout.addWidget(add_cmd_btn)
-        
-        edit_cmd_btn = QPushButton("✏️ Edit Command")
+
+        edit_cmd_btn = QPushButton("✏️ Edit")
         edit_cmd_btn.clicked.connect(self._edit_voice_command)
         cmd_btn_layout.addWidget(edit_cmd_btn)
-        
-        remove_cmd_btn = QPushButton("🗑️ Remove Command")
+
+        remove_cmd_btn = QPushButton("🗑️ Remove")
         remove_cmd_btn.clicked.connect(self._remove_voice_command)
         cmd_btn_layout.addWidget(remove_cmd_btn)
-        
+
         cmd_btn_layout.addStretch()
-        
-        reset_cmd_btn = QPushButton("🔄 Reset to Defaults")
+
+        reset_cmd_btn = QPushButton("🔄 Reset")
         reset_cmd_btn.clicked.connect(self._reset_voice_commands)
         cmd_btn_layout.addWidget(reset_cmd_btn)
-        
+
         commands_layout.addLayout(cmd_btn_layout)
 
         commands_group.setLayout(commands_layout)
-        layout.addWidget(commands_group)
+        right_column.addWidget(commands_group)
+        right_column.addStretch()
+
+        # Add columns to the two-column layout
+        left_widget = QWidget()
+        left_widget.setLayout(left_column)
+        right_widget = QWidget()
+        right_widget.setLayout(right_column)
+
+        columns_layout.addWidget(left_widget, stretch=1)
+        columns_layout.addWidget(right_widget, stretch=1)
 
         # ===== Always-On Mode Section =====
         alwayson_group = QGroupBox("🎧 Always-On Listening Mode")
@@ -18552,7 +18576,13 @@ class SupervertalerQt(QMainWindow):
         ahk_group.setLayout(ahk_layout)
         layout.addWidget(ahk_group)
 
-        # Save button
+        # Add stretch to left column to push content up
+        layout.addStretch()
+
+        # Add two-column layout to main layout
+        main_layout.addLayout(columns_layout, stretch=1)
+
+        # Save button (full width, below the two columns)
         save_btn = QPushButton("💾 Save Supervoice Settings")
         save_btn.setStyleSheet("background-color: #4CAF50; color: white; font-weight: bold; padding: 10px; border: none; outline: none;")
         save_btn.clicked.connect(lambda: self._save_voice_settings(
@@ -18561,15 +18591,13 @@ class SupervertalerQt(QMainWindow):
             lang_combo.currentText(),
             voice_cmd_enabled.isChecked()
         ))
-        layout.addWidget(save_btn)
+        main_layout.addWidget(save_btn)
 
         # Store references
         self.dictation_model_combo = model_combo
         self.dictation_duration_spin = duration_spin
         self.dictation_lang_combo = lang_combo
 
-        layout.addStretch()
-
         return tab
 
     def _populate_voice_commands_table(self):
@@ -49563,7 +49591,7 @@ class SuperlookupTab(QWidget):
             target_lang = getattr(main_window, 'target_language', 'Dutch')
 
             # Import and show MT Quick Lookup popup
-            from modules.mt_quick_popup import MTQuickPopup
+            from modules.quicktrans import MTQuickPopup
 
             # Create popup without Qt parent so it can appear independently over any app
             # Pass main_window as parent_app for API access

{supervertaler-1.9.194 → supervertaler-1.9.196}/docs/index.html

@@ -17,7 +17,7 @@
     <link rel="apple-touch-icon" href="apple-touch-icon.png">
     <link rel="manifest" href="site.webmanifest">
 
-    <link rel="stylesheet" href="styles.css?v=1.9.194">
+    <link rel="stylesheet" href="styles.css?v=1.9.196">
     <link rel="preconnect" href="https://fonts.googleapis.com">
     <link rel="preconnect" href="https://fonts.gstatic.com" crossorigin>
     <link href="https://fonts.googleapis.com/css2?family=Inter:wght@300;400;500;600;700;800&display=swap"
@@ -50,7 +50,7 @@
                         class="nav-discussions">💬 Discussions</a></li>
                 <li class="nav-action">
                     <a href="https://github.com/michaelbeijer/Supervertaler/blob/main/CHANGELOG.md" target="_blank"
-                        class="nav-version" title="Current version">v1.9.194</a>
+                        class="nav-version" title="Current version">v1.9.196</a>
                 </li>
                 <li class="nav-action">
                     <a href="https://github.com/michaelbeijer/Supervertaler" target="_blank" class="nav-icon-link"
@@ -74,7 +74,7 @@
                     Supervertaler
                     <span class="gradient-text">The Ultimate Translation Workbench</span>
                 </h1>
-                <p class="version-badge">v1.9.194</p>
+                <p class="version-badge">v1.9.196</p>
                 <p class="hero-subtitle">
                     AI-enhanced translation workbench. Work with ChatGPT, Claude, and Gemini simultaneously.
                     Built by translators, for translators and writers.
@@ -711,7 +711,7 @@
             <div style="text-align: center; margin: 40px 0;">
                 <div class="download-card featured"
                     style="border: 3px solid #6366F1; max-width: 700px; margin: 0 auto;">
-                    <div class="download-badge" style="background: #6366F1;">v1.9.194</div>
+                    <div class="download-badge" style="background: #6366F1;">v1.9.196</div>
                     <h3>🎨 Supervertaler</h3>
                     <p class="download-description" style="font-weight: 600; color: #6366F1; margin-bottom: 10px;">
                         Modern Professional Interface
@@ -823,6 +823,22 @@
             <p class="section-subtitle">See Supervertaler in action</p>
 
             <div class="screenshots-grid">
+                <div class="screenshot-card" style="border: 3px solid #6366F1;">
+                    <a href="screenshots/Supervertalerv1.9.196_2026-02-02.png" class="screenshot-link" target="_blank">
+                        <img src="screenshots/Supervertalerv1.9.196_2026-02-02.png"
+                            alt="Supervertaler v1.9.196 with QuickTrans" loading="lazy">
+                        <div class="screenshot-overlay">
+                            <span class="zoom-icon">🔍 Click to enlarge</span>
+                        </div>
+                    </a>
+                    <div class="screenshot-caption">
+                        <h3>⚡ QuickTrans & Modern UI <span class="badge"
+                                style="background: #6366F1; color: white; padding: 3px 8px; border-radius: 12px; font-size: 11px; margin-left: 5px;">v1.9.196</span>
+                        </h3>
+                        <p>QuickTrans instant translation popup with MT engines and LLMs. Global hotkey (Ctrl+Alt+M) works from any application.</p>
+                    </div>
+                </div>
+
                 <div class="screenshot-card" style="border: 3px solid #2196F3;">
                     <a href="screenshots/Supervertaler_v1.9.102_grid_2026-01-14.png" class="screenshot-link" target="_blank">
                         <img src="screenshots/Supervertaler_v1.9.102_grid_2026-01-14.png"
@@ -1207,7 +1223,7 @@
         </svg>
     </button>
 
-    <script src="script.js?v=1.9.191"></script>
+    <script src="script.js?v=1.9.196"></script>
 </body>
 
 </html>

{supervertaler-1.9.194 → supervertaler-1.9.196}/docs/superdocs/reference/faq.md

@@ -722,6 +722,34 @@ ollama pull qwen2.5:3b
 - Multiple API keys (GPT-4o, Claude, Gemini)
 - Sufficient API credits for testing runs
 
+### Why is my antivirus flagging Supervertaler as malware?
+
+This is a **known false positive** that affects applications built with PyInstaller (the tool used to create the Windows EXE).
+
+**Why it happens:**
+- PyInstaller bundles Python + all libraries into a single package
+- This binary pattern triggers heuristic detection in some AV software (especially Norton 360)
+- Generic detections like "Win64:Malware-gen" are pattern matches, not actual malware signatures
+- The executable is not code-signed (certificates cost ~$300/year)
+
+**Solutions:**
+
+1. **Use pip install (Recommended)**
+   - Install: `pip install supervertaler`
+   - Run: `supervertaler`
+   - No AV warnings because it runs from Python source
+
+2. **Whitelist the application**
+   - Add an exception in your antivirus settings
+   - The application is safe - you can verify by checking the [source code on GitHub](https://github.com/michaelbeijer/Supervertaler)
+
+3. **Report as false positive**
+   - Norton: https://submit.norton.com/
+   - Windows Defender: Click "More info" → "Run anyway"
+   - Other AV: Check vendor's false positive submission page
+
+**This is common for PyInstaller apps** - it affects tools like Calibre, many game mods, and scientific software. See the [PyInstaller FAQ](https://pyinstaller.org/en/stable/faq.html) for more details.
+
 ---
 
 ## Workflow & Integration

{supervertaler-1.9.194 → supervertaler-1.9.196}/modules/llm_clients.py

@@ -594,18 +594,20 @@ class LLMClient:
         context: Optional[str] = None,
         custom_prompt: Optional[str] = None,
         max_tokens: Optional[int] = None,
-        images: Optional[List] = None
+        images: Optional[List] = None,
+        system_prompt: Optional[str] = None
     ) -> str:
         """
         Translate text using configured LLM
-        
+
         Args:
             text: Text to translate
             source_lang: Source language code
             target_lang: Target language code
             context: Optional context for translation
             custom_prompt: Optional custom prompt (overrides default simple prompt)
-        
+            system_prompt: Optional system prompt for AI behavior context
+
         Returns:
             Translated text
         """
@@ -615,30 +617,30 @@ class LLMClient:
         else:
             # Build prompt
             prompt = f"Translate the following text from {source_lang} to {target_lang}:\n\n{text}"
-            
+
             if context:
                 prompt = f"Context: {context}\n\n{prompt}"
-        
+
         # Log warning if images provided but model doesn't support vision
         if images and not self.model_supports_vision(self.provider, self.model):
             print(f"⚠️ Warning: Model {self.model} doesn't support vision. Images will be ignored.")
             images = None  # Don't pass to API
-        
+
         # Call appropriate provider
         if self.provider == "openai":
-            return self._call_openai(prompt, max_tokens=max_tokens, images=images)
+            return self._call_openai(prompt, max_tokens=max_tokens, images=images, system_prompt=system_prompt)
         elif self.provider == "claude":
-            return self._call_claude(prompt, max_tokens=max_tokens, images=images)
+            return self._call_claude(prompt, max_tokens=max_tokens, images=images, system_prompt=system_prompt)
         elif self.provider == "gemini":
-            return self._call_gemini(prompt, max_tokens=max_tokens, images=images)
+            return self._call_gemini(prompt, max_tokens=max_tokens, images=images, system_prompt=system_prompt)
         elif self.provider == "ollama":
-            return self._call_ollama(prompt, max_tokens=max_tokens)
+            return self._call_ollama(prompt, max_tokens=max_tokens, system_prompt=system_prompt)
         else:
             raise ValueError(f"Unsupported provider: {self.provider}")
     
-    def _call_openai(self, prompt: str, max_tokens: Optional[int] = None, images: Optional[List] = None) -> str:
+    def _call_openai(self, prompt: str, max_tokens: Optional[int] = None, images: Optional[List] = None, system_prompt: Optional[str] = None) -> str:
         """Call OpenAI API with GPT-5/o1/o3 reasoning model support and vision capability"""
-        print(f"🔵 _call_openai START: model={self.model}, prompt_len={len(prompt)}, max_tokens={max_tokens}, images={len(images) if images else 0}")
+        print(f"🔵 _call_openai START: model={self.model}, prompt_len={len(prompt)}, max_tokens={max_tokens}, images={len(images) if images else 0}, has_system={bool(system_prompt)}")
 
         try:
             from openai import OpenAI
@@ -686,10 +688,16 @@ class LLMClient:
             # Standard text-only format
             content = prompt
 
+        # Build messages list
+        messages = []
+        if system_prompt:
+            messages.append({"role": "system", "content": system_prompt})
+        messages.append({"role": "user", "content": content})
+
         # Build API call parameters
         api_params = {
             "model": self.model,
-            "messages": [{"role": "user", "content": content}],
+            "messages": messages,
             "timeout": timeout_seconds
         }
 
@@ -742,7 +750,7 @@ class LLMClient:
                 print(f"   Response: {e.response}")
             raise  # Re-raise to be caught by calling code
     
-    def _call_claude(self, prompt: str, max_tokens: Optional[int] = None, images: Optional[List] = None) -> str:
+    def _call_claude(self, prompt: str, max_tokens: Optional[int] = None, images: Optional[List] = None, system_prompt: Optional[str] = None) -> str:
         """Call Anthropic Claude API with vision support"""
         try:
             import anthropic
@@ -786,12 +794,19 @@ class LLMClient:
             # Standard text-only format
             content = prompt
         
-        response = client.messages.create(
-            model=self.model,
-            max_tokens=tokens_to_use,
-            messages=[{"role": "user", "content": content}],
-            timeout=timeout_seconds  # Explicit timeout
-        )
+        # Build API call parameters
+        api_params = {
+            "model": self.model,
+            "max_tokens": tokens_to_use,
+            "messages": [{"role": "user", "content": content}],
+            "timeout": timeout_seconds  # Explicit timeout
+        }
+
+        # Add system prompt if provided (Claude uses 'system' parameter, not a message)
+        if system_prompt:
+            api_params["system"] = system_prompt
+
+        response = client.messages.create(**api_params)
         
         translation = response.content[0].text.strip()
         
@@ -800,7 +815,7 @@ class LLMClient:
         
         return translation
     
-    def _call_gemini(self, prompt: str, max_tokens: Optional[int] = None, images: Optional[List] = None) -> str:
+    def _call_gemini(self, prompt: str, max_tokens: Optional[int] = None, images: Optional[List] = None, system_prompt: Optional[str] = None) -> str:
         """Call Google Gemini API with vision support"""
         try:
             import google.generativeai as genai
@@ -809,10 +824,15 @@ class LLMClient:
             raise ImportError(
                 "Google AI library not installed. Install with: pip install google-generativeai pillow"
             )
-        
+
         genai.configure(api_key=self.api_key)
-        model = genai.GenerativeModel(self.model)
-        
+
+        # Gemini supports system instructions via GenerativeModel parameter
+        if system_prompt:
+            model = genai.GenerativeModel(self.model, system_instruction=system_prompt)
+        else:
+            model = genai.GenerativeModel(self.model)
+
         # Build content (text + optional images)
         if images:
             # Gemini format: list with prompt text followed by PIL Image objects
@@ -823,7 +843,7 @@ class LLMClient:
         else:
             # Standard text-only
             content = prompt
-        
+
         response = model.generate_content(content)
         translation = response.text.strip()
         
@@ -832,20 +852,21 @@ class LLMClient:
         
         return translation
     
-    def _call_ollama(self, prompt: str, max_tokens: Optional[int] = None) -> str:
+    def _call_ollama(self, prompt: str, max_tokens: Optional[int] = None, system_prompt: Optional[str] = None) -> str:
         """
         Call local Ollama server for translation.
-        
+
         Ollama provides a simple REST API compatible with local LLM inference.
         Models run entirely on the user's computer - no API keys, no internet required.
-        
+
         Args:
             prompt: The full prompt to send
             max_tokens: Maximum tokens to generate (default: 4096)
-            
+            system_prompt: Optional system prompt for AI behavior context
+
         Returns:
             Translated text
-            
+
         Raises:
             ConnectionError: If Ollama is not running
             ValueError: If model is not available
@@ -866,13 +887,17 @@ class LLMClient:
         print(f"🟠 _call_ollama START: model={self.model}, prompt_len={len(prompt)}, max_tokens={tokens_to_use}")
         print(f"🟠 Ollama endpoint: {endpoint}")
         
+        # Build messages list
+        messages = []
+        if system_prompt:
+            messages.append({"role": "system", "content": system_prompt})
+        messages.append({"role": "user", "content": prompt})
+
         # Build request payload
         # Using /api/chat for chat-style interaction (better for translation prompts)
         payload = {
             "model": self.model,
-            "messages": [
-                {"role": "user", "content": prompt}
-            ],
+            "messages": messages,
             "stream": False,  # Get complete response at once
             "options": {
                 "temperature": 0.3,  # Low temperature for consistent translations

{supervertaler-1.9.194 → supervertaler-1.9.196}/modules/unified_prompt_manager_qt.py

@@ -3795,6 +3795,26 @@ Output complete ACTION."""
                 from PyQt6.QtWidgets import QApplication
                 QApplication.processEvents()
 
+            # System prompt that explains the ACTION format to the AI
+            ai_system_prompt = """You are an AI assistant for Supervertaler, a professional translation workbench.
+
+You can execute actions using a special format. When you need to create, modify, or manage prompts, output ACTION blocks in this EXACT format:
+
+ACTION:function_name PARAMS:{"param1": "value1", "param2": "value2"}
+
+Available actions:
+- create_prompt: Create a new prompt. Required params: name, content. Optional: folder, description, activate
+- update_prompt: Update an existing prompt. Required params: name. Optional: content, folder, description
+- delete_prompt: Delete a prompt. Required params: name
+- list_prompts: List all prompts. Optional params: folder
+- activate_prompt: Set a prompt as active. Required params: name
+
+IMPORTANT:
+1. Output ONLY the ACTION block when asked to create/modify prompts - no explanatory text
+2. The ACTION must be on a single line (PARAMS JSON can be multiline if needed)
+3. Use valid JSON for PARAMS (double quotes for strings, escape special characters)
+4. Do not wrap in code fences or add any markdown formatting"""
+
             # Call LLM using translate method with custom prompt
             # The translate method accepts a custom_prompt parameter that we can use for any text generation
             self.log_message("[AI Assistant] Calling LLM translate method...")
@@ -3802,7 +3822,8 @@ Output complete ACTION."""
                 text="",  # Empty text since we're using custom_prompt
                 source_lang="en",
                 target_lang="en",
-                custom_prompt=prompt
+                custom_prompt=prompt,
+                system_prompt=ai_system_prompt
             )
 
             # Log the response

{supervertaler-1.9.194 → supervertaler-1.9.196}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "supervertaler"
-version = "1.9.194"
+version = "1.9.196"
 description = "Professional AI-enhanced translation workbench with multi-LLM support, glossary system, TM, spellcheck, voice commands, and PyQt6 interface. Batteries included (core)."
 readme = "README.md"
 requires-python = ">=3.10"

{supervertaler-1.9.194 → supervertaler-1.9.196}/supervertaler.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: supervertaler
-Version: 1.9.194
+Version: 1.9.196
 Summary: Professional AI-enhanced translation workbench with multi-LLM support, glossary system, TM, spellcheck, voice commands, and PyQt6 interface. Batteries included (core).
 Home-page: https://supervertaler.com
 Author: Michael Beijer