Bug Report

[trofimseldenis-ux]

Bug Report

Version: ('2.2.2', 'stable')
Platform: win32

Description

(Please describe what went wrong or what you expected to happen)

Steps to Reproduce

📋 Logs (click to expand)

🚀 Jarvis Log Viewer Ready
──────────────────────────────────────────────────

Starting event loop...
🚀 Jarvis daemon started
📋 Initializing daemon components...
✓ Daemon started
🧠 Using chat model: qwen3.5:0.8b
🎤 Using whisper model: tiny.en
📡 Discovering MCP tools from 2 server(s)...
  ✅ chrome-devtools: 29 tools available
  ✅ youtube-transcript: 1 tools available
💾 Initializing dialogue memory...
✓ Dialogue memory initialized
  📍 Location features are not available
     To enable location-based features:
     1. 🌐 Register for a free MaxMind account:
        https://www.maxmind.com/en/geolite2/signup
     2. 📥 Download the GeoLite2 City database (MMDB format)
     3. 📂 Save the database file as:
        C:\Users\trofi\.local\share\jarvis\geoip\GeoLite2-City.mmdb
📍 Location detection not available
     GeoLite2 database not found. Download from:
     https://www.maxmind.com/en/geolite2/signup
🔊 Initializing TTS engine (piper)...
✓ TTS engine started
🎤 Initializing voice listener (this may take a moment to load Whisper model)...
✓ Voice listener thread started (loading Whisper model in background)
  🔐 Checking microphone permission...
🎙️ Dictation enabled (hold Ctrl + Win to dictate)
  ✅ Microphone permission OK
  🔥 Warming up models...
  ℹ️  CUDA not available, using CPU mode
     Missing: cuBLAS, cuDNN
  💡 For GPU acceleration, click 'Reinstall GPU libraries' in the Jarvis tray menu
     🎤 Loading Whisper 'tiny.en' (device=cpu, compute=int8)...
     ⚡ CPU mode: using 20 threads with optimised decoding
     🎤 Whisper 'tiny.en' loaded on cpu
     💬 Chat model 'qwen3.5:0.8b' ready
     🧠 Intent judge 'qwen3.5:0.8b' ready
     🔧 Tool router 'qwen3.5:0.8b' ready
     📐 Embed model 'nomic-embed-text' ready
  🎤 Using default device: Микрофон (HyperX Cloud II Wirel
──────────────────────────────────────────────────
🎙️  Listening! Try:
      "How's the weather in [your city], Jarvis?"
      "I just ate a Big Mac, Jarvis."
      "What are you thinking, Jarvis?"
      "What do you know about me, Jarvis?"
  ⚠️  Small model in use (qwen3.5:0.8b). Assume it can't infer — spell out the steps for anything more involved:
      "Tell me tomorrow's weather, then find local events for tomorrow, then recommend ones that suit the weather, Jarvis."
  🌐 Chrome MCP detected. Name the destination URL so the browser tool can act directly:
      "Navigate to youtube.com, Jarvis."
📝 Heard: "Navigate to YouTube"

Additional Context

(Any other relevant information)