Prerequisites

• LM Studio 0.3.x or higher.
• Embedding Model: Download nomic-embed-text-v1.5.Q4_K_M.gguf (or similar) in LM Studio.
• Local Whisper (Optional, for YouTube):
  • Windows: Download whisper.cpp release (e.g., whisper-bin-x64.zip) and extract it. Set whisperBinaryPath to main.exe.
  • Fedora/Linux: Install yt-dlp and whisper-cpp via your package manager:
    Or build whisper.cpp from source to get the main binary.
  • In LM Studio, go to the Plugins tab → Local OpenBook → Settings.
  • Set whisperBinaryPath to the path of your whisper-cpp or main binary (e.g., /usr/bin/whisper-cpp). If it's in your PATH, the plugin will now try to find it automatically.

How to Use

• Start a Chat.
• Drag & Drop your research papers or notes.
• Add Web Sources (Optional):
  • User: "Add source https://en.wikipedia.org/wiki/Artificial_intelligence"
  • The plugin will fetch the text and add it to the chat.
• Ask Questions: "What does the text say about X?"
  • The plugin will search your documents and cite sources.
• Create a Podcast:
  • User: "Generate a podcast script about these documents."
  • The model will produce a dialogue script.

Configuration

• Reranker Enabled: Toggles the advanced reranking step. Disable for speed, enable for accuracy.
• Retrieval Limit: How many text chunks to retrieve (default: 5).

Known Limitations

• Web/YouTube Source Indexing: Currently, content fetched from the web/YouTube is added directly to the chat context. It is not indexed for RAG retrieval unless you copy the text into a .txt file and drag it back into LM Studio.
• Audio Generation: This plugin generates the script. You need an external TTS tool (or a future update) to convert the script to audio.