Local OpenBook Plugin - User Guide

This guide explains every feature of the Local OpenBook plugin for LM Studio and provides concrete examples of how to use them.

1. Add Web Source

Function: Extracts the main text content from a URL (article, blog post, documentation), strips out ads/menus, and adds it to your current chat context.

• Command: /add_source
• Parameters: url (The link), type: "web"

Example 1: Studying a Wikipedia Article

User Input:

Result: The plugin fetches the text of the article. Follow-up: "Summarize the history section of the article you just read."

Example 2: Analyzing a News Article

User Input:

Result: The plugin extracts the body of the news story. Follow-up: "What are the key arguments made by the author in this piece?"

---

2. Add YouTube Source

Function: Downloads the audio from a YouTube video, transcribes it purely offline using your local Whisper binary, and adds the full transcript to the chat context.

• Command: /add_source
• Parameters: url (YouTube link), type: "youtube"

Example 1: Summarizing a Lecture

User Input:

Result: "Downloading video... Transcribing..." → Adds full transcript. Follow-up: "List the 3 main concepts the professor explained in the video."

Example 2: Extracting Info from a Review

User Input:

Result: Transcript is added. Follow-up: "Does the reviewer mention battery life? If so, what did they say?"

---

3. Local File RAG (Retrieval Augmented Generation)

Function: Automatically analyzes PDF, TXT, or MD files you drag into LM Studio. It uses a "Reranker" to find the exact paragraphs relevant to your question, reducing hallucinations.

• Usage: Drag & Drop a file → Ask a question.

Example 1: Technical Manual Q&A

Action: Drag Dishwasher_Manual.pdf into the chat. User Input:

Result: The plugin finds the "Troubleshooting" section in the PDF and provides the exact reset procedure found in the text.

Example 2: Research Paper Analysis

Action: Drag Attention_Is_All_You_Need.pdf into the chat. User Input:

Result: The plugin retrieves the specific architectural details from the paper and explains them.

---

4. Generate Podcast Script

Function: Transforms the current context (files, web pages, or transcripts) into a lively, spoken-word dialogue script between two AI hosts ("Nova" and "Sage").

• Command: /generate_podcast_script
• Parameters: topic, hosts (optional)

Example 1: "Deep Dive" into a Topic

Prerequisite: You have added a YouTube transcript or Web article. User Input:

Result: A script appears:

Nova: Welcome back! Today we're diving into that article you shared about AI. Sage: I read it, and honestly, I'm a bit skeptical about the timeline they proposed. Nova: Really? But the author explicitly stated that...

Example 2: Custom Hosts for a Story

Prerequisite: You uploaded a short story PDF. User Input:

Result:

Detective Holmes: Watson, have you reviewed the case file? Watson: Yes, Holmes. The details about the attic seem peculiar. Detective Holmes: Precisely! Let us examine the evidence...