Logseq Ask PDF Plugin

A Logseq plugin that allows you to ask questions about PDF highlights (text or images) using Large Language Models (LLMs).

Demo

Note

The demo video uses the following paper as an example:

Frans, K., Park, S., Abbeel, P., & Levine, S. (2024). Unsupervised Zero-Shot Reinforcement Learning via Functional Reward Encodings. arXiv preprint arXiv:2402.17135.
URL: https://arxiv.org/abs/2402.17135
This paper is used for demonstration purposes only and is not affiliated with this plugin's development.

Important

This plugin is under active development. Contributions and pull requests are highly welcome!

Features

• Ask questions about PDF highlights directly within Logseq
• Supports asking both text and image highlights
• Use the entire PDF as context
• Seamless integration with Logseq's UI
• Compatible with OpenAI API and local models that are OpenAI API-compatible

Installation

1. Open Logseq
2. Go to Settings > Plugins
3. Search for "Ask PDF"
4. Click Install

Configuration

Before using the plugin, you need to set up your OpenAI API key:

1. Go to the plugin settings
2. Enter your OpenAI API key or the API key for your local OpenAI-compatible model in the designated field
3. (Optional) If using a local model, set the Embedding Model Host and LLM Model Host to your local server's URL

Usage

1. Highlight text or an image in your PDF (and paste it)
2. In Logseq, find the block with the corresponding UUID (it will look like ((uuid-uuid-...)))
3. Type /ask pdf in that block
4. Wait for the LLM to generate a response

Note

You can move your cursor or continue working in Logseq while the response is being generated.

License

MIT

Support

If you encounter any issues or have questions, please open an issue on GitHub.