This Logseq plugin integrates OpenAI's GPT-3 and DALL-E models into the note-taking workflow, enabling AI-assisted content generation, summarization, and image creation directly within the Logseq editor. It targets Logseq users seeking to leverage advanced AI capabilities for enhanced productivity and creativity in their knowledge management.

How It Works

The plugin provides slash commands (/gpt, /gpt-block, /gpt-page, /whisper, /dalle) to trigger AI actions. It sends selected text or entire pages to OpenAI's API (defaulting to gpt-3.5-turbo) and inserts or replaces content. Custom prompt templates can be defined using a prompt-template:: property and code blocks, allowing users to tailor AI responses for specific tasks like generating outlines or flashcards. Whisper API is used for speech-to-text transcription, and DALL-E for image generation.

Quick Start & Requirements

• Install via Logseq marketplace by searching for "gpt3" or "openai".
• Requires an OpenAI API key.
• Supports gpt-3.5-turbo, text-davinci-003, and gpt-4 (if available).
• Whisper supports m4a, mp3, mp4, mpeg, mpga, wav, webm.

Highlighted Details

• AI-assisted note-taking with GPT-3 and DALL-E.
• Supports custom prompt templates and block/page-level operations.
• Includes Whisper for audio transcription.
• Offers options to inject prefixes or markdown formatting into AI output.

Maintenance & Community

• Active status indicated by shields.
• Open to contributions via GitHub issues and pull requests.
• Author: @briansunter.

Licensing & Compatibility

• MIT License.
• Compatible with Logseq, an open-source, privacy-first knowledge base.

Limitations & Caveats

GPT-3 can produce subtly incorrect or misleading output and requires user verification. OpenAI API usage is subject to quotas and rate limits, and certain audio formats (like .aac from Logseq mobile) are not directly supported by Whisper. Users must manage their OpenAI API key and potential costs.