SmartScribe for Obsidian

An Obsidian plugin that uses AI to intelligently generate titles, tags, categories, and summaries for your notes, as well as optimize your writing. Supports multiple AI platforms.

Plugin ID: smartscribe

Supported AI Platforms

Platform	Name	Default Model
Claude	Anthropic Claude	claude-sonnet-4-6
OpenAI	OpenAI GPT	gpt-4o
DeepSeek	DeepSeek	deepseek-chat
Qwen	Alibaba Tongyi Qwen	qwen-plus
Zhipu AI	Zhipu GLM	glm-4
Spark	Xunfei Spark	pro

Features

AI Metadata Generation

• 🤖 Multi-Platform Support: Supports 6 mainstream AI platforms, freely switchable
• 🏷️ Tag Reuse: Prioritizes existing tags in your Vault for consistency
• 📂 Category Suggestions: Supports category recommendations based on folder structure
• ✏️ Preview & Edit: Preview and edit metadata before applying
• 🖱️ Context Menu: Right-click menu in file explorer and editor

AI Text Optimization

• ✨ Smart Writing Assistant: Optimize selected text or entire document with one click
• 📝 Smart Improvements: Enhances expression, structure, grammar, and readability
• 🎯 Dual-Panel Preview: Left panel shows optimized text (editable), right panel shows improvement suggestions
• 🔄 Direct Apply: Apply optimizations directly or preview first based on settings

Flexible Configuration

• ⚙️ Independent Configuration: Independent configuration for each platform (API Key, model, parameters)
• 🔧 Granular Control: Separate settings for metadata generation and text optimization features

Installation

Manual Installation

1. Download the latest release from GitHub
2. Extract the files to your Vault's .obsidian/plugins/smartscribe/ folder
3. Enable the plugin in Obsidian settings

Configuration

1. Open plugin settings
2. Select your preferred AI Platform
3. Configure the API Key for the selected platform
4. Choose the Model to use
5. Adjust parameters like Temperature and Max Tokens as needed
6. Configure feature options (max tags, folder structure, preview settings, etc.)

API Key Sources

• Claude: https://console.anthropic.com/
• OpenAI: https://platform.openai.com/
• DeepSeek: https://platform.deepseek.com/
• Qwen: https://dashscope.aliyun.com/
• Zhipu AI: https://open.bigmodel.cn/
• Spark: https://xinghuo.xfyun.cn/

Usage

Metadata Generation

Command Palette

• Generate AI Metadata for Current Note: Generate and apply metadata directly
• Generate AI Metadata for Current Note (with Preview): Show preview after generation, edit before applying

Context Menu

Right-click on a Markdown file in the file list and select "AI Generate Metadata"

Editor Menu

Right-click in the editor to access the AI metadata generation option

Text Optimization

Command Palette

• AI Optimize Text: Optimize selected text (or full document if no selection)
• AI Optimize Text (with Preview): Always show preview before applying

Context Menu

Right-click in the editor and select "AI Optimize Text" to optimize selected text or the entire document

How It Works

1. Select Text: Select text in the editor, or leave unselected to optimize the entire document
2. Trigger Optimization: Use command palette or right-click menu
3. Preview (Optional): If preview is enabled, a dual-panel modal appears:
   • Left Panel: Optimized text (editable, supports formatting)
   • Right Panel: Improvement suggestions (bullet points)
4. Apply: Click "Apply" to replace the original text with the optimized version

Generated Metadata Format

---
date: 2024-01-15 10:30:00
title: Generated Title
tags:
  - tag1
  - tag2
  - tag3
category: Category Name
keywords:
  - keyword1
  - keyword2
summary: Article summary...
share: true
---

How It Works

Metadata Generation

1. Analysis Phase: The plugin collects existing tags from your Vault using Obsidian's MetadataCache API
2. Generation Phase: Calls the selected AI platform API to analyze article content, passing existing tags and categories as references
3. Application Phase: Writes the generated metadata to the file's YAML frontmatter

Text Optimization

1. Content Capture: Captures selected text or the entire document content
2. AI Processing: Sends content to AI with optimization instructions
3. Result Parsing: Separates optimized text from improvement suggestions
4. Preview/Application: Shows preview (if enabled) or directly applies the optimized text

Settings

Model Settings

• AI Platform: Choose from 6 supported platforms
• API Key: Your API key for the selected platform
• Model: Select the model to use
• Base URL: Optional, for custom API endpoints or proxies
• Temperature: Generation randomness (0-2, lower is more deterministic)
• Max Tokens: Maximum tokens to generate

Metadata Generation Settings

• Max Tags: Maximum number of existing tags to reference (1-10)
• Include Folder Structure: Use file location as category suggestions
• Show Context Menu: Show metadata option in right-click menus
• Show Preview Before Generation: Show preview window before generating metadata

Text Optimization Settings

• Show Optimize Menu: Show "AI Optimize Text" in editor right-click menu
• Show Preview Before Optimization: Show preview window before optimizing text

Notes

• Requires a valid API Key, configured separately for each platform
• API calls will incur costs, please monitor your usage
• Recommended to backup important files before generation
• Adjust the Temperature parameter to control creativity of results
• Text optimization works best on substantial text (more than a few sentences)

Development

# Install dependencies
npm install

# Development mode (auto-build)
npm run dev

# Production build
npm run build

Support

If you encounter any issues or have feature requests, please submit an issue on GitHub.

License

MIT