Obsidian Auto-Processor

Automatically process notes in watched folders using OpenAI's GPT-5 API with custom slash commands. Monitor specific folders, detect new notes, run them through configurable command chains, and save processed outputs for review.

Features

• 🤖 GPT-5 Powered - Uses OpenAI's latest GPT-5 models (mini, nano, or full)
• 📁 Folder Watching - Automatically process new notes in specified folders
• ⛓️ Command Chains - Run up to 3 commands in sequence with reasoning context preservation
• 💰 Cost Tracking - Monitor API costs with monthly budget alerts
• 🎯 Custom Commands - Parse and use your own slash command library from markdown
• ⚙️ Flexible Configuration - Per-folder settings for model, reasoning effort, and verbosity

Installation

From Release

1. Download the latest release from the releases page
2. Extract files to <vault>/.obsidian/plugins/auto-processor/
3. Reload Obsidian
4. Enable the plugin in Settings → Community Plugins

From Source

1. Clone this repository into your vault's plugins folder:

   cd <vault>/.obsidian/plugins
   git clone https://github.com/yourusername/obsidian-auto-processor.git

2. Install dependencies:

   cd obsidian-auto-processor
   npm install

3. Build the plugin:

   npm run build

4. Reload Obsidian and enable the plugin

Setup

1. Configure API Key

1. Go to Settings → Auto-Processor
2. Enter your OpenAI API key (starts with sk-)
3. Click "Test" to verify connection

2. Load Command Library

1. Create or use an existing slash command markdown file
2. Format commands like this:

### /script60
**Label:** 60-second Explainer
**Category:** Structure
**Prompt:**
> Transform {{insert idea}} into a compelling 60-second explainer script.
> Focus on clarity and engagement.

---

### /punchup
**Label:** Add Punch
**Category:** Enhancement
**Prompt:**
> Add rhythm and punch to {{paste}}. Make it more engaging and dynamic.

3. In settings, set "Library File Path" to your markdown file
4. Click "Reload" to import commands

3. Create Watch Folder

1. Click "+ Add Watch Folder"

2. Configure:

   • Input Folder: Where to watch for new files (e.g., Inbox/Raw Scripts)
   • Output Folder: Where to save processed files (e.g., Processed/Review)
   • Command Chain: Comma-separated command names (e.g., script60, punchup, qprompt)
   • Enabled: Toggle to activate/deactivate

3. Save settings

Usage

Automatic Processing

When enabled, any new markdown file added to a watched folder will be automatically processed through its command chain.

Manual Processing

Use the command palette (Ctrl/Cmd + P):

• Process current file - Run the active file through configured commands
• Reload command library - Refresh command definitions
• Toggle auto-processing - Enable/disable automatic processing

Output Format

Processed files are saved with _processed.md suffix and include:

---
source: original-filename
processed: 2025-10-09T12:00:00.000Z
commands: script60 → punchup → qprompt
total_cost: $0.0124
status: success
---

# original-filename (Processed)

[Final output from last command in chain]

---

## Processing Details

### Step 1: /script60
**Cost:** $0.0042
**Tokens:** 150 input → 450 output (120 reasoning)

<details>
<summary>View Output</summary>

[Step 1 output]

</details>

[Additional steps...]

Configuration Options

API Settings

• Model: Choose between gpt-5, gpt-5-mini (default), or gpt-5-nano
• Reasoning Effort: minimal, low (default), medium, high
• Verbosity: low, medium (default), high
• Max Output Tokens: Maximum response length (default: 2000)

Cost Tracking

• View current month spending vs. budget
• Set monthly budget limit for warnings
• Per-command cost breakdown in outputs

Processing Options

• Auto-process new files: Enable/disable automatic processing
• Show completion notifications: Display notices when processing completes

Cost Optimization

Target cost per note: $0.003-0.008

Tips:

• Use gpt-5-mini for most tasks (75% cheaper than gpt-5)
• Set reasoning effort to low for simple transformations
• Use medium verbosity to balance detail and cost
• Limit command chains to 3 steps or fewer

Development

Build Commands

# Development mode (watch for changes)
npm run dev

# Production build
npm run build

Project Structure

src/
├── main.ts                 # Plugin entry point
├── settings.ts             # Data models & defaults
├── watcher.ts              # Folder monitoring
├── processor.ts            # Command chain execution
├── commandLibrary.ts       # Markdown parser
├── ui/
│   └── settingsTab.ts      # Settings interface
└── utils/
    ├── gpt5Client.ts       # API wrapper
    └── templateParser.ts   # Variable replacement

Troubleshooting

API Connection Failed

• Verify your API key is correct
• Check internet connection
• Ensure you have OpenAI API credits

Commands Not Loading

• Check library file path is correct
• Verify markdown format matches expected structure
• Look for console errors in Developer Tools (Ctrl/Cmd + Shift + I)

File Not Processing

• Confirm folder path uses forward slashes (e.g., Inbox/Raw not Inbox\Raw)
• Check that command names in chain match library exactly
• Ensure auto-processing is enabled
• Verify file is markdown (.md)

Budget Exceeded Warning

• Review cost tracking in settings
• Consider using gpt-5-nano for lower costs
• Reduce max_output_tokens setting
• Lower reasoning effort to minimal or low

License

MIT

Support

For issues, feature requests, or questions:

• GitHub Issues: yourusername/obsidian-auto-processor

---

Built with ❤️ for Obsidian