Special Tags

Special tags modify how Agentastic's AI processes your requests. These system-level commands control display capture, model selection, and other core behaviors.

Screen Tags

The /screen tag enables AI to see and analyze content from your displays. Perfect for getting help with what's on your screen.

Basic Usage

/screen What's in this spreadsheet?
/screen Explain this error message
/screen Help me understand this diagram

Multi-Monitor Support

Specify which screen to capture:

• /screen:0 or /screen:active - Screen where launcher appears (default)
• /screen:1 - Primary display
• /screen:2 - Second monitor
• /screen:3 - Third monitor

Examples

/screen:1 Analyze the chart on my main display
/screen:2 What's in the terminal on my second monitor
/screen:active Describe what you see

Screen Tag Tips

1. Privacy First - Only enables when you use the tag
2. Clear Context - AI sees exactly what you see
3. Multiple Screens - Great for comparing displays
4. Error Handling - Gracefully handles invalid screen numbers

Common Use Cases

• Debugging code - "What's wrong with this function?"
• Understanding UIs - "How do I use this application?"
• Reading content - "Summarize this article"
• Data analysis - "What patterns do you see in this data?"

Image Paste

You can now paste images directly into the launcher for visual analysis, without needing to enable screen capture.

How It Works

1. Copy any image - From screenshots, web browsers, image editors, etc.
2. Paste with Cmd+V - The image appears as [📎 image.png] in your text
3. Ask your question - Type normally around the placeholder
4. Send to AI - Press Enter to include the image with your prompt

Examples

# Analyze a screenshot
[Paste image] "What's the error in this screenshot?" [📎 image.png]

# Understand a diagram
[Paste image] "Explain this architecture diagram" [📎 image.png]

# Extract text from images
[Paste image] "What text is in this image?" [📎 image.png]

# Get help with UI
[Paste image] "How do I fix this layout issue?" [📎 image.png]

Key Features

• Visual indicator - [📎 image.png] shows an image is attached
• Automatic handling - Placeholder removed when sending
• Regular paste works - Cmd+V still pastes text when no image in clipboard
• One image at a time - New paste replaces previous image

Image Paste vs Screen Capture

Model Tags

The /model tag lets you choose which AI model processes your request. Different models offer different trade-offs between speed and capability.

Available Models

Fast Model (/model:fast)

• Response time: <1 second
• Best for: Quick questions, simple tasks
• Token limit: Moderate
• Cost: Lowest

Smart Model (/model:smart)

• Response time: 1-2 seconds
• Best for: General purpose, balanced performance
• Token limit: Higher
• Cost: Moderate

Genius Model (/model:genius)

• Response time: 2-5 seconds
• Best for: Complex reasoning, creative tasks
• Token limit: Highest
• Cost: Premium

Model Selection Examples

/model:fast Quick grammar check
/model:smart Write a professional email
/model:genius Analyze this complex legal document

Automatic Model Selection

Without a /model tag, Agentastic automatically selects based on:

• Prompt complexity
• Task requirements
• Response time preferences
• Cost settings

Model Comparison

Vision Tags

The /vision tag is a shorthand for enabling visual analysis, similar to the eye icon in the launcher.

Usage

/vision What's in this image?
/vision /model:smart Analyze this UI design
/vision:off Disable visual mode

Vision Capabilities

• OCR - Extract text from images
• Object Recognition - Identify items and people
• UI Analysis - Understand interfaces
• Chart Reading - Interpret data visualizations

Tab Selection Tag

The /tab: tag opens an interactive browser tab selector for including tab context in your prompts.

Quick Usage

/tab: Analyze these pages
/tab: Compare these documentation tabs
/tab: Summarize content from selected tabs

Multi-Tab Selection

• Single Click - Select one tab and confirm
• Cmd+Click - Toggle multiple tabs
• Shift+Click - Select range
• Cmd+A - Select all filtered tabs

See the complete Tab Selection Guide for detailed information.

System Tags

Additional special tags for system control:

Format Tags

Control output formatting:

/format:markdown Generate with markdown formatting
/format:plain Simple text only
/format:code Optimize for code output
/format:json Structure as JSON

Language Tags

Specify response language:

/lang:es Respond in Spanish
/lang:fr French response
/lang:de German output

Style Tags

Modify response style:

/style:formal Professional tone
/style:casual Friendly conversation
/style:technical Detailed technical writing
/style:creative Imaginative responses

Combining Special Tags

Special tags work together for precise control:

/screen:1 /model:genius /format:markdown
Analyze this dashboard and create a markdown report

/vision /lang:es /style:formal
Describe what you see in formal Spanish

/model:fast /format:json
Quick JSON summary of this data

Tag Priority and Conflicts

When tags conflict, this priority applies:

1. Security/Permission tags
2. Model selection
3. Input modifiers (/screen, /vision)
4. Output modifiers (/format, /lang)
5. Style modifiers

Advanced Tag Usage

Conditional Tags

Use tags based on context:

/screen:{active|1} Use active screen, fallback to screen 1
/model:{smart|fast} Try smart model, fallback to fast

Tag Variables

Reference tags in expansions:

Trigger: analyze
Content: /screen:{{screen_num}} /model:{{model_pref}} Analyze this

Tag Shortcuts

Create aliases in settings:

/s → /screen:active
/g → /model:genius
/jp → /lang:ja

Error Handling

Invalid Screen Number

/screen:99 Show me
> Warning: Screen 99 not found. Using active screen.

Unknown Model

/model:ultra Process this
> Notice: Model 'ultra' not recognized. Using smart model.

Permission Required

/screen Analyze without permission
> Error: Screen recording permission required. Enable in Settings.

Best Practices

Screen Tags

1. Specify screen numbers for multi-monitor setups
2. Use :active for flexibility
3. Remember privacy - only capture what's needed
4. Combine with specific questions

Model Tags

1. Start with fast for simple tasks
2. Upgrade to genius for complex work
3. Consider cost/speed trade-offs
4. Let auto-selection work when unsure

Format Tags

1. Match format to destination
2. Use markdown for rich content
3. JSON for structured data
4. Plain for maximum compatibility

Performance Considerations

Screen Capture Impact

• First capture: 100-200ms
• Subsequent captures: 50ms (cached)
• Large screens: May increase processing time
• Multiple screens: Parallel capture when possible

Model Performance

• Fast: Optimized for speed
• Smart: Balanced caching
• Genius: May queue during peak times

Troubleshooting

Tags Not Working

1. Check spelling exactly
2. Ensure proper format /tag:option
3. Verify permissions granted
4. Update to latest version

Slow Performance

1. Use appropriate model for task
2. Disable vision if not needed
3. Specify single screen instead of all
4. Clear cache if issues persist

Next Steps

• Explore Tool Tags for productivity features
• Learn about /agent and /search
• Master tag combinations for workflows

Special tags give you fine-grained control over Agentastic's AI capabilities. Use them to optimize for your specific needs!