# Voice & Screenshot Workflow ## Role You help me work efficiently with minimal typing. I'll speak my commands and show you screenshots for visual context. ## Screenshot Review Skill ### Setup Screenshots folder: ~/Screenshots/ (Adjust path to your system's screenshot location) ### Commands - "/ss" or "/screenshots" — Review the most recent screenshot - "/ss [N]" — Review the last N screenshots (e.g., "/ss 5") ### Skill Implementation When I invoke /ss or /screenshots: 1. Look at the screenshot folder at the path above 2. Sort by modification time (most recent first) 3. Read and analyze the specified number of images 4. Respond based on what you see ### Common Use Cases - Design review: "Does this look right?" - Code errors: "What's wrong here?" - Tool help: "How do I use this feature?" - Bug reports: "Why isn't this working?" ## Voice Prompting Tips ### Speak Naturally Voice works best with natural phrasing: - "Take a look at my last screenshot" - "What do you think of this design?" - "Fix the error showing in my terminal" ### Be Specific When Needed - "Look at the last 3 screenshots showing my workflow" - "The screenshot shows a form - help me fill it out" - "Review my code editor screenshot and spot issues" ## Useful Keyboard Shortcuts ### Background Agents - `Ctrl+B` — Send current task to background - `/tasks` — View all running tasks - `/agents` — Review agent status ### Parallel Work Run multiple agents simultaneously: - Start task, Ctrl+B to background - Start another task - Check progress with /tasks ## Multi-Screenshot Workflows ### Collecting Context When gathering information from multiple sources: 1. Screenshot each relevant screen 2. Call "/ss 10" (or however many) 3. Claude reviews all context together ### Example: Tweet Curation - Scroll through feed, screenshot 10 interesting tweets - "/ss 10 - summarize these for my AI recap" - Claude processes all screenshots together ## Rules 1. Screenshot folder must exist and be accessible 2. Images should be readable (not too small/blurry) 3. For sensitive info, review what's in screenshots 4. Clear old screenshots periodically

What This Does

Set up a hands-free Claude Code workflow using voice dictation for prompts and automatic screenshot review for visual context. Almost no typing required - speak your instructions and show Claude what you see.

Quick Start

Step 1: Download the Template

Click Download above to get the CLAUDE.md file, which includes the screenshot review skill.

Step 2: Set Up Voice Dictation

Install Wispr Flow (or similar voice tool):

Set Fn hold key as shortcut
Allow across all displays
Consider a dedicated mic for noisy environments
Pro account recommended

Step 3: Configure Screenshot Folder

Set your system to save screenshots to a specific folder (e.g., ~/Screenshots/).

Step 4: Start Talking

claude

Hold Fn, speak: "Review my last screenshot and tell me what's wrong"

The Setup

Component	Purpose
Wispr Flow	Voice-to-text dictation
Screenshot Folder	Centralized screenshot storage
Screenshot Skill	Reviews recent screenshots on demand
Keyboard Shortcuts	Quick agent management

Wispr Flow Setup

Download Wispr Flow from wispr.com
Set shortcut: Fn key hold works well
Enable everywhere: Allow across all applications
Pro account: Required for best accuracy
Optional: Dedicated USB mic for clarity

Alternative Voice Tools

macOS Dictation (built-in)
Windows Voice Typing (Win+H)
Talon Voice (for power users)

Screenshot Folder Setup

macOS

# Change default screenshot location
defaults write com.apple.screencapture location ~/Screenshots
killall SystemUIServer

Windows

Screenshots → Settings → Change where screenshots are saved

Linux

Configure your screenshot tool to save to ~/Screenshots/

Example Workflows

Design Review

[Take screenshot of design]
Voice: "Review my screenshot and rate this design out of 10"

Claude: Looking at your screenshot, I see a login form with...
Rating: 7/10
Suggestions:
- Increase contrast on submit button
- Add more spacing between fields
- Consider adding password visibility toggle

Debug Session

[Screenshot error message]
Voice: "Look at my last screenshot - what's causing this error?"

Claude: I can see a TypeError in the console. The issue is...

Multi-Screen Context

[Screenshot 5 different screens showing a workflow]
Voice: "/ss 5 - Document this workflow for the team wiki"

Claude: Based on your 5 screenshots, here's the workflow documentation:
1. Start in the Dashboard...
2. Navigate to Settings...
[etc.]

Tips

Speak clearly but naturally - no need for robotic speech
Name screenshots if you want to reference specific ones
Clear regularly to avoid confusion with old screenshots
Combine with context docs for even more power (see Context Library playbook)
Use background agents for long tasks while you continue working

Keyboard Shortcuts Reference

Shortcut	Action
`Fn` (hold)	Voice dictation (Wispr Flow)
`Ctrl+B`	Send task to background
`/tasks`	View running tasks
`/ss`	Review latest screenshot
`/ss N`	Review last N screenshots

Troubleshooting

Voice not working Check Wispr Flow permissions and that your mic is selected

Screenshots not found Verify the folder path in the skill matches your actual screenshot location

Images not readable Take higher-resolution screenshots or zoom in before capturing

Too many old screenshots Clear your screenshot folder periodically or adjust the skill to filter by date

Voice & Screenshot Workflow