🧠 MonoMind – Read Calmer, Feel Safer

Inspiration

Many people on the autism spectrum or with sensory sensitivities struggle with emotional and visual overstimulation while reading online.
We wanted to create a simple, private, and local tool that makes the web calmer — reducing intensity without removing meaning.
MonoMind was inspired by accessibility design principles and neurodiverse user feedback emphasizing clarity, predictability, and emotional safety.

What it does

MonoMind is a Chrome extension that transforms the reading experience using built-in on-device AI models:

• 🕊️ Soften Tone — rewrites emotionally intense or aggressive text into neutral, balanced language using the Rewriter API.
  
• 📖 Summarize Content — condenses long or complex pages into clear, concise summaries via the Summarizer API.
  
• 🧩 Remove Image — temporarily hides distracting images and videos to reduce sensory load.
  
• 🎨 Style Optimize (in development) — leverages the Prompt API to suggest calmer color palettes and layouts.
  

Everything happens locally inside Chrome — no data ever leaves your device.

How we built it

• Manifest V3 architecture with strict CSP compliance (no inline scripts).
  
• Modular structure with feature-based scripts (tone_rewriter.js, simplify.js, image.js, optimize.js).
  
• Uses Chrome’s experimental AI APIs:
  • Rewriter API → tone softening
    
  • Summarizer API → text simplification
    
  • Prompt API → visual optimization
    
• Built and tested on Chrome 127+ (Canary) with on-device Gemini Nano models.
  
• Written in vanilla JavaScript (ES Modules) for clarity and portability.
  

Installation

1. Open chrome://flags and enable:
   
   • #prompt-api-for-gemini-nano
     
   • #rewriter-api
     
   • #summarizer-api
     
   • #optimization-guide-on-device-model
     
2. Restart Chrome.
   
3. Open chrome://extensions, enable Developer mode, and click Load unpacked.
   
4. Select your monomind/ project folder.
   
5. The MonoMind icon will appear in your toolbar.

Challenges we ran into

• Experimental APIs: Rewriter, Summarizer, and Prompt APIs are still in early access, often requiring specific Chrome versions and flags.
  
• Origin Trial tokens: Maintaining valid tokens for local builds and MV3 CSP compliance was tricky.
• Prompt API multimodal complexity: integrating image inputs was difficult — Chrome’s on-device model often failed to interpret cross-origin images.
  

Accomplishments that we're proud of

• Achieved on-device AI processing, ensuring full privacy.
  
• Created a unique tool that combines emotional tone moderation, readability improvement, and sensory control — all inside the browser.
  
• Built a real prototype that works offline and demonstrates Chrome’s AI capabilities for accessibility.

What we learned

• Chrome’s built-in AI stack (Gemini Nano) can power real accessibility tools entirely offline.
  
• Simplicity in design (gentle colors, calm icons, plain text) is as impactful as complex features.
  
• Writing accessible AI experiences requires empathy.
  
• The Prompt API’s multimodal design (text + image) is powerful but still limited.

What's next for MonoMind

• Try to implement multimodal Prompt API support to let AI visually analyze web pages and suggest layout or color adjustments automatically.
  
• Implement user preferences for tone softness and image filtering.
  
• Extend support for audio content (reduce loud auto-play or strong sound cues).
  
• Publish to Chrome Web Store with accessibility certification.
  

Built With

• javascript
• promptapi
• rewriteapi
• summarizeapi