LightSession for ChatGPT

Keep ChatGPT fast by keeping only the last N messages in the DOM. 100% local, privacy-first Firefox extension.

Features

• Automatic Trimming: Keeps only the last N messages visible (default: 10, range: 1-100)
• Memory Optimization: Reduces memory usage by 25-50% on long conversations (100+ messages)
• Performance: Maintains 60fps scrolling with <16ms DOM operation batching
• Intelligent Preservation: Optionally preserves system and tool messages beyond normal limit
• Scroll Pause: Automatically pauses trimming when reviewing older messages
• Privacy-First: Zero network requests, all processing happens locally
• Reversible: Page refresh restores full conversation history
• Quick Toggle: Enable/disable via popup without losing settings

Installation

Firefox Add-ons (Recommended)

Coming soon to addons.mozilla.org

Manual Installation (Development)

1. Clone this repository:

   git clone https://github.com/11me/light-session.git
   cd light-session

2. Install dependencies:

   npm install

3. Build the extension:

   npm run build

4. Load in Firefox:

   • Open Firefox
   • Navigate to about:debugging#/runtime/this-firefox
   • Click "Load Temporary Add-on"
   • Select extension/manifest.json

Usage

Basic Usage

1. Open a long ChatGPT conversation (or create one)
2. Click the LightSession icon in Firefox toolbar
3. Extension is enabled by default - trimming happens automatically
4. Adjust the slider to change how many messages to keep (1-100)

Settings

• Extension Enabled: Toggle trimming on/off
• Keep last N messages: Adjust message retention limit (1-100)
• Preserve system/tool messages: Keep system/tool messages beyond normal limit
• Pause when scrolled up: Stop trimming while reviewing history
• Debug mode: Enable console logging for troubleshooting
• Refresh: Reload page to restore full conversation history

Keyboard Accessibility

• All controls are keyboard accessible (Tab/Shift+Tab to navigate)
• Enter/Space to toggle checkboxes and buttons
• Arrow keys to adjust slider

How It Works

LightSession uses a non-destructive trimming approach:

1. Detection: Uses multi-tier selector strategy to find message nodes (resilient to UI changes)
2. Classification: Identifies message roles (user, assistant, system, tool)
3. Calculation: Determines which messages to keep based on settings
4. Batching: Removes excess nodes in idle time (<16ms chunks for 60fps)
5. Markers: Replaces removed nodes with comment markers for debugging

Important: Trimming only affects the DOM (what's displayed). The actual conversation data remains on OpenAI's servers. Refreshing the page restores everything.

Privacy & Security

• Zero Network Requests: No data leaves your browser
• Local Storage Only: Settings stored in browser.storage.local
• No Telemetry: No analytics, tracking, or reporting
• Minimal Permissions: Only requires storage permission
• Domain Restricted: Only runs on chat.openai.com and chatgpt.com
• Open Source: All code is auditable

See PRIVACY.md for detailed privacy policy.

Development

Requirements

• Node.js ≥18.0
• npm ≥9.0
• Firefox ≥115

Build Commands

# Install dependencies
npm install

# Build (one-time)
npm run build

# Watch mode (rebuilds on file changes)
npm run watch

# Lint code
npm run lint

# Format code
npm run format

# Run in Firefox Developer Edition
npm run dev

# Run in Firefox stable
npm run dev:stable

# Package for distribution
npm run package

# Clean build artifacts
npm run clean

Project Structure

extension/
├── src/
│   ├── content/        # Content scripts (run in ChatGPT pages)
│   ├── background/     # Background script (settings management)
│   ├── popup/          # Popup UI (HTML/CSS/JS)
│   └── shared/         # Shared types, constants, utilities
├── dist/               # Compiled output (TypeScript → JavaScript)
├── icons/              # Extension icons
└── manifest.json       # Firefox extension manifest

specs/                  # Feature specifications and design docs
tests/                  # Manual test procedures

Architecture

• State Machine: Trimmer uses IDLE → OBSERVING → PENDING_TRIM → TRIMMING states
• Multi-Tier Selectors: A (data attributes) → B (test IDs) → C (structural + heuristics)
• Debounced Observer: MutationObserver batches DOM changes (75ms debounce)
• Idle Callback: Removes nodes during browser idle time (requestIdleCallback)
• Fail-Safe: Minimum 6 message threshold, aborts on uncertainty

Compatibility

• Browser: Firefox ≥115 (Manifest V3 with background.scripts)
• OS: Windows, macOS, Linux
• ChatGPT: Optimized for current UI (as of 2025), resilient to minor changes

Known Limitations

• Firefox Only: Chrome/Edge not supported (different WebExtensions API)
• UI Dependency: Major ChatGPT UI redesigns may require selector updates
• No Server-Side: Cannot reduce actual conversation size on OpenAI servers
• Manual Testing: No automated test suite (manual testing with Firefox DevTools)

Troubleshooting

Extension not trimming messages

1. Check if extension is enabled (popup toggle should be ON/green)
2. Verify you're on ChatGPT (chat.openai.com or chatgpt.com)
3. Ensure conversation has >6 messages (fail-safe threshold)
4. Check if streaming is in progress (trimming pauses during generation)
5. Enable debug mode in popup and check Firefox console (F12)

Settings not persisting

1. Check Firefox console for storage errors
2. Verify storage permission in manifest
3. Try disabling/re-enabling extension

Performance issues

1. Lower the "Keep last N messages" slider value
2. Enable "Preserve system/tool messages" (may reduce removable nodes)
3. Check Firefox DevTools → Performance tab for profiling

Contributing

Contributions welcome! Please:

1. Fork the repository
2. Create a feature branch (git checkout -b feature/my-feature)
3. Make your changes
4. Run linter (npm run lint)
5. Format code (npm run format)
6. Build successfully (npm run build)
7. Test manually in Firefox (npm run dev)
8. Commit changes (git commit -m "feat: add X")
9. Push to branch (git push origin feature/my-feature)
10. Open a Pull Request

License

ISC License - see LICENSE for details

Credits

Created as part of the Specify workflow for feature specification and implementation.

Support

• Issues: GitHub Issues
• Discussions: GitHub Discussions

---

Disclaimer: This is an unofficial extension not affiliated with OpenAI. Use at your own risk.