A sophisticated web application for browsing, searching, and analyzing Claude Code chat histories with automatic AI-powered summarization.

This web application provides a powerful interface to browse, search, and analyze your Claude Code chat histories across all projects. It automatically analyzes conversations using AI to generate summaries, bullet points, and make your chat history searchable and organized.

• 📊 Automatic Analysis: Conversations are automatically analyzed on page load with AI-generated summaries
• 🔍 Full-Text Search: Search across all conversations with real-time highlighting and navigation
• 📁 Multi-Project Support: Browse conversations organized by project, not just a single project
• 💡 Smart Summaries: Each conversation includes:
  • One-line summary for quick understanding
  • Detailed paragraph summary (2-5 sentences)
  • Up to 8 bullet points of key accomplishments
• 📥 Export to Markdown: Download any conversation as a formatted Markdown file
• 🎨 Dark Theme: Comfortable viewing with VS Code-inspired dark theme
• ⚡ Real-time Updates: Automatically detects and analyzes new messages
• 🔄 Incremental Analysis: Only analyzes changed conversations for efficiency

1. Clone or navigate to the claude folder:

cd /Users/jeffk/Developement/provider-search/claude

2. Install dependencies:

npm install

3. Start the server:

npm start

4. Open your browser to: http://localhost:3101

To have the Claude History Viewer start automatically on system boot:

1. Copy the launch daemon configuration:

cp com.dataintegrities.claude-history.plist ~/Library/LaunchAgents/

2. Load the launch daemon:

launchctl load ~/Library/LaunchAgents/com.dataintegrities.claude-history.plist

3. To stop the service:

launchctl unload ~/Library/LaunchAgents/com.dataintegrities.claude-history.plist

4. To check service status:

launchctl list | grep claude-history

Logs are stored in ./logs/ when running as a service.

For auto-reload during development:

npm run dev

claude/
├── server.js                    # Express server with AI analysis
├── package.json                 # Node dependencies
├── history-index.json          # Cached analysis data
├── com.dataintegrities.claude-history.plist  # macOS launch daemon
├── logs/                       # Service logs (when running as daemon)
├── views/                      # EJS templates
│   ├── index.ejs              # Homepage with project grouping
│   └── chat.ejs               # Individual chat viewer
├── public/                     # Static assets
│   ├── css/
│   │   ├── claude-sidebar.css  # Main styling
│   │   └── claude-chat.css     # Chat conversation styling
│   └── js/
│       ├── search.js          # Search functionality
│       └── chat-search.js     # In-chat search navigation
└── README.md                   # This file

The app reads Claude Code conversation histories from ALL projects:

~/.claude/projects/

Each project has its own subdirectory with encoded path names.

1. File Discovery: Scans for .jsonl files in the Claude history directory
2. Message Extraction: Parses each JSONL file to extract user and assistant messages
3. Preview Generation: Extracts the first user message from each chat for the preview
4. Sorting: Orders chats by modification time (newest first)

• User Messages: Extracts content from entry.message.content
• Assistant Messages: Handles both text and tool_use content types
• Tool Calls: Displays tool usage with formatted JSON input
• Markdown Rendering: Uses marked library for HTML conversion

• GET /: Homepage with all conversations grouped by project
• GET /chat/:project/:id: View individual chat
• GET /download/:project/:id: Download chat as Markdown
• GET /history-index.json: View analysis cache data
• GET /analyze-chats: Manually trigger analysis (usually automatic)

• Port: Server runs on port 3101 (configurable in server.js)
• Chat History: Reads from ~/.claude/projects/
• Cache: Analysis results stored in history-index.json
• Logs: Service logs in ./logs/ directory

• Express.js: Web framework for server and routing
• EJS: Template engine for dynamic HTML generation
• Marked: Markdown to HTML conversion
• Moment.js: Date formatting and relative time display
• AI Analysis: Built-in Claude analysis for summarization
• Nodemon: Development auto-reload

• Use the 🔍 Find link to search across all conversations
• Search highlights appear in yellow with navigation
• Use Ctrl+N/P or Alt+↓/↑ to navigate between search results

• New conversations are analyzed automatically when you load the page
• A subtle notification shows "X conversations updated" if changes were detected
• Click the notification to see detailed analysis results

• Click 📥 Download on any chat to export as Markdown
• Exports include all messages with formatting preserved

• Integration with Proxmox VM management agents (in development)
• Automated infrastructure management capabilities
• Remote server configuration through Claude conversations

• DNS: claudehistory.dataintegrities.com via Cloudflare
• Reverse Proxy: Nginx configuration for:
  • External domain routing to port 3101
  • Internal network restriction (192.x.x.x only)
  • SSL termination for secure access
• Security: Restrict access to local network only

• 🔐 Authentication layer
• 📊 Analytics dashboard
• 🏷️ Tag and categorize conversations
• ⭐ Favorite important chats
• 📱 Mobile-responsive improvements

This Claude Chat History Viewer is part of the larger Provider Search project ecosystem, providing a way to review and reference past Claude Code conversations during development work.

When continuing work after a Claude Code session crash or interruption:

1. Identify the Chat History Location:

   ~/.claude/projects/-Users-jeffk-Developement-provider-search-provider-search-flutter/
   

2. Find the Most Recent JSONL File:

   • Look for the newest .jsonl file by modification date
   • The crashed thread will be the most recently modified file

3. Locate the Continuation Marker:

   • Search for "CONTINUATION MARKER:" in the last conversation
   • The marker format is: CONTINUATION MARKER: CLAUDE-WORK-YYYY-MM-DD
   • This indicates where work should resume

4. Extract Context:

   • Read the last 100 lines of conversation before the crash
   • Note the current project state and what was being worked on
   • Identify any incomplete tasks or next steps

5. Resume Work:

   • Start from the claude folder directory (now its own git repo)
   • Reference the continuation marker and context
   • Continue with the identified next steps

• ✅ Implemented search highlighting features
• ✅ Search terms from homepage appear highlighted in yellow on chat detail pages
• ✅ Claude folder is now its own git repository at https://github.com/Data-Integrities/claude
• 🔄 Working from claude folder directory in separate VSCode instance

This documentation ensures continuity across Claude Code sessions and helps quickly resume work from where it left off.

• Check if port 3101 is already in use: lsof -i :3101
• Verify Node.js path: which node
• Check logs: tail -f logs/stderr.log

• Ensure history-index.json is writable
• Check for JavaScript errors in browser console
• Manually trigger analysis: visit /analyze-chats

The application is deployed at: https://claude-search.dataintegrities.com/

• Access restricted to local network (192.x.x.x addresses only)
• SSL/TLS enabled for secure access
• See DEPLOYMENT.md for infrastructure details

This is part of the Provider Search ecosystem. For issues or suggestions:

• GitHub: https://github.com/Data-Integrities/claude
• Contact: [email protected]

MIT License - Feel free to modify and use for your own Claude Code chat browsing needs.