ChatGPT History Archiver

Offline archive and viewer for your ChatGPT conversation history.

Quick Start

1. Capture Conversations

1. Navigate to chatgpt.com
2. Open browser DevTools (F12 or right-click → Inspect)
3. Switch to the Network tab
4. Browse through your conversations in ChatGPT (click each one to load it)
5. Click on Export HAR icon (resembles a download icon)
6. Save the HAR file to the data/ directory in this project

Tip: Repeat this process for different conversation categories or time periods.

2. Extract Conversations

Run the extraction script:

node index.js

This will:

• Process all HAR files in data/
• Extract conversation data (title, timestamps, messages)
• Generate output/metadata.json with conversation index
• Save individual conversations to output/conversations/{id}.json

3. View Your Archive

Option A: Static Hosting (Basic Features)

Open index.html directly or deploy to any static host (GitHub Pages, Netlify, etc.):

# Open directly in browser
open index.html

# Or serve locally
npx serve .

Available features:

• Tree navigation by category
• Title-based search
• Markdown rendering
• Conversation viewing
• Metadata display

Option B: PHP Server (Full Features)

Serve with PHP to enable advanced features:

# Using PHP built-in server
php -S localhost:8000

# Or deploy to any PHP-enabled host (Apache, nginx + PHP-FPM)

Additional features with PHP:

• Content Search: Full-text search through all message content
• Category Editing: Change conversation categories with automatic backups

Features

Core Features (Always Available)

• Tree Navigation: Conversations organized by category
• Title Search: Filter conversations by title, ID, or category
• Markdown Support: Properly rendered code blocks, lists, and formatting
• Role-Based Styling: Color-coded messages (user/assistant/system)
• Metadata Display: Timestamps, model info, and referenced URLs
• Deep Links: Click conversation IDs to open in ChatGPT

PHP-Only Features

• Content Search: Search through all message text across conversations
• Category Management: Edit conversation categories with dropdown selector
  • Change existing categories
  • Create new categories
  • Automatic timestamped backups before changes

Requirements

• Node.js (for extraction script)
• Modern web browser (for viewer)
• PHP 7.0+ (optional, for advanced features)

File Structure

├── data/                        # Place HAR files here
├── output/
│   ├── metadata.json            # Conversation index
│   ├── metadata.backup.*.json   # Automatic backups (PHP feature)
│   └── conversations/           # Individual conversation files
├── lib/
│   ├── parser.js                # HAR parsing logic
│   └── metadata.js              # Metadata management
├── index.js                     # Extraction script
├── index.html                   # Conversation viewer
├── make-stat.js                 # Statistics generation (optional)
├── view-stat.html               # Statistics dashboard (optional)
├── search.php                   # Content search endpoint (requires PHP)
└── update-category.php          # Category update endpoint (requires PHP)

Bonus: Statistics Dashboard

If you're curious about your ChatGPT usage patterns, there's an optional statistics viewer with visual charts.

Generate Statistics

node make-stat.js

This analyzes all conversations and generates output/stats.json with metrics including:

• Volume & activity (message counts, conversation distribution)
• Length metrics (character/word counts, longest/shortest conversations)
• Temporal patterns (activity over time, hourly/daily trends)
• Linguistic analysis (questions, code blocks, sentences)
• Interaction dynamics (prompt types, turn counts)
• Sentiment indicators (emojis, politeness, hedging language)
• Model usage breakdown

View Statistics

Open view-stat.html in your browser:

# Static (works without server)
open view-stat.html

# Or with server
php -S localhost:8000
# Then navigate to http://localhost:8000/view-stat.html

The dashboard includes interactive charts:

• Message distribution by role (doughnut chart)
• Conversation/message length histograms
• Activity timeline (messages over time)
• Hourly and weekly activity patterns
• Prompt type breakdown
• Sentiment indicators comparison
• Model usage distribution

All charts are powered by Chart.js and match the dark theme of the main viewer.

---

Notes

• HAR files contain full conversation data including messages and metadata
• The script handles both plain JSON and base64-encoded responses
• Duplicate conversations (across multiple HAR files) are deduplicated by most recent update time
• Only conversation endpoints (/backend-api/conversation/{id}) are extracted
• The viewer automatically detects PHP availability and enables/disables features accordingly
• Category updates create timestamped backups to prevent data loss