AI-Chat-Retro

ChatGPT backup viewer · DeepSeek history reader · Gemini export viewer · Claude export viewer · Grok export viewer — 100% offline, self-hosted, privacy-first

Review all your past AI chat conversations from their export backups — no cloud, no tracking, fully offline. Your data never leaves your machine.

Why AI Chat Retro?

If you've ever exported your ChatGPT data and struggled to actually read it, this tool is for you. AI Chat Retro is a free, open-source ChatGPT export viewer, DeepSeek chat history reader, Claude export viewer, and Grok export viewer that parses your AI chat backups and turns them into a clean, browsable, searchable interface — entirely offline.

• 100% local — no data leaves your machine, no accounts, no analytics
• Multi-service — reads exports from ChatGPT, DeepSeek, Gemini, Claude, and Grok in one place
• Self-hosted — runs on your laptop or homelab, Docker included
• Privacy-first — ideal for GDPR data export viewing and personal data portability
• Zero dependencies beyond Python and Flask — lightweight and fast

What it does

Loads every chat export under exports/ (ChatGPT, DeepSeek, Gemini, Claude, Grok) and shows them in a familiar chat UI: all your conversations in a sidebar, click any of them to scroll and re-read the full history, search across everything, and review chats from multiple services side by side. It behaves like the online chat apps — but uses only your local data and never touches the internet.

UI

The main window shows your chats in a searchable sidebar on the left (grouped by service tabs: All / ChatGPT / DeepSeek / Gemini / Claude / Grok), with the selected conversation rendered as user/assistant chat bubbles in the main pane. The top bar holds the brand, live search box, and the Stats button.

Prep

1. Download your AI chats as backup.zip from each service.
2. Save each zip into a folder named after the service under exports/, e.g.:
   • exports/chatgpt/
   • exports/deepseek/
   • exports/gemini/
   • exports/grok/ — the xAI zip unpacks to export_data/<user-id>/ with prod-grok-backend.json (all conversations + messages) plus prod-mc-asset-server/<uuid>/content (uploaded files); the layout is documented in grok-export-format.txt. Drop that export_data/ tree straight into exports/grok/ (a flat top-level prod-grok-backend.json also works).
   • exports/claude/ — unzip each conversations-<NNN>.zip under exports/claude/conversations/ (the manifest-*.json lists the one-time-use download URLs)
3. Extract each zip — inside you'll find conversations*.json (or MyActivity.json / conversations.json; Claude ships conversations/conversations-000/conversations.json, light_metadata/, and projects/). These files are your past chats; the app reads them directly (a Claude shard that is still zipped is auto-extracted on load).

A reference listing of the current sample export layout is in data-export-reference-conversations.txt.

How to Run

cd aichatretro
pip install flask
python3 app.py

Then open http://localhost:5000 in your browser.

• The app parses your exports/ folder every time it starts — no prebuilt data, no cache; each person's exports are loaded fresh.
• It binds to 127.0.0.1 by default, so it is reachable only from this machine.

Running inside WSL (Windows browser)

If the app lives under WSL but your browser is on Windows, localhost forwarding usually works — but if you ever see a blank/dark page (the page loads but no chats appear, or the startup-error banner shows), bind to all interfaces and reach it via the WSL IP instead:

AICR_HOST=0.0.0.0 python3 app.py
ip addr show eth0 | grep inet        # note the WSL IP, e.g. 172.23.63.0

Then open http://172.23.63.0:5000 (or http://localhost:5000 if that works) and hard-refresh with Ctrl+F5 once so the browser drops any cached old CSS/JS. The app is still local-only — it is reachable on your LAN, not on the public internet.

Docker (Docker Desktop on Windows)

The easiest way to run the app — no Python install needed on the host.

Prerequisites:

• Docker Desktop for Windows installed and running.

Steps:

1. Place your chat exports in the exports/ folder (see Prep above).
2. Open a terminal in the project folder and build + start:

   docker compose up -d --build

3. Open http://localhost:5000 in your browser (Docker Desktop forwards port 5000 to localhost).

Day-to-day commands:

Action	Command
Start (first time or after code change)	docker compose up -d --build
Restart (after adding new exports)	docker compose restart
Stop	docker compose down
View logs	docker compose logs -f
Rebuild from scratch	docker compose down && docker compose up -d --build

How it works:

• The container runs Flask with AICR_HOST=0.0.0.0 (required — without this, port mapping silently fails as a dead page).
• Your exports/ folder is mounted read-only into the container — the app never modifies your data.
• Code changes (HTML/CSS/JS) require docker compose up -d --build to rebuild the image.
• New exports only require docker compose restart (the app re-parses on every start).

Action	What happens
Browse	All chats listed in the sidebar, grouped by service tabs (All / ChatGPT / DeepSeek / Gemini / Claude / Grok)
Filter	Type in the top search box to narrow the list live
Full-text search	Press Enter in the search box for cross-chat snippet results
Review	Click any conversation to read it as chat bubbles — images, reasoning, citations, voice transcripts included
Stats	Top-right Stats button shows counts per service and per model

Use Cases

• ChatGPT backup viewer — read your exported ChatGPT conversations offline, no internet needed
• DeepSeek chat history reader — browse and search your DeepSeek export backups locally
• Gemini export viewer — view Google Gemini conversation exports in a clean chat UI
• Claude export viewer — read your Claude (Anthropic) data-export conversations in a clean chat UI
• Grok export viewer — browse and search your xAI Grok conversation exports, including uploaded files
• AI chat history search — full-text search across all your past AI conversations from one place
• GDPR data export viewer — easily read and review your personal AI data exports
• Self-hosted AI chat archive — keep a local, private archive of all your AI interactions
• Analytics — counts per service, per model, and date ranges (Stats button)
• RAG with your data — offline "Ask": ask natural-language questions over your exported chats
• Data portability — own and access your AI conversation data on your own terms

Tech stack

• Backend: Python · Flask · no third-party dependencies beyond Flask
• Frontend: vanilla HTML/CSS/JS single-page app, hand-written offline markdown renderer (no CDN)
• Parsers: one module per service (parsers/) normalizing every export format to one schema

See PLAN.md for architecture, the unified data schema, API surface, and edge cases handled. For the optional offline Ask / RAG feature (natural-language questions over your chat history, SQLite+FTS5 index), see rag/RAG-README.md.

Browser Support

Use a current version of Chrome, Edge, Firefox, or Safari (Safari 14.1+). Internet Explorer 11 is not supported.

---

Search Keywords

These are for discoverability — if you found this project by searching for any of these, it was intentional:

ChatGPT export viewer, ChatGPT backup viewer, view ChatGPT conversations offline, read ChatGPT data export, ChatGPT history browser, DeepSeek chat export viewer, DeepSeek backup reader, Gemini export viewer, Claude export viewer, Claude chat history reader, Claude data export viewer, Grok export viewer, xAI chat history reader, Grok conversations viewer, AI chat history viewer, AI conversation backup, self-hosted AI chat viewer, offline AI chat browser, ChatGPT JSON viewer, ChatGPT data portability, GDPR AI data export, local AI chat archive, ChatGPT chat reader, AI chat backup tool, no cloud AI chat viewer, privacy-first AI tool