ChatGPT2026-09-16·9 min read

Open-Source ChatGPT Export Tools in 2026: What They Really Do (and When to Use Something Else)

Search for a way out of your ChatGPT history and you’ll hit a wall of free GitHub projects before you ever reach a paid product: a declining userscript with 23,000 installs, a modern open-source extension that bulk-exports your entire account, a journalist’s script born because Business accounts can’t export at all, and console gists that dump everything to a ZIP. They are real, they are free, and several are genuinely good — this guide covers each one with credits where due. What this guide also covers, because you deserve the full picture: the specific places where open-source tools quietly stop — readable PDF and Word documents, anything outside ChatGPT, backups that run when you’re not watching — and what it costs you to DIY those yourself. All project details checked against their public repositories and pages on September 16, 2026; open-source projects change fast, so re-verify before relying on this page.

The short answer

  • You want raw data out of ChatGPT, free, and you’re comfortable installing things: open-source tools are legitimately enough. The modern ones (chatgpt-helper, scrapemychats) export every conversation including attachments, filter by project or date, and keep processing entirely in your browser — an auditable privacy story many paid tools can’t match.
  • You need the export to be a document — a PDF for a client, a .docx for a reviewer, a file that renders code blocks, tables and charts — or you need the same from Claude, Gemini, Perplexity and nine other platforms, or you want backups that run on a schedule and stay searchable: that’s product territory, and no open-source project in the niche currently does it.
  • The smart move for a lot of people is a hybrid, and it’s free to try: an OSS bulk dump as your raw-data safety net, plus a daily-driver exporter for documents and automation. We’ll show exactly how to combine them below.

The open-source landscape, one by one

Four projects cover nearly all the search demand for “chatgpt exporter github” and “open source chatgpt export”:

1. chatgpt-exporter (pionxzh) — the original userscript. One-click export of a ChatGPT conversation to Markdown/YAML, thousands of stars, ~23,000 GreasyFork installs. Honest caveat: maintenance has slowed markedly (install velocity is a fraction of its peak), and it works on the rendered page, which struggles with very long threads. Historic, influential, use it knowing its age.

2. chatgpt-helper (D1DX) — the current-generation OSS extension. Authenticates with your live ChatGPT session and calls the same web APIs the interface uses: bulk export of all conversations in six formats (JSON, JSONL, Markdown, HTML, CSV, TXT), filters by project, date range, keyword and archived/active status, and attachment downloads packaged in ~100 MB ZIP volumes with an attachment index. All client-side, no telemetry. As a raw archive tool it is excellent.

3. scrapemychats — written by a freelance journalist specifically because OpenAI doesn’t offer the standard data export for Business and Enterprise workspaces (coverage in tech press, July 2026). Archives your accessible chats plus file attachments into a local, offline-browsable HTML archive. Single-purpose, technical, and it solves a real gap.

4. Console gists (e.g. Get-Safer’s script) — paste-into-devtools or run-with-curl utilities that fetch every conversation plus uploaded/generated files into JSON + Markdown + HTML + ZIP. Maximum transparency, zero interface: the DIY option for people who read code first.

Where the open-source tools stop (the honest list)

None of these criticisms are moral failures — they’re scope. But scope is what you’re buying around:

NeedOSS toolsTypical paid exporter (2026)
Bulk raw export (JSON/MD/CSV)✅ Best in classPartially (some add bulk, none as flexible)
Attachments & generated files✅ (helper: ZIP volumes; scrapemychats: yes)Varies
Readable, share-ready PDF✅ server-rendered docs
Editable Word (.docx) with tables/code/math intact✅ (XWX generates locally; AI Exporter caps free tier)
Non-ChatGPT platforms (Claude, Gemini, DeepSeek…)✅ up to 13–15 platforms (XWX, AI Exporter)
Scheduled / incremental backup that runs itself❌ (you are the cron job)✅ only via desktop app (XWX Backup)
Search across archived chatsPartially (scrapemychats’ HTML viewer)✅ full-text vault with highlights (XWX)
Obsidian/Notion sync, YAML with AI tagsMD files you manage yourself✅ direct sync + frontmatter (XWX)
Someone answers when OpenAI ships a UI changeCommunity issues, eventuallySupport email, usually days

The structural point: OSS tools excel at getting data out once, raw. Everything after that — formatting, multi-platform coverage, automation, searching later, maintenance when platforms change — is manual labor you own, forever. That labor is cheap if you enjoy it and expensive if you don’t.

How to decide in two minutes

Your situationBest-fit route
“I want my data in git as JSONL, I’ll script the rest”chatgpt-helper — free, six formats, filters
“My company workspace has no export button”scrapemychats for a personal archive now; a desktop backup app for the recurring version (see our Business & Enterprise guide)
“One beautiful PDF, today, to email someone”Any paid-tier exporter; free tier of XWX works (daily quota) or the OSS HTML view for raw
“200 chats across ChatGPT + Claude + Gemini, always backed up, searchable”Backup-engine shape: extension + desktop scheduler (how it works)
“Free and I have no needs yet”Start with XWX’s free plan for unlimited Markdown/TXT/JSON exports and daily PDF/Word/Notion quotas — it coexists with every OSS tool above

The hybrid setup (and yes, we’re serious about it)

Because raw dumps and readable documents are different jobs, the strongest free-plus setup uses both sides of this market:

  1. Quarterly raw snapshot — OSS. Run chatgpt-helper’s bulk export (or a console gist) into raw/chats.jsonl; commit or zip it anywhere. This is your cryptographic-grade “the provider can never hold my history hostage” layer, independent of any vendor including us.
  2. Daily working layer — a live exporter. The conversations you actually reuse become Markdown with YAML front matter in your notes, one-click PDF/Word for delivery, and scheduled incremental backups so the gap between “raw snapshot” and “today” never grows.
  3. Verification, both directions. Search your archive for a phrase you know exists; open the OSS JSON for the same chat. If both agree, your backup story is real.

If step 1 already covers 100% of your needs — pure archive, no documents, no automation, no other platforms — you can stop there and pay nothing. That is genuinely the right call for some people, and saying so is cheaper than pretending otherwise.

note: Open-source means you can read exactly what a tool does with your data before it does it — for chat logs many people consider sensitive, that auditability is itself a feature. Paid tools should be held to the same standard: XWX’s Markdown, TXT, JSON and Word exports also run entirely in your browser; only PDF rendering touches a server, briefly.

Häufige Fragen

Is there an open-source ChatGPT exporter?+

Yes, several: pionxzh/chatgpt-exporter (the original userscript), D1DX/chatgpt-helper (bulk export, six formats, filters, attachments), scrapemychats (offline HTML archive, useful where Business workspaces have no export), and various console gists. All are free and process data locally; none produce polished PDFs, cover other AI platforms, or run scheduled backups.

Is chatgpt-helper safe to use with my ChatGPT account?+

It runs in your own logged-in browser session, calls ChatGPT’s web endpoints client-side, and ships no telemetry — you can read the source. The real caveat is any unofficial tool: a platform interface change can break it until a maintainer fixes it, which is the maintenance cost you accept when choosing OSS.

Can open-source tools export attachments and generated files?+

The best ones do. chatgpt-helper downloads files referenced by your export into ~100 MB ZIP volumes with an index; scrapemychats archives accessible attachments into its local HTML archive. This raw-file coverage is exactly where OSS tools match or beat paid products.

Why would I pay when GitHub has free exporters?+

You’re not paying for extraction — you’re paying for what happens after: readable PDF/Word output, thirteen platforms, backups that keep running on a schedule, searchable local archives, and fixes shipped when OpenAI changes something. If you don’t need those, free genuinely suffices; the hybrid above pairs OSS dumps with a light paid layer.

What’s the best free ChatGPT export in 2026?+

For raw data: chatgpt-helper, no contest. For everyday readable export: XWX’s free plan — unlimited Markdown, TXT, JSON and clipboard, daily quotas for PDF, Word and Notion, with no account required for the local formats.

Verwandte Artikel