How to Export AI Conversations to Obsidian (with YAML Frontmatter)

Obsidian is a folder of Markdown files. Everything else — backlinks, graphs, Dataview queries — works because the files are plain text with structured metadata at the top. That metadata block is called YAML frontmatter, and it’s what makes a note more than text:

---
title: Debugging a Python rate-limiter with ChatGPT
platform: chatgpt
model: gpt-4o
date: 2026-07-14
message_count: 12
word_count: 1860
source_url: https://chatgpt.com/c/...
tags:
  - ai/debugging
  - python
---

Notes with frontmatter can be queried with Dataview, filtered by platform or date, and grouped into dashboards. Notes without it are text you have to read to know what they contain.

Only an exporter that writes structured frontmatter gets it into your system in a way you can find again.

Open Obsidian, create a note, paste the conversation. It works — until the conversation has a code block, a table, or a formula. And there’s no metadata: no platform, no model, no date you didn’t type yourself, no link back to the original chat. For a daily PKM workflow, this scales badly.

ChatGPT, Claude, and Gemini all offer a full-account data export (ZIP, JSON, arrives by email). You can convert that to Markdown with a script or a third-party tool. It’s a fine full-account backup, but it’s batch-only, slow, and the converter decides what your metadata looks like — usually nothing.

Save the page as HTML or print to PDF, then drop the file into your vault. Now you have a note you can’t edit, query, or backlink. It’s a screenshot with extra steps.

Step 1 — Install. Search for XWX AI Chat Exporter in the Chrome Web Store and click Add to Chrome. No account, no signup.

Step 2 — Open the conversation you want to save. Go to ChatGPT, Claude, Gemini, DeepSeek, or any of the 13 supported platforms and open the chat. The extension auto-detects the platform.

Step 3 — Choose Markdown. Click the floating button and select Markdown. The file downloads with a .md extension and YAML frontmatter already written in — title, platform, model, timestamp, message count, word count, and the source URL. Nothing to fill in yourself.

Step 4 — Move the file into your vault. Drag the downloaded .md file into your Obsidian vault folder, or set your browser downloads to save directly into it. The note appears in your graph, links work, and Dataview can see it immediately.

Because every export has structured frontmatter, you can build a view of your AI notes. A simple example:

TABLE platform, model, message_count, source_url
FROM "AI Chats"
WHERE date >= date(2026-01-01)
SORT date DESC

That gives you a live, sorted list of every AI conversation you’ve exported since the start of the year — by platform, model, and length — without ever opening a single note. See Obsidian Dataview queries for AI chat archives for more.

Every Markdown export writes these fields automatically: title, platform, model, date / timestamps, message_count, word_count, and source_url — a link back to the original chat.

Two optional fields are available with Pro + your own API key: AI tags and summary, auto-generated by an LLM you configure (OpenRouter, OpenAI, Gemini, Qwen, Kimi, DeepSeek, and others). With those enabled, the frontmatter grows a tags: list and a summary block — which makes Dataview filtering by topic possible, not just by platform and date. See AI Tags & Summaries.

• Markdown export is 100% local. Unlike PDF, the Markdown file is generated in your browser — nothing is sent to a server.
• Obsidian renders math only if you enable it. The exported Markdown keeps $...$ delimiters; enable "Math" in Obsidian’s settings and equations render inline.
• Images need a downloadable source; if a host blocks it, the export uses a URL or placeholder.
• AI Tags & Summaries need Pro and your own API key. It’s not free out of the box, and the LLM you configure sees whatever content you choose to send to it. If you don’t want that, skip the feature — the base frontmatter still gives you everything above.

• PKM / second-brain users — capturing research and thinking in a place that’s queryable, not just a tab that vanishes.
• Students — archiving explanations and solved problems with the model and date attached, so a term’s worth of AI help becomes searchable.
• Researchers — keeping source-backed AI discussions with the source URL in frontmatter for later verification and citation.
• Developers — saving debugging conversations with code intact, then finding them by platform or date.
• Writers — collecting brainstorming and editing threads as editable notes they can reuse in drafts.

• PDF — when you need something shareable and printable. See Export ChatGPT to PDF.
• Word (.docx) — when you need to edit or fold the conversation into a report. See Export AI Chats to Word.
• Notion — pushes the conversation to your Notion workspace with database properties instead of frontmatter. See Sync AI Chats to Notion.
• JSON / TXT — structured and plain-text records for archival or scripts.