update
May 2, 2026
By Teun
MemHub turns ChatGPT, Claude, and Gemini history into Markdown
MemHub can import chat history from ChatGPT, Claude, and Gemini, extract memories, and export them as a Markdown ZIP for tools like Obsidian. The project positions itself as a context control panel for AI agents and stores extracted memory in an encrypted vector database.
MemHub is a tool that turns chat history from ChatGPT, Claude, and Gemini into structured Markdown that can be browsed in Obsidian-style apps or explored as a graph. According to the project’s guide, the goal is to convert AI conversations into an LLM-Wiki-style mindmap that users can treat as a second brain.
The workflow starts with exporting chat history from one of the supported platforms. The guide says users open ChatGPT, Claude, or Gemini in a browser, use the XTrace Memory extension, sign in, and click the Memory button in the top right to export the last N chats. It recommends choosing a smaller number if you are unsure how much data you want to include.
After export, users upload the data to MemHub and wait while the system processes it. The guide says MemHub searches through the data and extracts memories in real time, organizing the result as structured context rather than a plain transcript.
MemHub says it stores extracted AI memory and context in an encrypted vector database. A vector database is a system that stores data in a way that helps software find related items by meaning, not just exact text matches. The project also arranges the output in a file layout that works well with Markdown tools such as Obsidian.
Once the memories are created, users can export them as a Markdown ZIP file called “Mindmap Markdowns.” On the Memories page, the guide says to click Export and open the “Export Markdown” modal. The export dialog includes filters for scope, date range, platform, belief type, artifact type, content density, and link style.
The available scopes include Beliefs, Artifacts, and Episodes. Beliefs are described as facts, while artifacts and episodes are other content categories used by the system. Users can also choose whether the export should include full YAML and body content or a compact version, and whether links should use standard formatting or relationship-style wikilinks for items such as supersession and linked facts.
The guide recommends keeping the default settings, clicking Confirm export, and saving the ZIP file. After that, users unzip the archive into an Obsidian vault or any Markdown wiki.
To view it in Obsidian, the guide says to open the unzipped folder as a vault and then use the graph view. The graph view lets users see connections between notes rather than reading them one by one.
The repository notes that its documentation is licensed under CC BY 4.0. It also says XTrace, MemHub, and related trademarks belong to their owners, and that the repo is an unofficial community and education guide unless explicitly published by XTrace.
