# Import Cursor chat transcripts into Xyvenra Code

Xyvenra Code can import one or more exported Cursor Markdown transcripts from the **Import chats** control. The files are read in the browser and are not uploaded by the importer. Cursor documents exporting chats as Markdown; the current Xyvenra importer recognizes transcripts with `## User` and `## Assistant` sections.

1. In Cursor, export the conversations you want as Markdown.
2. In Xyvenra Code, choose **Import chats** and select the `.md` or `.markdown` transcript files. You can select several at once. `INDEX.md` is ignored.
3. Open an imported conversation from the **Saved chats** list. It stays in this browser unless you separately enable account sync.

The importer accepts up to 20 transcript files, 5 MB each and 20 MB total. It keeps up to 40 text turns per conversation and truncates an individual turn at 64,000 characters. Thinking-only details are excluded. Structured tool-call state, code-edit operations, images, attachments, model settings, and provider keys do not transfer; any command or edit text embedded in the transcript is imported only as text and is never executed. Unsupported or empty transcripts are skipped; if none can be read, Code shows an error.

This transfers readable chat text, not a complete Cursor workspace backup or a runnable agent session. Project files can be imported separately as local folder context or a source ZIP.