Excaliself vs ExcaliDash

ExcaliDash and Excaliself solve the same problem: the Excalidraw editor on your own server, with accounts, a list of drawings, and live collaboration. I maintain Excaliself. I checked ExcaliDash at v0.6.7 (October 2026) in its repository and docs.

Side by side

	Excaliself	ExcaliDash
License	MIT	LGPL-3.0
Containers	App and Postgres	Backend, nginx frontend. SQLite by default, or Postgres
Sign-in	Accounts with invite links. Google Workspace, Microsoft Entra ID, OIDC	Accounts. OIDC
Shared space	One Team space for all members, plus Private	None. Share each drawing or collection with accounts
Folders	Nested	Collections, flat
Search	Titles and the text inside drawings	Titles
Live collab	Built in, with guest links	Built in, with link sharing
Share links	Always edit. No expiry	View or edit, with an expiry
Comments	Pinned to shapes	None found
Version history	No	Snapshots, kept 2 days by default
AI	MCP server. Edits wait for Apply	No built-in MCP
Excalidraw+ import	Yes, with an API key (beta)	No
Images	In Postgres	In the database, or on S3

Where ExcaliDash is ahead

• Version history. Open Version History in a drawing, preview a snapshot, and restore it. SNAPSHOT_RETENTION_DAYS sets how long snapshots stay (docs). Excaliself has no history: a backup is the only way back.
• Finer sharing. Per-account view or edit access on a drawing or a collection, and links that expire (docs). An Excaliself share link always lets people edit, until you stop or regenerate it.
• S3 for images, and scheduled SQLite backups built in.
• A larger community: about 1,800 GitHub stars.

Where Excaliself is ahead

• A Team space. Every member sees every Team drawing and folder with no share step, and Team drawings are always live (Team space). In ExcaliDash, you share each item with each account.
• SSO for whole domains. With SSO_ALLOWED_DOMAINS, people from your company join on their first Google, Microsoft, or OIDC sign-in, with no invite (Single sign-on).
• Comments pinned to shapes, with replies, resolve, and unread dots (Comments).
• MCP. Claude and other MCP clients can find, read, create, and export drawings, and answer comments. Changes to a drawing wait until someone clicks Apply (Connect an AI client).
• Search inside drawings, not only titles.
• Excalidraw+ import of scenes, collections, and images (Import).
• A shared team library saved on the server (Team library).

Running them

Both run with Docker Compose. Excaliself is one app container next to Postgres. It is also a Railway template and a Coolify compose file. Run one Excaliself replica: the live state of each open drawing is in the app’s memory. ExcaliDash’s Socket.IO server has no Redis adapter in the code I read, so I expect the same limit there, but its docs do not say.

Which one?

Pick ExcaliDash if you need version history or view-only links now. Pick Excaliself if your team wants one shared space, SSO for your domain, comments, or an AI client that can draw. To try Excaliself, run:

docker run --rm -p 127.0.0.1:3000:3000 -e PREVIEW_MODE=true ghcr.io/maciekzygmunt/excaliself

and open http://localhost:3000/preview/login. The comparison of all Excalidraw+ alternatives also covers Nextcloud Whiteboard.