Skip to content

Security: rebasepro/dadaki

SECURITY.md

Security policy

Supported versions

Dadaki is pre-1.0. Security fixes land on main and go out in the next release; older beta tags are not patched in place.

Version Supported
1.0.0-beta.x ✅ latest tag only
earlier ❌

Reporting a vulnerability

Please do not open a public issue for a security problem.

Use GitHub's private vulnerability reporting: go to the Security tab and choose Report a vulnerability. That opens a private thread with the maintainers, and you can attach a proof of concept there without publishing it.

Please include:

  • what an attacker can do, and what they need in order to do it;
  • the affected package (@dadaki/editor, @dadaki/app, @dadaki/mcp) and version or commit;
  • steps or a proof of concept — an .svg, a .dadaki file, or an MCP call transcript is usually the fastest route.

You can expect an acknowledgement within a few days. We will tell you whether the report is accepted, and let you know before any fix ships so you can check it. Credit in the release notes if you would like it; say so in the report.

Scope

This repository is the editor: it runs entirely in the browser, with no backend and no account. The hosted product at dadaki.com lives in a separate repository — a vulnerability specific to the hosted service is still welcome through the same channel.

Things that are in scope and worth looking at:

  • The SVG importer. It parses untrusted markup. Anything that turns an imported file into script execution, a network request, or a read of the host page is a real finding.
  • The file container. The reader parses untrusted binary; a malformed file should fail cleanly rather than corrupt state or exhaust memory.
  • The embedding contract. createEditor runs inside a host application — anything that lets a document reach outside the editor's own DOM or state matters.
  • The MCP server. It listens on a local port and pairs by token; anything that lets an unpaired caller drive an editor, or that leaks the token, counts.

Out of scope: findings that require the user to paste attacker-supplied code into their own devtools console, denial of service from a deliberately enormous document, and reports generated by a scanner with no demonstrated impact.

There aren't any published security advisories