Join the beta
Security

How DeskWink protects you

DeskWink gives an AI app access to a real computer, so the limits are enforced on your computer and the decisions stay with you. This page summarises the design and explains how to report a vulnerability.

The design in six points

  • You choose the folders

    Your AI sees only folders you add, on your computer. The cloud can remove access, never grant more. Sensitive files such as SSH keys, .env files and browser profiles stay blocked by default.

  • You approve every change

    Files are never changed until you approve the exact change. DeskWink saves a backup of the original on your computer first.

  • Commands are approved on your computer

    Running commands is off by default. When it is on, each command waits for you to approve it in a terminal on that computer. The web console can only reject.

  • No file contents stored

    File contents pass through DeskWink in transit but are never written to our database, logs or backups.

  • Your computer connects out

    DeskWink opens no inbound port on your computer, does not start at login, and does nothing while it is stopped. There is no queue of old requests.

  • Emergency disconnect

    Disconnect all your computers from the console in one action, run deskwink stop on the computer, or revoke a single AI app at any time.

The full detail, including the five layers every request passes through and the limits of what DeskWink can protect, is in the security model. What we store, and for how long, is in the Privacy Policy.

How we work

  • Identity is decided on the server. We never trust a user, workspace or computer ID sent by the caller.
  • Credentials are kept apart: your website sign-in, your computer’s key and your AI app’s permission cannot stand in for one another, and an AI app’s access is never passed on to your computer.
  • Our logs and database hold metadata only: no file contents, no differences between versions, no command text or output, no tokens.
  • Security rules are written as tests before the features they protect, and those tests must keep passing.
  • DeskWink runs on its own accounts, network and keys, separate from our other projects.
Beta

DeskWink is in closed beta and has not had an independent security audit. Please judge it by the design described here and by what you can verify yourself.

Report a vulnerability

If you think you have found a security problem in DeskWink, please tell us. Write to [email protected]. We read every report and will reply when we can; we cannot promise a response time.

What to include

  • What you found and where (address, command or version).
  • Steps to reproduce it, and what you expected to happen.
  • The impact as you see it.
  • How to contact you.

In scope

  • The website and services at deskwink.com, app.deskwink.com, auth.deskwink.com, mcp.deskwink.com and connect.deskwink.com.
  • The DeskWink CLI (@deskwink/cli), especially anything that lets an AI app or another program read, write or run something outside the folders and approvals you set.

Out of scope

  • Other companies’ products, including the AI apps you connect to DeskWink.
  • Denial of service, spam, and social engineering of our team or users.
  • Attacks that need physical access to your computer, or an already compromised computer or account.
  • Missing best-practice settings with no demonstrated impact.

Ground rules

  • Test only with your own account and your own computers. Do not access, change or keep other people’s data. If you reach someone else’s data by accident, stop and tell us.
  • Do not disrupt the service, and do not use automated scanning at a rate that could.
  • Give us reasonable time to fix the problem before you publish details. Tell us what you plan to publish and when.

What you can expect

  • We will use your report only to understand and fix the problem, and we will credit you if you want to be named.
  • If you follow these rules in good faith, we will not take legal action against you for your research.
  • We do not run a paid bug bounty at the moment.

The same information is available to tools at /.well-known/security.txt. DeskWink is operated by Viva Essence Ltd; see company details.