Skip to main content

How we keep you safe.

Non-custodial, offline-first, and open source. Verify the threat model and the cryptographic core yourself.

Three rules. No exceptions.

Non-custodial

We never hold the key to your funds, only you do. Seedcase can't access, recover, or transmit your seed phrase on your behalf, by design.

Offline-first

No internet connection is required, ever. Fragmentation, encryption, reconstruction: every operation runs entirely on your machine.

Open source core

The cryptographic engine is open source and auditable on GitHub. Anyone can inspect how fragments are created and reconstructed.

Shamir is safe by design.

A fragment below the threshold reveals nothing about your seed phrase. That is not a probability. It is an information-theoretic guarantee: the secret is not hidden in the share. It is absent.

The same scheme protects the DNS root keys that underpin the internet. Seven trusted community members hold shares. Five must meet to reconstruct.

Read the 1979 paper(opens in a new tab)
  1. 1. course
  2. 2. graduate
  3. 3. academic
  4. 4. acid
  5. 5. detect
  6. 6. wireless
  7. 7. pregnant
  8. 8. warmth
  9. 9. aunt
  10. 10. domestic
  11. 11. educate
  12. 12. ticket
  13. 13. firm
  14. 14. mental
  15. 15. perfect
  16. 16. else
  17. 17. gravity
  18. 18. alarm
  19. 19. buyer
  20. 20. mountain
A SLIP-39 fragment as Seedcase writes it. These twenty words, alone, reconstruct nothing.

Don't trust. Verify.

The seed is never written to disk

It exists in memory only during fragmentation, then it is erased. Never cached, never logged, never stored in plaintext. View the repository

No network code in the cryptographic core

The engine that splits and reconstructs your seed phrase is a standalone library with zero dependencies: nothing in it can open a connection. The desktop app inherits that rule, and you can check it yourself. Browse the core on GitHub

SLIP-39, checked against the official test vectors

Seedcase invents nothing: its fragments are exactly those defined by the SLIP-39 standard. The Seedcase core passes the official tests and specification provided by SatoshiLabs. See the test vectors

An independent audit is planned before the V1 stable release. The full report will be published. We will announce the date once contracted.

What we protect against. And what we don't.

Seedcase protects against

  • Loss of one or several fragments

    An M-of-N scheme is designed for this. As long as M fragments survive, your seed phrase can be fully reconstructed.

  • Theft of a single fragment

    Below the threshold, an attacker learns nothing about your seed phrase.

  • Single point of failure

    No fire, no flood, no lost notebook can wipe out your access if your fragments are distributed across locations.

  • Service shutdown or company failure

    Your fragments do not depend on us: any SLIP-39 compatible tool reconstructs them.

Seedcase does not protect against

  • A compromised machine during fragmentation

    If your computer has malware reading memory during fragmentation, no software can save you. Use a clean, preferably air-gapped device for the initial fragmentation. An old laptop, wiped and never reconnected to the internet, works perfectly.

  • Coercion of M or more fragment holders

    If someone forces M people to give up their fragments simultaneously, the seed phrase can be reconstructed. You can get around this with an encryption password. Gathering the fragments is then no longer enough: reconstruction still requires the password.

You've seen the model. Request access.