Access Your Password Vault from Another Device
2026-11-11 · 4 min read · sync, ipfs, cross-device, password-manager

To access your vault from another device, you restore the same encrypted vault onto it, and there are three ways: pull the managed encrypted sync, resolve the vault from your own IPFS node via IPNS, or import an encrypted .vkvault file you carried over by hand. All three deliver identical ciphertext. You decrypt it locally on the new device with your master password, the same way as on the first one.
This page answers the question people actually ask next, where the passwords are physically stored, and the one the search boxes suggest, how to get them on a second phone.
Where your passwords are actually stored
It helps to be concrete before talking about sync. With VaultKeepR, your vault is one encrypted file. On each device it lives in that device's local storage, in the browser profile or the app's sandbox. Nothing else exists on the device: no plaintext cache, no password file in a folder. Encryption is local, XChaCha20-Poly1305 with the key derived from your master password via Argon2id (t=3, m=64 MiB, p=4).
Every synced copy of that file, wherever it lives, is the same ciphertext. This is what zero-knowledge means here, and it is covered in detail in what zero-knowledge encryption is.
The three cross-device modes
Mode 1: managed encrypted sync
The default. You sign in on the new device, and it pulls the encrypted vault from VaultKeepR's cloud, along with a managed pointer at /api/vault-cid that tells it which version is current. Zero setup, works everywhere, and the server side only ever stores ciphertext: the decryption key is derived from your master password on your device and never travels. The free tier includes this sync.
Mode 2: your own IPFS node, resolved via IPNS
The mode for people who want the storage under their own control without operating a classic server. You run vaultkeepr-node (v0.1.0, released with GitHub releases and sha512 checksums) on a machine you own, configured through one file, vaultkeepr-node.json: rpcUrl, token, gatewayUrl. You can bind it to your LAN. The RPC token is mandatory.
Your encrypted vault is published to that node with publishToNode, old versions are unpinned with unpinFromNode, and your other devices resolve the current version through IPNS with resolveIpnsOnNode. If the node is unreachable, a public IPFS gateway serves the same content by identifier as a fallback, since content addressing means any copy will do. You host the storage and the pointer; nobody else holds the routing.
Mode 3: the encrypted file, moved by hand
The mode with no infrastructure at all. Export a .vkvault file, an encrypted archive of your vault, move it however you like: USB stick, your own cloud drive, email to yourself, it is all the same ciphertext. Import it on the other device. This mode is slow, fully manual, and completely independent: it works between two devices that have never heard of any network.
How to choose between the modes
| Setup | Works offline at switch time | You control storage | Best for | |
|---|---|---|---|---|
| Managed sync | None | No, needs the sync endpoint | No | Default, phones, simplicity |
| Own node + IPNS | One config file | Gateway fallback if node is down | Yes | Self-hosters who still want no server checklist |
| .vkvault file | None | Yes | Yes, fully | Air-gapped moves, occasional transfers |
Practical rules of thumb:
- Use managed sync unless you have a specific reason not to. It is the mode every other mode falls back to.
- Use your own node if you already have a home server and want your ciphertext on your disk. The self-hosted overview covers the trade-offs.
- Use the file export as your escape hatch and backup: it works when nothing else does, and it is also how you test that a migration copied everything.
All three can coexist. Many people run managed sync day to day and keep a .vkvault export as a cold backup.
Offline, as a fact rather than a promise
Since the vault on each device is a local encrypted file, it stays readable without any network: you can open it and read your passwords on a plane. Edits made while disconnected queue up and sync when a connection is available. That is a consequence of the storage design, not a feature mode you toggle.
The honest limit: the first restore on a new device does need one of the three channels above. If you never synced, never published, and never exported, there is nothing on the new device yet.
If you lose the master password mid-transfer
Every mode above transfers only ciphertext, so a lost device or a compromised transfer channel exposes nothing readable. The master password is the key, and losing it means recovery through Shamir 3-of-5 fragments: three of your five recovery shares, held across your devices and people you trust, reconstruct access. No email reset exists, because an email reset would mean the key exists somewhere outside your devices. Download VaultKeepR and set up the fragments when you set up the vault, not after.
Related reading: decentralized sync with CRDT and IPFS, the self-hosted route without a server, and the zero-knowledge model.