Storage. The server keeps its data in SQLite (a file in the data
directory, the default) or PostgreSQL (database, KEEPORG_DATABASE_URL or
-database). Every row, the KDBX versions included, is sealed with
AES-256-GCM under a key derived from the master key and named by an HMAC.
The master key is required to start and a different one is refused. A data
directory from 0.10 or earlier is imported on first start. Several servers
may share one PostgreSQL database. Go goes to 1.26 for the pure-Go
drivers (modernc.org/sqlite, jackc/pgx).
Hardening, after an audit of the server, both clients and the deployment:
- Single sign-on approves a device only after an explicit Approve in the
browser that started it, bound by a cookie and PKCE; identities are never
linked by username, by verified email only when allowed, never to admins.
- Sign-in attempts are limited per address and per account, passwords are
checked outside the store lock with uniform timing, one-time codes work
once, and starting two-factor needs the password.
- Merge vaults bound KDF parameters, payload size and XML shape, check
pushes before storing them, and run at most two merges at once. A version
is written in the same transaction as its metadata.
- Tokens in URLs only on the event stream, which ends on revocation; CSP,
frame-ancestors, no-store and HSTS; sign-in CSRF; 404 for databases
without access; new trusted_proxies.
- Clients open only the server's approval page, keep case-variant and
retired file names apart, never overwrite a save made during a pull,
require https except locally, send the stream token in a header, bound
their reads and escape notifications and terminal output.
- The .deb creates the master key, systemd is sandboxed further, nginx sets
modern TLS and HSTS, and CI deploys from main only.
Server 0.12.0, desktop 0.6.0, Android 0.4.0 (versionCode 4).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The web interface could create a vault and never remove one, so the only
way to undo a mistake was the CLI. There is now a Delete link on every
vault you administer, leading to a page that says what is about to go —
owner, versions stored, grants pointing at it — and asks for your
account password before it goes.
The gate is re-authentication, not a confirmation checkbox. A session
token is enough to read and write a vault; it should not be enough to
erase every version of one, because a session is what an unlocked
laptop leaves lying around. The one-time code is required too when the
account has one enrolled.
It lives in confirmDeletion, so the rule holds wherever deletion is
reachable: DELETE /api/v1/databases/{id} now takes {password, totp?}
and answers 401 without it, and keeporg db -delete prompts. That is a
breaking change to the endpoint, which is why the server goes to 0.7.0
and the desktop clients to 0.4.0.
An account that signs in only through an identity provider has no
password to type. Rather than lock those deployments out of deleting
anything, they confirm by typing the vault's exact name — the weaker
gate, and deliberately the fallback.
Deleting takes the record, its grants, every stored version and the key
a merge vault left with the server, and tells the watchers over SSE.
Copies already in people's mirror directories stay where they are: the
server stops tracking a vault, it does not reach into anybody's home
directory.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
