Security

How we protect your account and your data.

The controls the platform applies today: how people sign in, how credentials are stored, what gets logged and for how long.

Access

Every account signs in with a password and a code from a phone.

  • Two steps on every account

    Besides the password, every sign-in asks for a code from an authenticator app (TOTP). It doesn't rely on email or SMS, and each code works only once.

  • Strong passwords, never stored

    At least 12 characters, with letters and numbers. The database keeps only an Argon2id hash, which the password can't be recovered from.

  • Lockout after wrong attempts

    Ten wrong passwords or codes lock the account for 30 minutes. Attempts are also limited per address and per account.

  • Short sessions

    A session ends after 15 minutes without use and lasts 12 hours at most. Changing the password or email, or using “Sign out of every device”, ends every open session.

  • Everyone sees what they need

    Owners, admins and members have different permissions. Data sources, TV screens and the audit trail are for admins.

  • Forgotten password, no email needed

    The organization owner creates a link to set a new password, valid for 30 minutes and once. After it, the authenticator code is still required.

Encryption

Credentials are encrypted before they reach the database.

  • Always over HTTPS

    TLS 1.2 and 1.3 only, with authenticated ciphers and forward secrecy (ECDHE). Browsers are told not to fall back to HTTP for a year (HSTS).

  • Credentials encrypted at rest

    Source passwords and tokens, webhook headers and URLs, authenticator secrets and incident notes are encrypted with AES-256-GCM before they're written.

  • Secrets don't come back to the screen

    Once saved, a credential is never sent back to the browser. The screen only shows that one is set.

  • Codes stored only as hashes

    Invitation links and TV screen codes are checked against a SHA-256 hash: the database doesn't keep the value.

  • A key outside the database, with rotation

    The encryption key isn't kept in the database. It can be changed: a procedure re-encrypts every value with the new key.

Application and network

Barriers between an attack and your data.

  • Only the server's scripts run

    Every page ships with a Content Security Policy and a fresh code on each visit. Injected scripts don't run, and other sites can't show these screens inside their own.

  • Queries within limits

    Queries to your sources never reach the cloud metadata service and, by default, not the internal network either. Credentials don't follow a redirect to another domain, and each response is read up to 1 MB.

  • Live updates for the organization only

    The live connection refuses requests from other sites. Whoever leaves the organization stops getting updates right away.

  • Each service with the minimum

    Each service signs in to the database as its own user, with only the permissions it uses. The one that runs queries can't read accounts or the audit trail.

  • Card numbers masked

    When a value read from a response carries a card number (13 to 19 digits that pass the Luhn check), only the last 4 digits are kept. A scan finds and masks any already in the database.

Audit and retention

Who did what, and when, is on record.

  • Audit trail

    Sign-ins, failed attempts, account lockouts and every configuration change are logged with author and time. Passwords, tokens and webhook URLs are left out of the log.

  • Alerts as it happens

    The organization's owner picks where security alerts go: Slack, Discord, webhook or email. Account lockouts, repeated wrong attempts, password, role or credential changes arrive the moment they happen, with no passwords or tokens in the message.

  • A record that can't be rewritten

    The application can only add to the trail, never change or delete an entry. The trail is kept for at least 12 months.

  • Retention periods

    Raw query results go after 48 hours, and sent notifications within 90 days. History follows the period the organization sets.

TV screens

The wall dashboard without leaving an account signed in.

  • Pairing by code

    Each screen pairs with an 8-character code that works once and expires in 30 minutes. An admin can switch a screen off at any time.

  • Status only

    The screen shows the status and whether an incident has been acknowledged, without saying by whom and without the notes.