Skip to main content
Log InStart Free

Security

Security you can read, not just trust.

Designed with enterprise security and governance requirements in mind. Every control on this page says whether it exists today or is planned.

Visit the Trust CenterPlatform architecture

DooHorizon does not currently hold a security certification. We say so plainly, and publish our compliance roadmap in the Trust Center.

Security philosophy

Security is a property of how the platform is built, not a layer added to it.

  • Deny by default: every application page and API route requires a session, and each handler checks it again.
  • Least reach: server paths, network destinations and database statements are restricted unless an administrator opens them.
  • Collect little: an account needs a username and an email; uploaded files expire.
  • Say what is true: this site names what is built, what is planned and what has not been certified.

Identity & access

One account across the Doo products, with separate administration.

  • Passwords are stored as salted scrypt hashes, never in plain text.Today
  • Sessions are signed, httpOnly cookies; administrators use a separate cookie and role.Today
  • Project roles control who can view, edit and publish in VizzDoo and MLDoo.Today
  • Self-registration is off by default on a networked server.Today
  • Single sign-on (SAML / OIDC) and SCIM provisioning.Planned

Encryption

Secrets are sealed on the server and stay there.

  • Connection credentials are encrypted with AES-256-GCM using a key derived from the server secret.Today
  • The browser only ever holds the sealed value; another server cannot read it.Today
  • Exported flow and project files (.fdt, .vzd, .mld) are encrypted and authenticated, so their mappings and connection details cannot be read or altered outside DooHorizon.Today
  • A connection shared with a colleague carries placeholders, never the owner's secrets; the server fills them in at run time.Today
  • Production deployments are expected to terminate TLS; the site sends HSTS so browsers insist on it.Today
  • Customer-managed encryption keys.Planned

Tenant isolation

Your workspace is yours.

  • Each account has its own workspace with a storage quota; server paths from the browser are resolved and confined to it.Today
  • Files, folders and connections are shared only with licensed users you choose, as viewer or editor; access is checked on every read, so revoking it takes effect at once.Today
  • Short-lived working files from runs are deleted automatically after a retention period.Today
  • Dedicated and private deployments for organizations that need physical separation, where offered.Today

Audit logs

What happened, to what, and when.

  • Every pipeline run records rows in, rows out and what could not be read.Today
  • MLDoo keeps a hash-chained audit trail, so a changed record is evident.Today
  • A single cross-product audit view in GovernanceDoo.Planned

Infrastructure security

Controls around where code runs and what it can reach.

  • Outgoing HTTP and database connections are checked against private-network ranges.Today
  • Databases are read-only unless an administrator enables writes.Today
  • Python used for charts and model training runs in a sandbox with time and memory limits.Today
  • The desktop agent makes outbound connections only; it never listens for inbound ones.Today

Application security

The ordinary disciplines, applied consistently.

  • Security headers on every response: content-type sniffing, framing, referrer and permissions policies.Today
  • A Content-Security-Policy on the public website.Today
  • Formulas are evaluated by a purpose-built parser, never with eval.Today
  • Inputs are validated with schemas at the API boundary; post-login redirects accept same-site paths only.Today
  • Independent penetration testing.Planned

AI security

AI is held to the same rules as everything else on the platform.

  • Model serving endpoints answer only to their own deployment token.Today
  • Optional language-model features are off unless configured, and say what they send to a provider.Today
  • Agent permissions, guardrails, human approval and run traces in AIDoo.Planned

Contact the security team

Report a vulnerability or ask a security question at hello@flowdoo.tools. Please include enough detail to reproduce the issue, and give us reasonable time to respond before disclosing it.

Working through a security review? Talk to us and we will walk your team through the architecture.

Security → Architecture → Pricing → Demo.

Everything an evaluation needs is public. When you are ready, we are too.

Start FreeBook Enterprise Demo