Security

# Built to pass your security review

OpenSteward runs inside your own infrastructure, so your security team reviews the code and the deployment rather than a vendor's promises. This page is the short version; the [security guide](https://docs.opensteward.dev/security/) has the threat model and the hardening checklist.

### Your boundary

Customer data and OAuth tokens move between your application and the upstream platform only. There is no OpenSteward cloud in the data path, and no telemetry.

### Verified offline

The license file is signed and checked on the install itself. Nothing phones home, so air-gapped installs work, and an expired license never stops production: features step down after a 30-day grace period and nothing is deleted.

### Auditable source

Every line is on GitHub. Images are multi-arch, digest-pinned, cosign-signed and SBOM-attested, and the build gates on a vulnerability scan.

## Hardened by default

- Non-root containers with read-only root filesystems, dropped capabilities and seccomp, and a default-deny NetworkPolicy in the Helm chart. Published ports bind to `127.0.0.1`.
- Production images refuse to start on a placeholder secret and name the variable at fault. First-run admin creation can require a bootstrap token, so a fresh deployment cannot be claimed by whoever reaches it first.
- Outbound requests are checked against SSRF: callback URLs must be HTTPS and may not resolve to loopback, private or link-local addresses. OAuth redirect URIs are allow-listed.
- Inbound webhooks are verified against a per-platform signature registry that fails closed. Secrets are compared in constant time, and credentials are redacted from logs.
- Personal data is masked by default in the admin console; reveals and exports are audited, and every change lands in a filterable, exportable audit log.
- SSO with OIDC, SAML or LDAP, SCIM provisioning, custom roles and per-user tenant scope.

## Report a vulnerability

Please report security problems through [GitHub Security Advisories](https://github.com/open-steward/opensteward/security/advisories/new) on the OpenSteward repository, not in public issues. We aim to acknowledge reports within 5 business days and follow a coordinated 90-day disclosure window. If GitHub is not an option, write to <support@opensteward.dev>. The same details are in [security.txt](https://opensteward.dev/.well-known/security.txt).

## This website

opensteward.dev and docs.opensteward.dev are static pages. They set no cookies and load no analytics, fonts or scripts from anyone else. A strict Content-Security-Policy allows only this site's own files, plus platform logos from Brandfetch on the docs' Platforms page, and HSTS keeps every visit on HTTPS. Signing up for a license happens on [license.opensteward.dev](https://license.opensteward.dev/), under its own [terms](https://license.opensteward.dev/terms) and [privacy notice](https://license.opensteward.dev/privacy).
