Civicade

Security

Security you can ask hard questions about.

Public records work involves sensitive information about residents and staff. This page describes our approach in plain language and labels every control honestly.

How to read this page

  • Available in current build implemented and tested in the current CivicRecords build. CivicRecords is preparing for its first pilots and is not yet in production use, and these controls have not been independently audited.
  • Planned on our roadmap, not available today.
  • Design objective how we intend to operate a hosted service; to be confirmed as the service is deployed.

Civicade does not hold SOC 2, FedRAMP, StateRAMP or any other certification, and no external penetration test has been performed yet.

Access

Who can see and do what.

Tenant separation

Available in current build
Each municipality's data is separated in the application. The active municipality is determined on the server from a person's verified membership—never from a web address—and automated tests check that one municipality can't read, download, search or export another's data. Database-level row security is planned as an extra layer.

Role-based permissions

Available in current build
Roles are bundles of specific permissions—such as viewing requester details, viewing unredacted originals or approving a release. People can't grant permissions they don't hold themselves, and these checks run on the server, not just in the interface.

Authentication & two-step verification

Available in current build
Strong password rules, lockout after repeated failures, and optional authenticator-app two-step verification (TOTP) for every user. Password resets and invitations use single-use, expiring links.

Single sign-on

Available in current build
Sign-in through your identity provider with OpenID Connect or SAML 2.0, with an option to require it for your staff. Tested against simulated identity providers; verified with yours during setup.

Required two-step verification policy

Planned
A setting that lets a municipality require two-step verification for every staff account is planned.

Data & documents

Handling records carefully.

Secure document handling

Available in current build
Uploads are checked by their actual content, not their file extension; risky formats are rejected; there's an integration point for malware scanning with quarantine. Documents are never publicly accessible—downloads go through permission-checked links and are logged.

Redaction that removes content

Available in current build
Redaction creates a separate copy with the underlying text and image content removed and metadata stripped, then checks the result automatically before release. Originals stay with restricted access.

Encryption

Design objective
Encrypted connections (HTTPS) for all traffic and encrypted storage for databases and files in the hosted service. Sensitive secrets such as two-step verification keys are encrypted by the application today.

Backups & resilience

Design objective
Automated database backups with point-in-time recovery and versioned file storage, with restore procedures tested before we serve production customers.

Accountability

A record you can rely on.

Audit logging

Available in current build
Sign-ins, permission changes, status changes, uploads, downloads, redactions, approvals and releases are recorded with who, what and when. The log is append-only: entries can't be edited or deleted, and passwords and secrets are never written to it.

Controlled support access

Available in current build
Civicade staff can't browse a municipality's data. Support access requires the municipality's approval, is time-limited and read-only, excludes documents, and is recorded in the audit log.

Engineering

Secure development and vulnerability management.

Secure development practices

Available in current build
Automated tests for permissions and data separation, code review, dependency auditing, and an internal security review of the product with fixes tracked to completion.

Independent testing

Planned
An external penetration test before production launch, and a path toward independent security attestation as we grow.

AI governance

Assistance that never takes the decision.

Off by default, your choice

Available in current build
AI assistance is optional and disabled unless a municipality enables it, feature by feature. Sending data to an external AI provider additionally requires the municipality to record its approval.

Human review, always

Available in current build
AI can suggest a category or highlight possible personal information. It never decides exemptions, denies requests, applies redactions or releases records. Every core workflow works with AI turned off.

Read more: AI-assisted redaction: why human review must stay in control.

Security inquiries

Questionnaires, reviews and vulnerability reports.

We're glad to complete your security questionnaire and walk your IT team through the architecture. If you believe you've found a vulnerability, please contact us privately rather than disclosing it publicly.

Bring your IT team to the demo

Walk us through how requests reach your office today. We'll show you the product, answer questions honestly—including what isn't built yet—and talk through what a pilot would involve.