Security

Your exercise is yours alone.

Built for GRC and security teams who have to answer for what a platform can see. Every exercise, inject, response, decision and after-action report is scoped to its owning organization and enforced in the database on every request.

SOC 2 Type I: in progress · Trust Center · Roadmap

Trust summary at a glance

The questions a security review asks first, answered plainly. Where the answer is no, it says no.

Independent audit or attestation

Self-assessed

Our SOC 2 Type I examination is in progress; no report has been issued yet. No ISO/IEC 27001 certificate, no HITRUST. Control mappings on this page are a self-assessment.

Single sign-on (SAML or OIDC)

Not available

Not available today. On our public roadmap.

Directory provisioning (SCIM)

Not available

Not available today. Accounts are created by invitation from an Operator.

Multi-factor authentication

Available

Enforced for every account with an authenticator app (TOTP), checked by the database on every request.

Hosting and database

Available

Lovable for application hosting, Supabase for Postgres, authentication and storage, running on Amazon Web Services. All named in the subprocessor table below.

AI processing

Available

Google Gemini models reached through the Lovable AI gateway. No model is trained on customer exercise content.

Data residency

Available

Stored and processed in the United States. Ask us for the specific region.

Tenant separation

Self-assessed

Enforced in the database on every request, per organization. Self-assessed, not independently tested.

Certifications and independent assessment

Our current posture and how to request a security review.

Cyber Ambush holds no independent certification, audit report or attestation: no SOC 2, ISO/IEC 27001 or HITRUST. Control mappings here are a self-assessment by Sterling Readiness Group, LLC, not reviewed by an independent auditor — treat them as vendor statements, not assured evidence. We answer questionnaires (SIG Lite, CSA CAIQ, your own template) directly and say plainly where a control is planned rather than in place. Our SOC 2 Type I examination is in progress; we claim no certification until it's issued. Request a security review.

Subprocessors

Every named vendor that processes data on behalf of Cyber Ambush.

  • Lovable

    Lovable Labs Incorporated · United States

    Application hosting, build and content delivery for cyberambush.com

    Data: Application code and every request served to the application

  • Supabase

    Supabase, Inc. · United States

    Postgres database, authentication (including multi-factor authentication) and file storage

    Data: Organizations, accounts, exercises, injects, responses, decisions, after-action reports

  • Amazon Web Services

    Amazon Web Services, Inc. · United States

    Cloud infrastructure beneath the database and storage provider

    Data: The same records at rest, encrypted by the platform

  • Lovable AI gateway

    Lovable Labs Incorporated · United States

    Routes AI requests to the model provider for the planning assistant, response assessment and after-action report drafting

    Data: Scenario and inject text, planning guidance, Participant responses

  • Google (Gemini models)

    Google LLC · United States

    Large language model inference reached through the AI gateway

    Data: The exercise text listed in the row above, for the duration of the request

  • Resend

    Resend, Inc. · United States

    Transactional email delivery (invitations, reminders, after-action notices)

    Data: Recipient name and email address, exercise name, join link

Stored and processed in the United States; contact us for the specific region. We give notice before adding or replacing a subprocessor, per our data processing addendum.

How exercise data reaches the AI

What is sent to the model, what is not, and who owns the final call.

  • The planning assistant, response assessment and AAR drafting send scenario text, injects, guidance and Participant responses to Google Gemini models, reached through the Lovable AI gateway. No other model provider receives exercise content.
  • Cyber Ambush does not train any model on customer exercise content; text is sent only for the duration of the request. Terms in force are available on request.
  • The platform is designed for synthetic content — don't enter regulated personal data, live case material or production secrets.
  • AI assessment output is advisory; a human Sponsor or Operator owns every final rating.

Authentication and access

How accounts are signed in and enforced today, and what is coming next.

Today

  • Email and password sign-in with enforced multi-factor authentication (authenticator app, TOTP) for every account, checked by the database on every request. No single sign-on or directory provisioning today.
  • Passwords are never stored by the application.
  • Accounts are created by invitation from an Operator, who can revoke access at any time.
  • One role per account: Sponsor, Operator, Participant or Evaluator (read-only, for Observers too). Every read and write is enforced server-side.

On the roadmap

Not available today
  • SAML and OIDC single sign-on for enterprise and public-sector customers.
  • SCIM provisioning from the customer's directory.

Both are on our public roadmap, not available today. If your identity standard requires them before purchase, Cyber Ambush doesn't meet it yet.

Retention and deletion

How long exercise records are kept and how to request deletion.

  • Exercise records, decisions and AARs are retained for as long as your organization keeps its account, so they remain available as audit evidence. An AAR can be exported at any time.
  • On written request — during the subscription or after termination — we permanently delete an organization's exercise data and personal data within 30 days and confirm when it's done.
  • We retain data beyond that only where applicable law requires it, and only as long as required.

Reporting a security issue

Coordinated disclosure and how we handle reports.

Cyber Ambush is operated by Sterling Readiness Group, LLC. Report a vulnerability to support@sterlingllc.com or use the form if you prefer: Report a security vulnerability.

  • We acknowledge reports within two business days. We don't currently operate a paid bug bounty.
  • If a personal data breach affects data we process for your organization, we'll notify you without undue delay, per our Data Processing Addendum.
  • Machine-readable contact details are published at /.well-known/security.txt.

Account isolation

Controls that keep one organization's exercise out of another's console. Implemented and tested by us; not verified by an independent assessor.

Every account uses multi-factor authentication

Access to exercise data requires a second factor, enforced in the database as well as the interface.

IA-2(1)PR.AA-03

Every record is filtered in the database, not the screen

Access rules live on the data itself. A request for another organization's exercise, inject, response, decision or AAR returns nothing. There is no client-side filter to bypass and no shared query that could leak rows.

AC-3AC-4PR.AA-05

Organization membership decides the boundary

Each account belongs to one organization. Sponsors and Operators can only reach that organization's exercises; the server checks on every read and write.

AC-2AC-6GV.RR

Participants see only the exercise they joined

Participants read the brief, released injects and roster only for exercises they joined with a code — nothing pre-release, from other exercises, or from other Participants' submissions.

AC-6PR.AA-05

Evaluations are not exposed to the people being evaluated

Scoring is stored apart from Participant submissions and readable only by Sponsors and Operators, so an exercise can't be gamed from the inside.

AC-6(1)IR-3

Roles cannot be self-granted

Anyone can update their own name and department, but attempts to change their own role or organization are rejected by the database.

AC-5AC-6(10)GV.RR-02

Live updates inherit the same rules

Real-time traffic follows the same access rules as ordinary reads, so nothing is broadcast to an account that couldn't already request it.

SC-8AC-4

Contact form is write-only

Inquiries submitted from the public contact form can be created but not read back through the public interface, so form data cannot be enumerated.

SI-10SC-7

Control mapping — NIST SP 800-53 Rev. 5

  • AC-2 / AC-3Account management & access enforcement

    Accounts are provisioned per organization with a single role each, enforced server-side on every request.

  • AC-6Least privilege

    Participants get the narrowest view that still lets them play; only Sponsors and Operators can plan, approve, release or score.

  • AU-2 / AU-3Audit events & content

    Exercise records keep who released each inject, committed each response, approved guidance and called ENDEX, and when.

  • IA-2 / IA-5 / IA-2(1) / IA-2(2)Identification & authenticator management

    Email and password sign-in with enforced MFA (TOTP) for every account, checked by the database on every request. Passwords are never stored by the application.

  • IR-2 / IR-3 / IR-8IR training, testing and plan

    Each completed exercise is documented, scored evidence of IR training and annual testing against your plan.

  • SC-8 / SC-13Transmission confidentiality & cryptography

    All traffic is served over HTTPS; data at rest is encrypted by the managed cloud platform.

  • CA-7 / RA-5Continuous monitoring & vulnerability management

    Automated access-rule and dependency scans run against the platform; findings are triaged on a risk basis.

What auditors ask for

The evidence an exercise produces, lined up against the expectation it satisfies.

Compliance expectations & evidence

StandardExpectationEvidence produced
NIST CSF 2.0GV.RR, ID.IM, RS.MA, RS.CO outcomes are exercised and improved on a cycle.Objective-by-objective ratings and corrective actions in each after-action report.
NIST SP 800-53 Rev. 5IR-2, IR-3, IR-4, IR-6, IR-8 implemented and tested.Dated exercise record, Participant roster, inject timeline and improvement plan.
NIST SP 800-171 Rev. 33.6.1 – 3.6.3 incident handling capability, tested periodically.Exercise export retained as the periodic test artifact.
ISO/IEC 27001:2022A.5.24 – A.5.27 incident management planning, assessment and learning.After-action report plus tracked corrective actions with owners and due dates.
SOC 2 (Common Criteria)CC7.3 – CC7.5 incident evaluation, response and recovery.Exercise timeline, decision log and remediation follow-through.
CIS Controls v8Control 17.1 – 17.8 incident response process, roles and exercises.Assigned roles in the exercise plus the scored exercise record.
PCI DSS v4.0Requirement 12.10 incident response plan tested at least annually.Annual tabletop record with findings and plan updates.
HIPAA Security Rule164.308(a)(6) security incident procedures and response.Documented response walk-through and notification-clock decisions.

Our own analysis, not an auditor's opinion — these certify neither your program nor ours.

Running exercises securely — practitioner guidance

  • Use a distinct account per person — shared logins destroy the audit trail.
  • Treat join codes as short-lived: share them where Participants already trust, and issue a new one rather than reusing an old code.
  • Keep real casework out of injects — use realistic but synthetic names, systems and data.
  • Nominate a Sponsor genuinely authorized to make the scenario's decisions, or the escalation test proves nothing.
  • Export the AAR at the end of the exercise and file it with your compliance evidence while fresh.
  • Close corrective actions in your own tracker with an owner and a date.
  • Register two authenticator devices per person so a lost phone doesn't lock anyone out.

Scope & honest limits

  • SOC 2 — In progress; we claim no certification until the report is issued.
  • Not a certification — Cyber Ambush produces supporting evidence. It doesn't issue an attestation, and no platform can certify your program for you.
  • Not a legal opinion — Notification clocks and regulatory references are exercise material for practice. Confirm real obligations with counsel.
  • Exercise data only — Designed for synthetic scenario content. Don't enter live case material, regulated personal data or production secrets.
  • Demo mode is a sample — The DEMO toggle shows a fixed sample exercise, not written to your organization's records.

Want the framework detail behind the scoring? See how exercises map to NIST CSF 2.0, or ask us a security question.