Public snapshot 9 real matchesPublication status 8 scored research notesIndex-grade rows 0
methodology

Scenario Families

Scenario families keep Purple Games from measuring one narrow range. Each family composes reusable infrastructure, telemetry, background activity, vulnerabilities, and objectives.

methodology in progress

source boundary

Reflects the environment archetypes draft. Archetypes are research design targets, not all public benchmark coverage today.

Archetypes A-G

Archetypes are reusable range shapes ordered by complexity and research cost. Scenarios add vulnerabilities and objectives on top of an archetype.

  • A, Cloud-native startup: small GCP-style app, database, worker, bucket, cloud audit logs, and light host telemetry.
  • B, SaaS B2B: larger service graph with API tiers, workers, queues, warehouse access, cache, and artifact storage.
  • C, Corporate office: AD-like identity, workstations, file server, DMZ host, jumpbox, SIEM, and EDR-style telemetry.
  • D, Hybrid corporate: cloud plus on-prem style networks with identity bridging and visibility gaps.
  • E, Container-native: Kubernetes and service-mesh surfaces with container, audit, and workload-identity telemetry.
  • F, SOC-heavy: a telemetry substrate that can be applied over A-D for SIEM-grade defensive evaluation.
  • G, OT/ICS: deferred beyond v1 because realistic physical-process modeling is a separate research problem.

Composition rules

Scenario manifests select an archetype, version, overrides, background traffic profile, documented vulnerabilities, objectives, and safety controls.

  • Scenarios can adjust host counts within defined bounds and choose off, quiet, normal, or busy background traffic.
  • Scenarios can add hosts or vulnerability surfaces without weakening safety posture.
  • Every archetype and scenario should map to specific MITRE ATT&CK technique IDs for coverage reporting.

Defender realism

Background traffic and telemetry are part of the benchmark, not decoration. Defender agents must separate malicious activity from legitimate operational noise.

  • Traffic profiles include web activity, auth events, file-system activity, email-like signals, and normal egress.
  • Commercial EDR behavior may be approximated with open tooling when licenses make direct use impractical.
  • Broken services, latency, IAM friction, and noisy tools are recorded because real cyber work contains those conditions.

methodology pages