AI Skill Report Card

Architecting Enterprise Security Programs

A-85·Nov 18, 2026·Source: Web
14 / 15

When asked to address a security problem, first classify it into one or more domains, then apply the matching framework:

DomainCore Framework
GRC / ComplianceRisk register → control mapping (NIST CSF / ISO 27001) → gap analysis → remediation roadmap
Threat ModelingSTRIDE or PASTA applied to data flow diagrams
Zero TrustVerify explicitly → least privilege → assume breach, mapped across identity/device/network/app/data pillars
IAMIdentity lifecycle → authentication strength → authorization model (RBAC/ABAC) → privileged access controls
Cloud SecurityShared responsibility model → CSPM findings → IaC guardrails → workload identity
Vulnerability MgmtDiscover → prioritize (CVSS + exploitability + asset criticality) → remediate → verify
Incident ResponseNIST IR lifecycle: Prepare → Detect/Analyze → Contain/Eradicate/Recover → Post-incident
Endpoint/IntuneCompliance policies → conditional access → configuration baselines → app protection

Example request: "Design Zero Trust access for remote contractors accessing financial data." Immediate output: Identity verification (phishing-resistant MFA) → device compliance (Intune-managed, encrypted) → conditional access policy (block unmanaged devices, require compliant + MFA) → least-privilege RBAC scoped to specific data classification → session controls (CAE, time-bound access) → continuous monitoring (sign-in risk, UEBA).

Recommendation▾
Add an example showing a 'bad outcome' or common failure mode to contrast with correct approach
14 / 15

Progress:

  • Clarify scope: what asset, threat, regulation, or incident is in question
  • Identify applicable domain(s) from the table above (problems often span 2-3)
  • Gather/assume context: industry, regulatory obligations, existing maturity, tech stack
  • Apply the relevant framework to produce a structured assessment
  • Prioritize recommendations by risk reduction per effort (quick wins vs. strategic initiatives)
  • Present findings as: Current State → Gaps/Risks → Recommendations → Roadmap (30/60/90 day or phased)
  • Flag residual risk and compensating controls explicitly — never claim zero risk
Recommendation▾
Include a brief template/checklist for risk scoring (e.g., explicit CVSS + exploitability + asset criticality formula) rather than just naming the inputs
18 / 20

Example 1: Input: "Our SaaS startup needs a security program for SOC 2 but has no formal controls yet." Output:

  • Current State: ad hoc access, no logging centralization, no vendor risk process
  • Gaps: no access review cadence, no incident response plan, unencrypted backups
  • Recommendations (prioritized): (1) centralize identity on SSO with MFA, (2) implement audit logging + SIEM ingestion, (3) formalize IR runbook, (4) quarterly access reviews, (5) vendor risk questionnaire process
  • Roadmap: 30 days — SSO/MFA + logging; 60 days — IR plan + access review process; 90 days — vendor risk program, first internal audit dry run
  • Residual risk: control maturity not yet validated by auditor; recommend SOC 2 Type I before Type II

Example 2: Input: "A ransomware incident was just detected on a file server." Output:

  • Immediate (Contain): isolate host from network, disable compromised accounts, preserve volatile evidence
  • Analyze: identify initial access vector (phishing, RDP, vulnerable service), scope lateral movement via EDR telemetry
  • Eradicate: remove persistence mechanisms, patch exploited vulnerability, rotate credentials
  • Recover: restore from verified clean backups, validate integrity before reconnecting
  • Post-incident: timeline writeup, root cause, control gaps (e.g., missing network segmentation, no EDR on server), update IR runbook

Example 3: Input: "Review our Intune endpoint security configuration." Output:

  • Check compliance policies cover OS version, disk encryption, firewall, jailbreak/root detection
  • Verify conditional access requires compliant device + MFA for all cloud app access
  • Validate configuration baselines (CIS/Microsoft Security Baseline) applied via settings catalog
  • Confirm app protection policies for BYOD scenarios (prevent data copy to unmanaged apps)
  • Gap check: unmanaged/personal devices with access, missing Autopilot enrollment, no conditional access for legacy auth protocols
Recommendation▾
Consider trimming the domain table slightly or linking to deeper sub-resources for very large domains like GRC vs IAM to keep scope manageable
  • Always map recommendations to a recognized framework (NIST CSF, ISO 27001, CIS Controls, MITRE ATT&CK) for defensibility and auditability.
  • Default to least privilege and explicit verification — never recommend implicit trust based on network location alone.
  • Quantify risk where possible (likelihood × impact × asset value) rather than using vague severity labels alone.
  • Treat privacy as a design constraint, not an afterthought — apply data minimization and purpose limitation when IAM/cloud designs touch personal data.
  • For any architecture recommendation, state the threat it mitigates and the residual risk it leaves.
  • Prefer phased rollouts (pilot group → broader deployment) for identity/endpoint policy changes to avoid lockouts.
  • Don't recommend a tool/product without first defining the control objective it serves.
  • Don't present Zero Trust as a single product purchase — it's an architecture spanning multiple pillars.
  • Don't skip the "assume breach" lens in IR — always include detection/containment steps, not just prevention.
  • Don't ignore operational impact (user friction, help desk load) when designing conditional access or MFA policies.
  • Don't conflate compliance (passing an audit) with actual security posture — call out the distinction explicitly when relevant.
  • Don't give incident response advice without first establishing containment before deeper investigation.
0
Grade A-AI Skill Framework
Scorecard
Criteria Breakdown
Quick Start
14/15
Workflow
14/15
Examples
18/20
Completeness
19/20
Format
15/15
Conciseness
13/15