agentleFS
vinayaklatthevinayaklatthe/microsoft-security-skills/skills/defender-for-cloud-hardening/SKILL.md

defender-for-cloud-hardening

Skill179 starsChanged 5 months agoMITView on GitHub

Guidance for Microsoft Defender for Cloud — cloud security posture management (CSPM) and cloud workload protection (CWPP) across Azure, AWS, and GCP. Covers Foundational vs Defender CSPM, Secure Score, Defender plans (Servers, Storage, Containers, Databases, App Service, Key Vault, APIs, AI), agentless scanning, attack path analysis, and remediation prioritisation. WHEN: Defender for Cloud, CSPM, cloud workload protection, Secure Score, harden cloud posture, Defender plans, attack path analysis, multicloud security, remediate recommendations, cloud security hardening, improve Azure Secure Score, harden Azure subscription, protect VMs and containers, multicloud CSPM across AWS and GCP, cloud security graph. DO NOT USE for SaaS posture (use cloud-app-security-posture), Sentinel SIEM (use sentinel), or extending Azure to on-prem servers (use azure-arc).

Files
SKILL.md

Microsoft Defender for Cloud (Hardening)

Microsoft Defender for Cloud is a cloud-native application protection platform (CNAPP) combining CSPM (posture, Secure Score, attack paths) and CWPP (threat protection for servers, containers, databases, storage, APIs, AI) across Azure, AWS, and GCP.

When to use

Assessing and improving cloud security posture, and enabling workload threat protection across a multicloud estate. Use this skill to choose which Defender plans to enable, how to prioritise remediation, and how to operationalise findings.

Do not use this skill for SaaS posture (cloud-app-security-posture), Sentinel SIEM (sentinel), or extending Azure management to on-prem (azure-arc).

Pick the Defender plan by workload

WorkloadPlanEnable when
Subscriptions / accounts (posture)Foundational CSPM (free)Always; everywhere
Multicloud + attack paths + agentless scanningDefender CSPMTier 1+ environments
Windows / Linux servers (EDR, file integrity, JIT)Defender for Servers Plan 2Production servers
Storage accounts (malware, sensitive-data scanning)Defender for StoragePublic-facing / sensitive data
AKS / EKS / GKE clustersDefender for ContainersAny production cluster
SQL, Cosmos DB, open-source DBsDefender for DatabasesInternet-exposed or sensitive
App Service appsDefender for App ServiceInternet-facing apps
Key VaultDefender for Key VaultTier 0 / regulated
APIs behind APIMDefender for APIsPublic APIs
Azure AI / OpenAI workloadsDefender for AI ServicesAI workloads handling org data

Rule of thumb: Foundational CSPM everywhere on day one (it's free). Then enable Defender for Servers Plan 2 + Defender CSPM in production - those two unlock the most value (attack paths, EDR integration). Other plans by risk and budget.

Approach

  1. Connect all clouds — Onboard Azure (auto), AWS (connector + CloudFormation), GCP (connector + service account). Without all clouds connected, attack paths can't trace across them. Verify: AWS / GCP connectors show "Connected"; recommendations appearing within 24 hours.
  2. Foundational CSPM baseline — Get Secure Score, recommendations mapped to Microsoft Cloud Security Benchmark (MCSB), and asset inventory. Use as the universal baseline. Verify: Secure Score baseline captured for each subscription; recommendation count by severity logged.
  3. Enable Defender plans by workload risk — Use the picker table. Don't enable every plan on day one - cost spikes and alert fatigue. Production-first.
  4. Tag critical assets so the cloud security graph weights them appropriately — criticality=high, environment=prod, data-classification=sensitive. Attack path analysis uses these to rank. Verify: critical assets tagged consistently; attack paths page shows them weighted.
  5. Remediate by attack path, not by raw count — Use attack path analysis to find chains from internet → workload → sensitive asset. Fix the chain breakers first. Secure Score % is a vanity metric without attack-path context. Verify: top 5 attack paths reviewed weekly; chain-breaker fixes tracked separately from generic recommendation count.
  6. Enforce with Azure Policy — Drive recommendations via Azure Policy DeployIfNotExists or Audit. Assign workload-owner accountability with governance rules + SLA.
  7. Operate — Stream alerts and recommendations to Sentinel for SOC correlation. Use workflow automation (Logic Apps) for auto-response (e.g. open ticket, notify owner).

Guardrails

  • Start with foundational CSPM everywhere; enable paid plans where workload risk justifies cost. Don't blanket-enable every Defender plan on every subscription.
  • Use attack path analysis to focus effort - Secure Score alone can mislead prioritisation. A 95% Secure Score with one internet-exposed unpatched VM is worse than 80% with no live attack path.
  • Tag critical assets so the cloud security graph weights them appropriately. Untagged critical asset = invisible to attack-path prioritisation.
  • Don't enable Defender for Servers Plan 2 without Defender for Endpoint license / plan awareness. Cost and EDR-policy coordination matter.
  • Recommendations without owners die. Assign via governance rules with a named owner and SLA.
  • Sentinel + Defender for Cloud is the operational pair. Posture finds, SIEM correlates with telemetry.

Common anti-patterns

  • "Maximise Secure Score by exempting recommendations" - Score goes up; risk stays. Fix, don't exempt.
  • "Enable every Defender plan everywhere" - Cost explodes; alert fatigue kills SOC. Production-first, by risk.
  • "Remediate top 100 recommendations in order" - Misses attack-path priority. Sort by attack path first.
  • "Don't connect AWS / GCP - we'll get to it later" - Cross-cloud attack paths invisible. Connect day one even if you don't enable paid plans there yet.
  • "Skip tagging critical assets" - Cloud security graph can't weight risk. Tag.
  • "Defender for Cloud is a one-time deployment" - Posture drifts daily; treat as a continuous operations programme.

Example prompts

  • Improve my Azure Secure Score in Defender for Cloud.
  • Which Defender plans should I enable to protect VMs and containers?
  • Use attack path analysis to prioritise remediation.
  • Extend CSPM across AWS and GCP with Defender for Cloud.
  • Tag critical assets and configure the cloud security graph for risk weighting.
  • Stream Defender for Cloud alerts into Sentinel with workflow automation.

Microsoft Learn

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

Reviews can't be read right now.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.