7.1. Glossary

Agentless scanning — Vulnerability, secret, and malware detection performed by reading a disk snapshot rather than installing software inside the running VM. (4.2.4)

Application Security Group (ASG) — A logical grouping of VMs by application role, referenced in NSG rules instead of hardcoded IP addresses. (3.3.1)

Attack path analysis — A Defender CSPM capability that identifies compounding risk from combinations of individually low-severity findings. (5.1.1)

Automation rule — A Sentinel configuration defining the triggering conditions and orchestration order for incident response, including which playbooks to invoke. (5.2.3)

Azure Arc — A service that projects on-premises or multicloud servers into the Azure resource model, enabling Azure security tooling to reach them. (4.2.3)

Azure Bastion — A managed service providing browser-based RDP/SSH access to VMs without exposing a public IP or management ports to the internet. (4.2.2)

Azure Firewall — A managed, stateful Layer 3-7 firewall service supporting FQDN filtering and threat intelligence, typically deployed centrally. (3.3.5)

Azure Machine Configuration — A service that audits and enforces in-guest OS-level settings against a defined baseline. (4.2.5)

Azure Policy — A governance service that enforces allowed resource configurations regardless of who has RBAC permission to deploy. (2.3.1)

Blast radius analysis — A Defender XDR capability, unique to Entra Agent ID, that maps what systems could be affected if a specific AI agent's credentials were compromised. (4.1.3)

Certificate (Key Vault) — A managed object type bundling a private key with an X.509 certificate, with support for automated CA-integrated renewal. (2.2.3)

Common Event Format (CEF) — A standardized log format used by many network devices and third-party security tools to send data to Sentinel. (5.2.2)

Compliance framework evaluation — Assessment of an environment against a named external standard (ISO 27001, NIST, PCI DSS) via Defender for Cloud. (2.3.2, 5.1.1)

Conditional Access — The Microsoft Entra ID policy engine that evaluates context (risk, location, device) on every sign-in to grant, block, or require additional controls. (2.1.2)

Content hub solutions — Packaged bundles of Sentinel analytics rules, workbooks, connectors, and playbooks for specific data sources or scenarios. (5.2.1)

Custom log table — A Sentinel table for ingesting data that doesn't match a built-in schema. (5.2.3)

Data collection rule (DCR) — A Sentinel/Azure Monitor mechanism for configuring collection of Windows Security events, including via Windows Event Forwarding. (5.2.2)

Defender CSPM — Defender for Cloud's Cloud Security Posture Management capability: continuous assessment, secret scanning, and attack path analysis. (2.2.4, 5.1.1)

Defender EASM — External Attack Surface Management; discovers internet-facing assets from an attacker's outside-in perspective, including unknown/forgotten assets. (5.1.3)

Defender for AI Service — A Cloud Workload Protection plan detecting anomalous AI usage, jailbreak, and prompt-injection attempts. (4.1.5)

Defender for Containers — Detects misconfigurations and runtime risks in containerized workloads. (4.3.1)

Defender for Databases — A workload protection plan providing vulnerability assessment and advanced threat protection for Azure database services. (3.2.2)

Defender for Key Vault — Monitors vault access patterns for anomalous or malicious activity. (2.2.4)

Defender for Servers — A workload protection plan for VMs including vulnerability scanning and EDR, deployable agent-based or agentless. (4.2.4)

Defender for Storage — Threat protection for storage accounts, including malware scanning and anomalous access detection. (3.1.2)

Defender Vulnerability Management — Scans known, inventoried Azure VMs for missing patches and misconfigurations. (5.1.3)

Eligible assignment (PIM) — A PIM role assignment granting zero standing access until explicitly activated, subject to configured requirements. (2.1.1)

Encryption at host — Disk encryption applied at the host infrastructure level, covering temp disks and caches that guest-level encryption misses. (4.2.1)

Enterprise application — The tenant-local service principal instance of an application, holding granted permissions and Conditional Access scope. (2.1.4)

FQDN filtering — Azure Firewall's capability to allow or block traffic based on destination domain name rather than IP address alone. (3.3.5)

Foundry guardrails — Application-layer controls in Microsoft Foundry that filter content and mitigate prompt injection for AI agents and models. (4.1.5)

Global Secure Access / Microsoft Entra Private Access — A Zero Trust Network Access capability granting per-application, identity-aware access without full network extension. (3.3.3)

Just-in-time (JIT) VM access — Temporarily narrows an NSG rule to allow VM management access only during an approved window from an approved source. (4.2.2)

Managed identity — A credential-free identity for an Azure resource to authenticate to Entra ID, available as system-assigned or user-assigned. (2.1.6)

Microsoft Entra Agent ID — A distinct identity type for AI agents, with dedicated Conditional Access enforcement and blast radius analysis. (4.1.3)

Multi-user authorization (MUA) — A backup security feature requiring a second, independent approver for sensitive backup operations. (2.3.5)

Network Security Group (NSG) — A stateful Layer 3-4 packet filter applied to a subnet or network interface. (3.3.1)

Network Watcher (effective security rules) — A diagnostic tool showing the actual combined outcome of all layered network rules affecting an interface. (3.3.5)

OAuth permission grant / consent — The mechanism by which an application's requested permissions become actual granted access, via user or admin consent. (2.1.5)

Overprivileged access assignment — A standing RBAC or role grant broader than the identity's actual activity justifies. (2.3.4)

Passwordless authentication — Phishing-resistant authentication methods (Windows Hello for Business, FIDO2) that remove the password entirely. (2.1.3)

Playbook — A Logic Apps-based automated action invoked by a Sentinel automation rule. (5.2.3)

Private Endpoint — A network interface with a private IP inside your VNet representing a private connection to a PaaS resource, removing its public reachability. (3.3.4)

Privileged Identity Management (PIM) — A service converting standing privileged role assignments into time-bound, on-demand activations. (2.1.1)

Purview DSPM for AI — Data Security Posture Management identifying oversharing and data exposure risk specific to Copilot and AI apps. (4.1.1)

Purview Audit — Microsoft 365 audit logging (mailbox access, admin actions, eDiscovery) queryable directly from Defender XDR. (5.2.4)

Regulatory compliance dashboard — A Defender for Cloud view showing compliance posture against assigned external standards, control by control. (2.3.2)

Resource lock — A CanNotDelete or ReadOnly setting protecting a resource from accidental deletion or modification regardless of RBAC permissions. (2.3.3)

SAS token (Shared Access Signature) — A time-bound, scoped bearer credential for storage access, generated without exposing the account key. (3.1.1)

Secure boot — Verifies only signed, trusted bootloaders and OS components load during VM startup. (4.2.1)

Secure score — A single number summarizing environment hardening against Microsoft's general security benchmark, independent of any named regulatory framework. (2.3.2)

Security admin rules — Centrally-managed Azure Virtual Network Manager rules capable of overriding local NSG configuration. (3.3.2)

Security Copilot plugin — An individually-enabled extension of Security Copilot's ability to query a specific data source. (5.3.2)

Service principal — See Enterprise application. (2.1.4)

Shared key access — A storage account-wide credential granting full, unscoped access; a common security hardening target to disable. (3.1.1)

Sign-in risk — An Identity Protection signal indicating an anomalous sign-in, usable as a Conditional Access condition. (2.1.2)

Soft-delete — Retains a deleted object recoverable for a configurable window; does not alone prevent immediate purging. (2.2.1)

Transparent Data Encryption (TDE) — Encryption at rest for Azure SQL databases. (3.2.1)

Virtual Trusted Platform Module (vTPM) — A virtualized hardware root of trust supporting measured boot attestation and key storage. (4.2.1)

Vulnerability assessment (databases) — Scans Azure SQL resources for misconfigurations and missing patches, part of Defender for Databases. (3.2.2)

Web Application Firewall (WAF) — Edge-level filtering of known attack patterns (OWASP Top 10) for web apps and APIs; not a substitute for application-level auth. (4.3.3)

Windows Event Forwarding (WEF) — Centralizes Windows Security events from many machines before ingestion into Sentinel via data collection rules. (5.2.2)

Zero Trust — A security model replacing location-based trust with continuous, explicit verification of identity, device, and context on every access request. (1.1)