Glossary

Terms below also appear as hover tooltips throughout the challenges. Point at any underlined acronym to see its meaning in place.

Term	Meaning
Azure SRE Agent	Azure's AI-powered reliability agent. Investigates, plans and remediates across the whole Azure estate (not just AKS), with approval, or, once trusted, autonomously within guardrails.
AKS	Azure Kubernetes Service, the managed Kubernetes cluster that runs Aetherion's microservices.
APIM / front door	Azure API Management, the single public entry point that authenticates (subscription key) and routes all client traffic to the services.
HPA / autoscaler	Horizontal Pod Autoscaler, the Kubernetes control loop that adds or removes pods (replicas) as load changes.
Replica	One running copy (pod) of a service. More replicas share the load; the HPA adjusts the count.
Blast radius	How much of the system an incident, or a fix, can affect. Good remediation keeps it as small as possible.
RBAC	Role-Based Access Control, Azure's permission model. Roles (e.g. Reader) decide what an identity is allowed to do.
Managed identity	An Azure identity for a workload (here, the agent) so it can authenticate to Azure without stored secrets.
AAU	Azure Agent Unit, the consumption unit for SRE Agent usage. Managing AAUs keeps the agent cost-effective (Challenge 6).
OBO	On-behalf-of: an action running under your credentials rather than the agent's. Not available in this environment — every action here runs as the agent's own managed identity, which is why scoping that identity matters.
Review mode	Run mode where the agent proposes a plan and waits for explicit human approval before any write. The default.
Autonomous mode	Run mode where the agent executes approved classes of action without pausing. Used only for well-understood, validated faults.
Skill	A packaged, reusable runbook (a SKILL.md) the agent can draw on when a situation matches the skill's description. Name it explicitly if you want it used.
Subagent	A specialist custom agent created on the Agent Canvas, with its own instructions and tools. You use it by asking the main agent to delegate to it by name.
Knowledge base	The .md / .txt files (in knowledge/) that ground the agent in Aetherion's architecture and runbooks.
SRE Agent Administrator	The human role that approves changes to what the agent is: new skills, run-mode changes, permission grants.
Operations Center	The live GUI showing service health, operational-risk gauge, flight map, incidents and business impact.
Operational-risk gauge	The risk indicator on the Operations Center, driven by real service health, never fabricated telemetry.
Grafana / App Insights / Log Analytics	The observability stack: Grafana dashboards visualise metrics; Application Insights holds request/dependency/latency telemetry; Log Analytics stores container and platform logs.
Activity Log	Azure's record of control-plane changes, who changed what and when, used to correlate incidents with deployments.
Tier 0 / P1 / P2	Business criticality and incident priority. Tier 0 services (check-in, crew scheduling) have immediate business impact; P1 is a critical outage, P2 a major degradation.
SEV / severity	Incident severity. SEV1 critical, SEV2 major, SEV3 minor.
Fault mode	The kind of change behind an incident: a bad release, a resource limit cut too far, a canary revision serving the wrong role, a missing database index, or an API Management policy override.
Reactive → Assisted → Autonomous	The operational maturity arc the hack walks through, from human-driven response to trusted autonomy.