RVAS — Azure SRE Agent

Autonomous SRE with Azure SRE Agent

Ten hands-on challenges: configure SRE Agent building blocks (skills, subagents, connectors, scheduled tasks), then trigger and observe 6 autonomous incident scenarios. Finish with a capstone where you build your own production-ready SRE agent.

Start the Lab Read the Briefing

11Challenges

6Subagents

8Skills

5Scheduled Tasks

Autonomous PaaS response

S1–S2 · Container Apps · App Insights

Proactive workflow automation

S3 · GitHub issue triage · scheduled tasks

IaaS service recovery

S4 · VMs · Syslog · az run-command

Network diagnostics & remediation

S5–S6 · VNet Flow Logs · NSG · UDR

Capstone: Build your own agent

C10 · Your use case · End-to-end demo

Challenges

Challenges 00–03 cover SRE Agent configuration primitives. Challenges 04–09 are live incident scenarios. Challenge 10 is the capstone: bring your own use case and build a production-ready agent. Coaches: press Shift+C to unlock solution guides.

Setup CH — 00

Prerequisites &
Lab Setup

Deploy the Grubify infrastructure with Terraform, configure the Azure SRE Agent data plane — subagents, skills, connectors, scheduled tasks — authorize the GitHub OAuth connector, and generate baseline traffic to seed the observability pipeline.

terraform sre-agent config GitHub OAuth VNet Flow Logs

Open challenge →

Config CH — 01

Skills &
Knowledge Docs

Write your own skill YAML and knowledge document from scratch, apply them to the SRE Agent, and verify the new capability appears in the portal.

skill YAML knowledge docs sre-agent-config

Config CH — 02

Subagents &
Incident Filters

Define a subagent YAML and pair it with an incident filter. Verify that matching alert titles are automatically routed to your new specialist agent.

SubAgent YAML IncidentFilter alert routing

Config CH — 03

Connectors, Repos
& Scheduled Tasks

Register a data connector, link a code repository, and create a scheduled task. Manually trigger the task from the portal to confirm execution.

DataConnector Repository ScheduledTask

Autonomous CH — 04

Autonomous App
Incident Response

Inject a PaaS fault into the Grubify API and watch the SRE Agent detect the 5xx spike, execute the investigation runbook, and remediate — no human prompt required.

Container Apps App Insights aca-incident-handler

Autonomous CH — 05

Code Correlation
& GitHub

Push the agent from infrastructure diagnosis into developer territory: read source code, trace the 5xx to a specific file and line, and open a GitHub pull request.

GitHub OAuth code-analyzer PR creation

Scheduled CH — 06

Proactive Workflow
Automation

Create customer GitHub issues and trigger the scheduled issue-triager subagent. Observe automatic classification, labeling, and structured bot comments.

issue-triager scheduled task GitHub

Autonomous CH — 07

IaaS Service
Recovery

Stop NGINX on the web-tier VMs and let the agent detect the Syslog alert, SSH-equivalent into the VM via run-command, and restart the service.

VM run-command AMA / Syslog iaas-vm-handler

Interactive CH — 08

Network Diagnostics
(Interactive)

Inject a UDR asymmetry fault and drive a diagnostic conversation with the network-traffic-analyst agent. Find the blackhole route using VNet Flow Logs and Traffic Analytics KQL.

VNet Flow Logs Traffic Analytics UDR

Autonomous CH — 09

Autonomous NSG
Remediation

Block PostgreSQL traffic with an NSG deny rule and observe the agent autonomously identify the deny flow in NTANetAnalytics and remove the rule — no human in the loop.

NSG NTANetAnalytics network-analyst

Capstone CH — 10

Build Your Own
Production-Ready SRE Agent

Bring a real reliability problem from your own environment. Design and implement a custom subagent — skills, knowledge docs, incident filter or scheduled trigger — then demo it end-to-end to the group and defend its production readiness.

Capstone Custom subagent Your use case

Subagents

Each subagent is a specialist with its own system prompt, tool grants, and incident routing. The network-traffic-analyst supports both interactive and autonomous modes.

aca-app-incident-handler

PaaS Incident Handler

Investigates Grubify Container Apps 5xx incidents. Queries App Insights, identifies bad revisions, deactivates or rolls back, optionally analyzes source code and opens GitHub PRs.

Write CLI GitHub OAuth App Insights

iaas-vm-incident-handler

IaaS VM Recovery Agent

Handles NGINX and other VM service failures detected via Syslog. Uses az vm run-command to restart services and verify recovery.

Write CLI Log Analytics

network-traffic-analyst

Network Observability & Remediation

Analyzes VNet Flow Logs, Traffic Analytics KQL, and NSG/UDR configs. Supports interactive investigation (S5) and autonomous remediation (S6).

Write CLI NTANetAnalytics

issue-triager

GitHub Issue Classifier

Runs on a schedule to triage [Customer Issue] GitHub issues. Classifies as Bug/Performance/Feature/Question, adds labels, and posts a structured bot comment.

GitHub OAuth Scheduled

cost-optimization-agent

Cost & Efficiency Analyst

Reviews Azure resource costs on a weekly schedule. Queries Cost Management, correlates with workload usage, and reports optimization opportunities.

Scheduled Cost Management

azure-resource-config-auditor

Compliance & Config Auditor

Audits Azure resource configuration against best practices. Checks SKU selection, redundancy settings, diagnostic settings, and security posture on demand.

Read-only

Solutions

Solution Guides

Mini-lectures, expected agent behavior, common failure modes, and debrief discussion guides. See the Coach README for the full index.

Prerequisites & Lab Setup

TF deploy walkthrough, config-sre-agent pitfalls, OAuth authorization, pre-flight checks.

Skills & Knowledge Docs

Skill YAML anatomy, knowledge doc conventions, apply targeting, portal verification steps.

Subagents & Incident Filters

SubAgent YAML anatomy, IncidentFilter routing logic, write-capability governance discussion.

Connectors, Repos & Scheduled Tasks

Connector chain, repo link auth, scheduled task cron, manual portal trigger, debrief prompts.

Autonomous App Incident Response

Incident routing architecture, expected agent behavior, debrief: autonomous vs. manual.

Code Correlation & GitHub

Inline code-analyzer architecture, OAuth prereq check, PR governance discussion.

Proactive Workflow Automation

Sample issue titles, expected triage behavior, idempotency check, production path discussion.

IaaS Service Recovery

AMA/Syslog path, run-command extension, expected remediation, write-capable agent governance.

Network Diagnostics (Interactive)

TA timing warning, expected prompts, IsFlowCapturedAtUDRHop KQL, interactive vs. autonomous debrief.

Autonomous NSG Remediation

NSG delete blast radius, governance provocation, interactive↔autonomous contrast, debrief guide.

Build Your Own SRE Agent (Capstone)

Use-case approval checklist, design review guide, production-readiness criteria, demo facilitation, and full debrief discussion guide.