Custom skills for Azure SRE Agent that fill gaps in its built-in capabilities, focused on proactive governance, cost intelligence, and architecture quality.
⚠️ Community project: These skills are not officially maintained by Microsoft. They have been tested on a live Azure SRE Agent instance but may require adjustments as the platform evolves. Contributions and bug reports are welcome!
📦 Now in the marketplace: These skills have been accepted into the official Azure SRE Agent plugin marketplace.
📝 Blog post: From reactive firefighting to proactive operations: custom skills for Azure SRE Agent
Azure SRE Agent is Microsoft's AI-powered site reliability agent with 40+ MCP connectors, sandboxed execution, and memory. While it excels at reactive incident response (investigating alerts, running diagnostics), it lacks proactive operational skills out of the box.
This pack adds 8 custom skills that transform your SRE Agent into a proactive operations partner:
| # | Skill | What it does | Value |
|---|---|---|---|
| 01 | Well-Architected Review | 5-pillar WAF assessment with scoring | Architecture quality gate |
| 02 | Compliance & Governance | Policy, RBAC, tagging, locks, naming audit | Audit readiness |
| 03 | Capacity Planning | Quota utilization, growth projection, scaling prep | Prevent capacity outages |
| 04 | FinOps Intelligence | Cost optimization + chargeback combined | Reduce waste, track spend |
| 05 | Incident Postmortem | Blameless postmortem generator (5 Whys) | Accelerate learning loops |
| 06 | Defender Secure Score | Secure Score monitoring + improvement plan | Continuous security improvement |
| 07 | Digital Native Governance | Startup governance maturity check (15 checks, scored 0–100) | Enterprise readiness for startups |
| 08 | AI Foundry & OpenAI Posture | Security, reliability & cost posture for Azure OpenAI/Foundry (15 checks) | AI workload production readiness |
Each skill includes a sample output so you can preview the report format before running it.
 |
 |
| Digital Native Governance — maturity score with category breakdown | FinOps Intelligence — waste detection with prioritized savings |
 |
 |
| Defender Secure Score — quick wins with score impact | AI Foundry Posture — per-check Pass/Fail assessment |
| Persona | Use case |
|---|---|
| Solution Engineer | Run a governance/WAF review before a customer workshop |
| SRE / Platform Engineer | Weekly automated checks on security, cost, and compliance |
| CTO / Engineering Lead | Pre-audit maturity assessment ("are we production-ready?") |
| FinOps Practitioner | Monthly cost optimization + chargeback reports |
| Azure Advisor | SRE Agent Skills | |
|---|---|---|
| Output | Flat list of per-resource recommendations | Scored reports with maturity levels and priority |
| Correlation | None — each recommendation is isolated | Connects findings across domains ("fix 2.2 first, it blocks 2.5 and 3.5") |
| Remediation | Link to documentation | Ready-to-paste az CLI commands |
| Scheduling | Passive (you go check) | Runs on your schedule, surfaces issues proactively |
| Coverage | Cost, Security, Reliability, OpEx, Performance | + FinOps chargeback, Postmortem, AI posture, Digital Native governance |
| Context | Generic (same advice for everyone) | Asks about your scenario (startup? enterprise? pre-audit?) |
Advisor is a linter. These skills are closer to a staff SRE who reads everything, correlates findings, and delivers a prioritized report.
- Azure SRE Agent enabled on your subscription (docs | portal)
- At least Reader role on target subscriptions
- For cost skills: Cost Management Reader role
- Go to sre.azure.com and open Skill Builder (left sidebar under Builder).
- Click + Create skill, then paste the full contents of any
SKILL.mdfile from this repo (start with Digital Native Governance — it's the fastest to see results). - Save the skill, set a schedule (or run on-demand), and point it at your subscription. You'll get a scored report in minutes.
See Try your first skill above for step-by-step setup. Repeat for each skill you want to install.
Once installed, just ask naturally:
"Run a Well-Architected review on my production subscription"
"Are we compliant with our governance standards?"
"Can our infrastructure handle 3x traffic for Black Friday?"
"Where can we save money? And how much did each team spend?"
"Write up a postmortem for what just happened"
"What's our secure score and how do we improve it?"
"Are we production-ready? Run a governance maturity check"
"Is our OpenAI deployment secure? Check our AI Foundry posture"
The SRE Agent matches your intent to the skill's description and activates it automatically.
- P0 (must-have): Capacity Planning, Compliance & Governance — genuine gaps with no built-in equivalent
- P1 (high-value): WAF Review, FinOps Intelligence, Digital Native Governance, AI Foundry Posture — significant value beyond built-ins
- P2 (nice-to-have): Incident Postmortem, Defender Secure Score — lighter differentiation but useful
Azure SRE Agent supports max 5 concurrent skills (oldest auto-unloads). This pack has 8 skills, but they're designed for different use cases — you'll rarely trigger more than 2-3 simultaneously.
Each SKILL.md is self-contained. Common customizations:
- Accepted exceptions: Each skill supports an optional exceptions list to suppress known false positives (e.g., single-AZ by design, dev/test without backups). The agent skips excepted checks in scoring and lists them separately in the report.
- Adjust tagging standards in Compliance skill (Step 3) to match your organization
- Change cost allocation tags in FinOps Intelligence (Pre-check section)
- Set quota thresholds in Capacity Planning (default: flag at 70%)
- Customize postmortem template to match your org's incident process
- Adjust governance checks in Digital Native Governance for your organization's maturity stage
- Configure PTU thresholds in AI Foundry Posture (default: 70–85% optimal utilization)
See docs/customization.md for detailed guidance.
These skills become even more powerful when combined with SRE Agent's Scheduled Tasks feature:
| Skill | Recommended cadence | Trigger phrase |
|---|---|---|
| Compliance & Governance | Weekly | "Run governance audit" |
| FinOps Intelligence | Monthly (1st business day) | "Generate monthly cost report" |
| Capacity Planning | Bi-weekly | "Check capacity headroom" |
| Defender Secure Score | Weekly | "Check secure score and suggest improvements" |
| Digital Native Governance | Monthly | "Run startup governance maturity check" |
| AI Foundry Posture | Bi-weekly | "Check our OpenAI posture and token consumption" |
See docs/scheduled-tasks.md for setup instructions.
See docs/architecture.md for detailed design choices, safety model, and skill structure conventions.
Contributions welcome! See CONTRIBUTING.md for guidelines. A CI workflow automatically validates SKILL.md structure (front matter + required sections) on every PR.
Ideas for new skills:
- SLA/SLO monitoring dashboard
- Network topology validator
- Migration readiness assessment
- Disaster recovery drill runner
- Container Apps / AKS health check
- Data pipeline monitoring (Data Factory, Synapse)
MIT — use freely, attribute if you like.
Built by @ricmmartins | Powered by Azure SRE Agent | 📝 Blog post: Your Startup Doesn't Have an SRE Team — Now What?