Azure Chaos Studio Expert Guidance

Problem

Azure Chaos Studio validates failure recovery, but failed experiments often require checking Chaos Agent status, Azure RBAC, VNet/Private Link, AKS authentication, throttling, quotas, and version compatibility at the same time. This skill organizes scattered Microsoft Learn documentation into a local index, helping engineers locate specific limits or configuration steps instead of reconstructing commands from memory.

How it works

The skill uses the SKILL.md category index as the entry point:
- Troubleshooting: L33-L40 for known Chaos Agent issues, installation health checks, and common experiment failures.
- Limits and quotas: L41-L49 for throttling, region availability, OS compatibility, and version matrix.
- Security: L50-L64 for identity/RBAC, CMK encryption, VNet, AKS authentication, IP allowlists, and Relay Bridge security.
- Configuration and integration: L65-L82 for ARM/Bicep experiment templates, fault parameters, Private Link, Azure Policy, REST/CLI, and Azure Monitor/Application Insights telemetry.
At runtime it prefers mcp_microsoftdocs:microsoft_docs_fetch and falls back to fetch_webpage, appending the from=learn-agent-skill query parameter to retrieve Markdown documentation.

Boundaries

The skill needs network access to fetch remote docs, so offline environments can only use the local index and prompts. If metadata.generated_at is older than 3 months, update the local reference first. It supports Azure Chaos Studio operations and coding tasks, but it does not replace change approval, permission governance, or chaos experiment risk controls.