Azure Reliability Advisor

The skill clearly limits scope to Azure Functions and App Service, requires resource-group, subscription, or app-name filtering, and adds user confirmation before CLI changes, IaC deployment, code edits, storage migration, and multi-region rollout. It discloses permissions, costs, downtime, migration duration, and some data-migration risks. Points are deducted because it directs the skill to execute deployments and migrations itself, rollback coverage is incomplete, several examples could affect production resources directly, and sensitive configuration and credential data flows are not sufficiently described.

The documentation provides workflows, error messages, unsupported-service handling, and verification commands, making the happy path plausible. Points are deducted because nothing was executed and key commands are therefore unverified; the parent workflow says to change compute before storage migration while configure-zone-redundancy imposes a conflicting gate requiring migration before compute changes; command, property, and example consistency is not demonstrated.

Trigger phrases, audience, supported services, prerequisites, and the non-fit boundary for Container Apps are reasonably explicit, with CLI, Bicep, Terraform, and multi-region paths. Points are deducted because Chinese-language interaction, mainland-China reachability, and China-region Azure compatibility are not addressed; evidence for variation across resource types, SKUs, regions, and real project layouts is limited.

The skill has front matter, an MIT license, author and version metadata, quick reference material, progressive reference files, boundaries, cost notices, error handling, and navigable organization. Points are deducted because there is no changelog, explicit maintainer responsibility, or clear update path; repository statistics in the README do not fully align with the object description, and several references and planned features indicate an evolving documentation surface with incomplete dependency and MCP availability notes.

By design, it can assess reliability, produce a feature-pivoted checklist, and, after confirmation, generate or apply CLI/IaC remediation followed by reassessment. Points are deducted because static review cannot validate Azure queries, CLI commands, Bicep, Terraform, or Front Door configurations; multi-region examples depend on project-specific modules and variables not supplied here, while storage, Event Hubs, code, and configuration synchronization still require substantial manual work.

The files include concrete queries, commands, code samples, verification steps, version metadata, and repository-level CI/test indicators, providing some auditability. Points are deducted because this review is static and contains no skill-specific executed test results, coverage evidence, or independently reproducible third-party execution evidence. Repository badges cannot establish that this skill's key paths were tested, so the score remains within the static-review ceiling.