[New Skill]: azure-service-retirement

## Skill Name

azure-service-retirement

## Contacts

mugurehiya

## Intended Agent Scenarios

This skill helps users discover actual Azure service retirement recommendations and affected resources by invoking the Azure MCP Server tool `advisor_recommendation_list`.

It supports:
- Azure service retirement and shutdown deadlines
- Deprecations with a retirement date, end of support, and end of life
- Retirements that have passed or are approaching
- Affected Azure resources and ARM resource IDs
- Service Health retirement tracking IDs
- Required upgrades or migrations that remediate an explicitly identified retirement
- Prioritization of expired and nearest retirement deadlines

The skill is exclusively for retirement scenarios. It must not route general upgrades, routine patching, modernization, unrelated SKU changes or migrations, general Azure Advisor requests, or reliability, cost, security, and performance recommendations. Although the Advisor subcategory is `ServiceUpgradeAndRetirement`, upgrade-only requests remain out of scope unless tied to a specific retirement.

## Timeline

Q4 2026

## Additional Context

Dependency: Azure MCP Server `advisor_recommendation_list` with retirement filters including `subCategory`, `trackingIds`, and `retirementDate`.

A proof-of-concept skill has been implemented locally under `plugins/azure-skills/skills/azure-service-retirement`. Build, frontmatter, reference, and per-file token validations pass. Adding the skill to the existing `azure-skills` plugin raises its formatted skill-description count to 20,373 characters against the repository's 20,000-character soft cap, so onboarding guidance is requested on plugin placement or description-budget remediation.

Review the proof of concept under plugins/azure-skills/skills/azure-service-retirement and the existing azure-skills plugin description. Confirm the build, frontmatter, reference, and per-file token validations, then investigate the 20,373-character description against the 20,000-character soft cap. Done means the skill has an agreed plugin placement or description-budget remediation while retaining the passing validations.

Written by the indexing model from the issue text.