The description lists no concrete actions or capabilities. It only specifies trigger phrases and vaguely mentions 'migrate infrastructure between providers, graduate from PaaS to IaC, consolidate providers, or partially move specific services' without describing what the skill actually does (e.g., generates Terraform configs, creates migration plans, etc.).

The 'when' is thoroughly covered with extensive trigger phrases and use cases. However, the 'what' is almost entirely missing — the description never explains what the skill actually does (what outputs it produces, what steps it takes, what artifacts it generates). It only describes when to invoke it, not what it accomplishes.

Excellent coverage of natural trigger terms including specific commands ('arn infra migrate'), natural phrases ('move to AWS', 'switch providers', 'graduate from PaaS'), specific platforms ('heroku', 'fly.io'), and variations ('infrastructure migration', 'cloud migration', 'provider migration'). These are terms users would naturally say.

The description is highly specific to infrastructure migration between cloud providers and PaaS graduation. The extensive list of specific trigger terms and the narrow domain (cloud infrastructure migration) make it very unlikely to conflict with other skills.

The skill is extremely verbose at ~400+ lines. It over-explains many steps that Claude could infer, includes extensive boilerplate for issue creation commands repeated across GitHub/Jira, and provides lengthy scenario descriptions that could be condensed into tables or brief lists. Much of the workflow narration (e.g., explaining what consolidation or partial migration means) is unnecessary for Claude.

The skill provides concrete gh CLI commands and structured agent invocation patterns, which is good. However, many critical steps like 'Apply the generated IaC to staging', 'Configure data replication', 'Update DNS records', and 'Shift traffic from source to target' are vague directives without executable specifics. The IaC generation and deployment steps delegate to agents without concrete fallback commands.

The workflow is well-sequenced across 10 clearly numbered steps with explicit dependencies. It includes validation checkpoints (staging verification before cutover, pre-cutover checklist), feedback loops (rollback procedure with retry/skip/abandon options), and clear state transitions (label-based status tracking). The rollback handling at Step 9 is thorough with per-service granularity.

The skill references external files like migration-scenarios.md, migration-checklist.md, experience-derivation.md, and providers.md, which is good progressive disclosure. However, no bundle files are provided to verify these exist, and the main SKILL.md itself is monolithic — much of the scenario-specific guidance (Graduate, Provider migration, Consolidation, Partial) and the detailed issue creation templates could be split into separate reference files.