What problem does it solve? Agent Skill folders often drift from structural standards, lack trigger evaluations, miss safety guardrails, or fail cross-host compatibility, making them hard to trigger correctly and risky to run. This Skill inspects, aligns, and validates existing or scaffolded skill folders while preserving their existing behavior. ## Core Features & Use Cases - Structural Alignment: Checks SKILL.md front matter, description trigger quality, scope, workflow clarity, and progressive disclosure against the Agent Skills portable minimum or stricter repository policy. - Cross-Host Validation: Validates Codex agents/openai.yaml metadata, Claude invocation policy, canonical trigger evals, and real npx installation parity using disposable environments. - Safety and Review Lanes: Applies mandatory code-review and apply-security lanes, adds guardrails for destructive actions and secrets, and verifies vendor-specific claims against official documentation. - Use Case: After scaffolding a new skill with skill-creator, run this Skill against the folder to repair its Help contract, create trigger evals, set the correct invocation policy, and confirm it installs cleanly for both Codex and Claude Code. ## Quick Start Use the align-skill skill to review and validate the skill folder at skills/my-skill and report its alignment evidence.
