skill-specification

Cross-platform skill specification uses YAML frontmatter and formal SKILL.md structure to ensure Claude, Codex, Cursor, Amp, and future agents consume and validate skills consistently. Define name, description, license, and compatibility metadata once; all platforms read the same canonical format.

Skill specification enforces lowercase hyphen-case naming, description length between 1–1024 characters, and optional metadata fields. Built-in validation hooks check naming patterns and field constraints automatically, preventing incompatibilities before deployment.

Yes. Skill specification provides evolutionary robustness patterns and cross-platform compatibility rules that satisfy all four platforms' schema requirements simultaneously. Progressive disclosure and self-validation ensure interoperability without platform-specific rewrites.

SKILL.md requires YAML frontmatter with name, description, license, and compatibility fields, plus optional versioning and metadata. The specification defines directory layout and field patterns; quick-start examples show minimal to extended configurations for gradual adoption.

YAML frontmatter provides machine-parseable, platform-agnostic metadata that agents consume consistently. It separates schema declaration from implementation, allowing Claude, Codex, Cursor, and Amp to validate and interpret skill intent without custom parsing logic.

Validation tooling catches naming, description length, and schema violations before deployment. Without compliance, agents may reject or misinterpret the skill. The specification ensures robust interoperability; non-compliant skills risk silent failures or platform-specific behavior.