docs-skills

Required frontmatter fields for Claude Code skill authoring include name, description, allowed-tools, context, and disable-model-invocation. These standardize skill configuration and ensure uniform behavior across different projects and repository deployments.

Subagent execution and dynamic context injection are configured by defining specific frontmatter keys in your skill definition. This documentation specifies how to structure these fields to enable runtime context injection and manage how skills execute within the subagent environment.

The best way to distribute Claude Code skills as plugins is by following the official packaging and sharing guidelines. This ensures consistent frontmatter, correct execution patterns, and standardized deployment processes across multiple projects for repository maintainers.

Claude Code skill configuration works without external dependencies. The documentation defines required frontmatter keys and execution models internally, allowing skill authors to standardize creation and distribution natively without needing additional tooling or package managers.

Standard frontmatter is needed for Claude Code skill creation because it solves the lack of consistent standards across configuration and distribution. Defining required fields like name, description, and context ensures uniform skill behavior and reliable execution.

The disable-model-invocation frontmatter key limits specific model execution behaviors within Claude Code skills. Skill authors use this field to explicitly control subagent execution patterns and prevent unauthorized context injection during runtime processing.