skill-creator

To create a modular AI skill for Kimi, you define metadata in YAML frontmatter, write a Markdown body, and populate optional bundled resources like scripts or references. You start by creating a directory named after your skill and adding a SKILL.md file with proper frontmatter.

The best way to package a Kimi skill is to use a standardized workflow that bundles your SKILL.md and optional resources into a distributable .skill file. This process ensures consistent quality, documentation, and packaging practices across teams.

Validating frontmatter and naming conventions for an existing skill involves applying a standardized workflow for iteration and discovery. The skill creator framework provides validation steps to ensure your updates maintain consistent quality and documentation.

The structure of a Kimi skill includes YAML frontmatter for metadata, a Markdown body for instructions, and optional bundled resources such as scripts, references, and assets. This modular design keeps the SKILL.md lean while supporting complex functionality.

You do not need bundled resources to build a Kimi skill. Resources like scripts, references, and assets are optional. You only need a SKILL.md file with proper frontmatter and a Markdown body to define the core skill logic.

The skill creator framework supports iterative improvement by encouraging feedback loops during the discovery, iteration, and packaging workflow. This approach keeps your SKILL.md lean while allowing continuous updates to existing skills.