C4: Workflow and Knowledge Skillification
When teams invest significant effort in prompt engineering and context setup, this knowledge often remains in individual conversations rather than being codified into reusable artifacts. This leads to repetitive prompting, inconsistent approaches across team members, expertise loss when senior developers leave, onboarding friction for new members, and no improvement loop for successful patterns.
Agent Skills solve this by providing a standardized mechanism for extracting, sharing, and iterating on AI workflow knowledge.
Principle
- Teams SHOULD extract reusable AI workflows into Agent Skills rather than keeping effective prompts and context setups in individual conversations.
Agent Skills are modular capabilities that extend AI Agent functionality, representing professional expertise that can be extracted, standardized, and shared across projects. Each Skill contains a SKILL.md file with instructions and metadata, optional supporting files (reference guides, scripts, templates), and is organized in a directory structure. Unlike slash commands which require explicit input, Skills are automatically discovered and used based on context.
Related Principles
- G1: Single Source of Truth
- G2: Version-Controlled Documentation
- C1: Context Engineering Competency
- C3: AI First