Skills Overview

Understanding AI agent skills - domain knowledge + workflow automation

A skill is an executable SOP that combines domain knowledge with workflow automation. It encodes both how experts think and how work gets done.

The Dual Nature of Skills

DimensionWhat It ProvidesExamples
Domain KnowledgeExpert reasoning, best practices, decision frameworksBrand voice guidelines, analysis methodologies
Workflow AutomationExecutable SOPs that orchestrate agent toolsResearch → analyze → write → review pipelines

Tool Orchestration

Skills orchestrate the agent's native tools:

ToolWhat Skills Orchestrate
Bashgit, docker, npm, aws, kubectl, and the entire CLI ecosystem
ReadConfiguration, documents, data sources
WriteReports, code, documents, structured outputs
WebSearchResearch, fact-checking, current information
WebFetchDocumentation, APIs, external resources

Bash is the universal connector—bridging the agent to any tool with a command-line interface. This makes skills a natural replacement for workflow automation tools like n8n or Zapier.

For Every Knowledge Worker

Skills aren't just for programmers:

  • Copywriters: Brand voice, content frameworks, editorial standards
  • Data Analysts: Analysis methodologies, visualization patterns, reporting SOPs
  • Marketers: Campaign workflows, audience research, content strategies
  • Product Managers: PRD templates, prioritization frameworks
  • Engineers: Code review standards, deployment SOPs, architecture patterns

Where Skills Live

LocationPurpose
./.claude/skills/Project-specific skills
~/.claude/skills/Global skills for all projects

Skill Examples

Real-world skills in SkillUse:

  • commit — Git workflow with Conventional Commits standards
  • publish — Release workflow with validation checks
  • work — Parallel development with git worktree orchestration
  • issue — Issue management with structured templates
  • epic — Project planning with progress tracking

Next Steps