Skills Reference

Skills are reusable knowledge packages activated semantically based on their description. The agent runtime reads each skill’s description field and decides at every turn whether the skill is relevant to the current task. A well-written description with clear trigger keywords and explicit use/don’t-use guidance is the primary mechanism for controlling when a skill loads.

Tip

Skills are never invoked explicitly. They load on-demand when the agent’s context matches the skill description. Unlike instructions (deterministic, path-based) or agents (explicit invocation), skills activate automatically.

Skill Catalog

Skill	Purpose	Used By
adr-workflow	ADR creation and management	plan, specify
board-config	Platform detection (GitHub/ADO)	devsquad, kickoff, decompose, implement
complexity-analysis	Story complexity estimation	kickoff, specify, plan, decompose
diagram-design	Architecture diagram guidance	plan, security
documentation-style	Markdown formatting rules	All doc-producing agents
engineering-practices	DevOps/SRE/CI/CD decisions	plan
git-branch	Branch management	implement
git-commit	Conventional Commits	implement
init-config	Project config initialization	init
init-docs	Documentation template init	init
init-scaffold	Community file scaffolding	init
next-task	Next task suggestion	implement
pull-request	PR finalization and review	implement
quality-gate	Artifact quality validation	specify, plan, decompose, review
reasoning	Decision recording and handoff	All agents
security-review	Security assessment workflow	plan, implement, review
work-item-creation	Work item creation rules	kickoff, decompose, implement, refine
work-item-workflow	Implementation workflow	decompose, implement

By Domain

Architecture and Planning ADR management, diagram design, engineering practices, and decision recording.

Work Items and Estimation Board configuration, complexity analysis, work item creation, and workflow management.

Quality and Security Artifact validation, security assessment, and documentation formatting.

Development Branch management, commit standards, PR finalization, and task progression.

Initialization Project configuration, documentation templates, and community file scaffolding.

What to Read Next

• Choosing the Right Extension for when to use skills vs instructions, hooks, or agents
• Extension Recipes for creating custom skills