Skip to content

Command Reference

Commands for AI-assisted development.

The canonical command definitions (the markdown files Cursor executes) live in commands/ at the repo root; those files are the source of truth. This documentation describes what each command does. For full instruction content, see the repository commands/ folder. Command files are plain markdown with structured sections; validate with python schemas/validate.py commands/<name>.md (see schemas/README).

Quick Reference


Product

Command What It Does
/create-task Create tasks with specified type (epic, story, bug, task, etc.)
/decompose-task Decompose large tasks into well-defined subtasks

Planning

Command What It Does
/refine-task Refine tasks to meet Definition of Ready with clear acceptance criteria
/create-plan Create technical implementation design

Development

Command What It Does
/start-task Begin implementation with branch and context
/complete-task Commit, push, create PR

Quality

Command What It Does
/create-test Generate unit tests (adapts for backend/frontend)
/review-code AI-assisted code review

Utilities

Command What It Does
/mcp-status Check MCP server authentication status
/setup-asdlc Initialize repository for ASDLC adoption


How Commands Work

Commands are markdown files that tell AI agents what to achieve. The AI reads your project structure, issue tracker, implementation plans, and team conventions, then executes contextually - same command, different projects, intelligent adaptation.

Each command aligns with specific ASDLC patterns and pillars (Factory Architecture, Standardized Parts, Quality Control); see individual command pages for details.

See the full product flow →