
Run multiple kanban tasks end-to-end in Rolling Wave order — refine each task based on the prior card's actual implementation, then implement, then verify, then refine the next. Use for epic-level batch execution. --big-bang flag disables rolling wave for simple independent tasks.
Codebase exploration skill for uncertain implementation direction. Deeply explores the codebase, produces a direction report, and creates phased kanban tasks. Use when you don't know exactly how to implement something. NOT for direct implementation.
Scan kanban boards for stagnant tasks and optionally mark them. Detects tasks with no agent activity for N days (default 3), outputs a markdown report table, and appends Heartbeat entries to agent_log unless --dry-run.
프로젝트의 전체 아키텍처, 목표, 주요 결정사항을 wiki/ 디렉토리에 합성하여 정리합니다. 첫 실행 시 전체 생성, update로 변경분 반영. 매 카드가 아닌 프로젝트 수준의 지식을 정리합니다.
--- name: kanban-run description: Run the AI team pipeline for kanban tasks — orchestration loop with 6 agents (Planner, Critic, Builder, Shield, Inspector, Ranger), single-step execution, and code review. Use /kanban-run to execute tasks through the 7-column pipeline. AUTO-TRIGGER when: user says "implement task NNN" or any task ID + implement/build/do combination; or user confirms with "yes/ok/go/do it" after Claude proposes implementing a specific kanban task. license: MIT --- ## Auto-Trigge
--- name: kanban-local description: Local markdown-file kanban for toy and personal projects. No server, no PostgreSQL, no auth — all state lives in KANBAN.md in the project root. Use for solo/small projects where the remote kanban board is overkill. Auto-trigger when KANBAN.md exists and user says "태스크 추가", "add task", "칸반 보여줘", "다음 할 일", "task list", or similar task-management phrases. Sub-commands: init, list, add, move, done, show, edit, refine, run, rm, stats. --- # kanban-local 로컬 `KANBA
Manage project tasks in a local SQLite DB (~/.claude/kanban-dbs/{project}.db). Supports task CRUD (add, edit, move, remove), board viewing, session context persistence, and statistics. For pipeline orchestration use /kanban-run, for requirements refinement use /kanban-refine. Run /kanban-init first to create the local DB.
Refine backlog requirements through structured user interview. Turns rough task descriptions into concrete, actionable requirements with goal, scope, acceptance criteria, and edge cases.
Turn vague intent into a precise, executable spec through 5 structured phases, then optionally create a kanban task. Use when starting something new that needs proper scoping before refinement. Complements /kanban-refine (which refines existing tasks).
Initialize the current project in local SQLite kanban. Creates ~/.claude/kanban-dbs/{project}.db and writes local config. Usage: /kanban-init or /kanban-init my-project-name. Run with /kanban-init.
Sync kanban board tasks to EOB weekly worklogs. Reads completed/in-progress kanban cards, maps them to worklog projects, estimates hours from timestamps, aggregates by day, and submits via the EOB API.
Full project pipeline — SRS → Plan → Tasks + TDD → Rolling Wave Execute. Use when starting a new project or major feature from scratch. Default mode is Rolling Wave (implement one → verify → refine next → repeat). Use --big-bang for old all-upfront refine style.
Backend 코드 리뷰. PR의 백엔드 코드를 API 설계, DB, 보안, 동시성, 에러 처리 관점으로 분석. 사용법: /review-backend <PR_URL>
PR code review. Input URL → code analysis → post structured review comment + save MD. Auto-detects domain (backend/frontend/PLC). Usage: /review-pr <PR_URL>
Frontend 코드 리뷰. PR의 프론트엔드 코드를 컴포넌트 설계, 상태 관리, 렌더링 성능, 접근성 관점으로 분석. 사용법: /review-frontend <PR_URL>
CODESYS ST 코드 리뷰. PR의 PLC 코드를 Infrastructure 아키텍처, 네이밍 규칙, 시퀀스 데드락, 모듈화 관점으로 분석. 사용법: /review-plc <PR_URL>