Dev Workflow Skills

Skills that accelerate the daily development loop -- from brainstorming and planning through TDD, debugging, code orientation, and maintenance tasks like changelog generation and dependency management.

Overview

Dev Workflow skills are the day-to-day companions that live inside your development cycle. Unlike orchestration skills (which manage multi-step projects) or cloud skills (which handle deployment), these skills operate at the individual task level: understanding an error, choosing a data structure, writing a test plan, or summarizing a log file.

Skill Catalog

Planning Skills

brainstorming

Runs a structured ideation session. The skill uses a diverge-then-converge pattern: first generating as many ideas as possible, then filtering and ranking them by feasibility, impact, and effort.

# Start a brainstorming session
"새로운 알림 시스템을 브레인스토밍해줘"

# With constraints
"Redis 없이 실시간 알림을 구현할 방법을 브레인스토밍해줘"

writing-plans

Generates a structured implementation plan with phases, milestones, file-level tasks, and acceptance criteria. Outputs a markdown plan that can be piped into a goal.

# Generate an implementation plan
"OAuth2 로그인 기능 구현 계획 세워줘"

# Plan with time estimates
"이 리팩터링 플랜 만들어줘, 시간 추정도 포함해서"

Development Skills

tdd

Drives a full Test-Driven Development cycle. The skill follows the Red-Green-Refactor loop: write a failing test first, implement the minimum code to pass, then refactor while keeping tests green.

# Start TDD for a new feature
"TDD로 사용자 인증 모듈 개발해줘"

# TDD with a specific test framework
"Jest 기반으로 TDD 해줘, API rate limiter 만들어야 돼"

The TDD skill follows this loop automatically:

1. Red -- Write a failing test that describes the desired behavior
2. Green -- Write the minimum implementation to make the test pass
3. Refactor -- Clean up duplication, improve naming, extract helpers
4. Repeat -- Move to the next behavior until the feature is complete

git-worktrees

Manages git worktrees for parallel multi-branch development. Instead of stashing or committing WIP changes, create isolated worktrees for each branch.

# Create a worktree for a feature branch
"feature/notifications 브랜치로 워크트리 만들어줘"

# List active worktrees
"현재 워크트리 목록 보여줘"

# Clean up merged worktrees
"머지된 워크트리 정리해줘"

dispatching-parallel-agents

Fans out independent sub-tasks to parallel agents and collects results. Useful for large refactors, multi-file migrations, or running checks across multiple packages.

# Parallel lint + test + type-check
"lint, test, type-check 병렬로 돌려줘"

# Fan out file migrations
"이 10개 파일을 병렬로 새 API 형식으로 마이그레이션해줘"

Debugging Skills

debugging-helpers

Automated debugging assistance that generates hypotheses, suggests diagnostic steps, and proposes fixes based on error context and code structure.

# Debug a runtime error
"이 TypeError 디버깅 도와줘"

# Debug with stack trace
"이 스택 트레이스 분석해줘:
  TypeError: Cannot read properties of undefined (reading 'map')
    at UserList (UserList.tsx:42)"

debugging-checklist

Provides systematic checklists tailored to specific bug categories. Each checklist walks through verification steps from most likely cause to least likely.

# Get a debugging checklist
"메모리 릭 디버깅 체크리스트 줘"
"race condition 체크리스트로 점검해줘"

Navigation & Context Skills

codebase-orientation

Maps an unfamiliar codebase by analyzing directory structure, entry points, module boundaries, and dependency relationships. Produces a concise orientation document.

# Orient to a new project
"이 프로젝트 구조 파악해줘"

# Focus on a specific area
"src/api 디렉터리 구조 설명해줘"

Output includes:

• Directory tree with annotations
• Entry points and build configuration
• Key abstractions and their relationships
• Data flow diagrams (when applicable)
• Recommended reading order for onboarding

context-compression

Compresses long conversation history or large file contents into compact summaries that preserve essential context while reducing token usage.

# Compress current conversation
"컨텍스트 압축해줘"

# Summarize a large file for context
"이 파일 요약해서 컨텍스트로 만들어줘"

deep-research

Performs multi-source research with fan-out web searches, source verification, adversarial claim checking, and a final cited report. Best for complex technical questions that require cross-referencing multiple sources.

# Research a technical topic
"WebSocket vs SSE 성능 비교 리서치해줘"

# Deep dive with specific focus
"Bun vs Node.js 번들 사이즈 차이를 deep research 해줘"

Diagnostics Skills

error-message-explainer

Parses error messages, stack traces, and panic logs, then explains them in plain language with actionable fix suggestions.

# Explain an error
"이 에러 뭔지 설명해줘: ENOENT: no such file or directory"

# Explain a complex stack trace
"이 Rust panic 설명해줘:
  thread 'main' panicked at 'index out of bounds: the len is 3 but the index is 5'"

config-file-explainer

Reads a configuration file and annotates each field with its purpose, valid values, and common pitfalls.

# Explain a config file
"tsconfig.json 설명해줘"
"이 docker-compose.yml 각 필드 설명해줘"
"wrangler.toml 설정 분석해줘"

log-summarizer

Condenses verbose log output into structured summaries organized by timestamp, severity, and component. Highlights anomalies and recurring patterns.

# Summarize application logs
"이 로그 요약해줘"

# Filter and summarize
"ERROR 레벨 로그만 요약해줘, 최근 1시간"

Maintenance Skills

linter-fix-guide

Explains linter and formatter warnings in context, then auto-applies fixes across the codebase. Supports ESLint, Prettier, Ruff, Clippy, golangci-lint, and more.

# Fix all linter warnings
"린터 경고 전부 고쳐줘"

# Explain a specific rule
"no-unused-vars 규칙이 왜 중요한지 설명해줘"

# Fix with scope
"src/components 폴더만 린트 수정해줘"

data-structure-chooser

Recommends the optimal data structure based on your access patterns, size constraints, and language. Provides Big-O comparisons and implementation examples.

# Get a recommendation
"빈번한 삽입/삭제가 있고 정렬 상태를 유지해야 해. 어떤 자료구조가 좋아?"

# Compare options
"HashMap vs BTreeMap, 내 유즈케이스에 뭐가 맞아?"

dependency-install-helper

Resolves dependency installation failures including version conflicts, peer dependency issues, lockfile corruption, and platform-specific build errors.

# Fix install failure
"npm install 실패했어, 도와줘"

# Resolve version conflict
"peer dependency 충돌 해결해줘"

# Clean reinstall
"node_modules 완전히 초기화하고 다시 설치해줘"

changelog-generator

Generates changelogs from git history following the Keep a Changelog format with Conventional Commits parsing.

# Generate changelog for latest release
"changelog 만들어줘"

# Generate for a specific range
"v1.2.0부터 v1.3.0까지 changelog 생성해줘"

# With category grouping
"changelog을 Added/Changed/Fixed/Removed로 분류해서 만들어줘"

Meta Skills

skill-creator

Bootstraps a new CLI-JAW skill from a template. Generates the SKILL.md orchestrator file, trigger patterns, and a references/ directory for deep guidance docs.

# Create a new skill
"새 스킬 만들어줘: docker-compose-helper"

# Create with trigger phrases
"'도커 컴포즈', 'docker compose'를 트리거로 하는 스킬 만들어줘"

Generated structure:

skills/
  docker-compose-helper/
    SKILL.md           # Orchestrator with role, triggers, and workflow
    references/
      compose-spec.md  # Deep reference material
      examples.md      # Example invocations

Common Patterns

Chaining skills together

Dev workflow skills compose naturally. A typical feature development flow might chain several skills in sequence:

# 1. Brainstorm approaches
"알림 시스템 설계를 브레인스토밍해줘"

# 2. Create an implementation plan
"2번 방안으로 구현 계획 세워줘"

# 3. Develop with TDD
"TDD로 개발 시작해줘"

# 4. Debug any issues
"이 테스트 실패 디버깅해줘"

# 5. Generate changelog
"작업 내용으로 changelog 업데이트해줘"

Combining with goals

For larger tasks, wrap workflow skills inside a goal for progress tracking and persistence:

/goal create "Implement notification system"
# The goal orchestrator will invoke brainstorming, writing-plans,
# tdd, and changelog-generator as needed throughout the lifecycle.