β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ•— β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ•—β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ•—β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ•— β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ•— β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ•— β–ˆβ–ˆβ•— β–ˆβ–ˆβ•—β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ•—β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ•— β–ˆβ–ˆβ•”β•β•β–ˆβ–ˆβ•—β–ˆβ–ˆβ•”β•β•β•β•β•β•šβ•β•β–ˆβ–ˆβ•”β•β•β•β–ˆβ–ˆβ•”β•β•β–ˆβ–ˆβ•—β–ˆβ–ˆβ•”β•β•β•β–ˆβ–ˆβ•— β–ˆβ–ˆβ•”β•β•β–ˆβ–ˆβ•—β•šβ–ˆβ–ˆβ•— β–ˆβ–ˆβ•”β•β–ˆβ–ˆβ•”β•β•β•β•β•β–ˆβ–ˆβ•”β•β•β–ˆβ–ˆβ•— β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ•”β•β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ•— β–ˆβ–ˆβ•‘ β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ•”β•β–ˆβ–ˆβ•‘ β–ˆβ–ˆβ•‘ β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ•”β• β•šβ–ˆβ–ˆβ–ˆβ–ˆβ•”β• β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ•— β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ•”β• β–ˆβ–ˆβ•”β•β•β–ˆβ–ˆβ•—β–ˆβ–ˆβ•”β•β•β• β–ˆβ–ˆβ•‘ β–ˆβ–ˆβ•”β•β•β–ˆβ–ˆβ•—β–ˆβ–ˆβ•‘ β–ˆβ–ˆβ•‘ β–ˆβ–ˆβ•”β•β•β–ˆβ–ˆβ•— β•šβ–ˆβ–ˆβ•”β• β–ˆβ–ˆβ•”β•β•β• β–ˆβ–ˆβ•”β•β•β–ˆβ–ˆβ•— β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ•”β•β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ•— β–ˆβ–ˆβ•‘ β–ˆβ–ˆβ•‘ β–ˆβ–ˆβ•‘β•šβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ•”β• β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ•”β• β–ˆβ–ˆβ•‘ β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ•—β–ˆβ–ˆβ•‘ β–ˆβ–ˆβ•‘ β•šβ•β•β•β•β•β• β•šβ•β•β•β•β•β•β• β•šβ•β• β•šβ•β• β•šβ•β• β•šβ•β•β•β•β•β• β•šβ•β•β•β•β•β• β•šβ•β• β•šβ•β•β•β•β•β•β•β•šβ•β• β•šβ•β•

Engineering Discipline Extension μ—”μ§€λ‹ˆμ–΄λ§ λ””μ‚¬μ΄ν”Œλ¦° μ΅μŠ€ν…μ…˜

Strict engineering discipline and agentic orchestration for the pi coding agent. Multi-agent workflows, executable plans, and milestone-level decomposition. pi μ½”λ”© μ—μ΄μ „νŠΈλ₯Ό μœ„ν•œ μ—„κ²©ν•œ μ—”μ§€λ‹ˆμ–΄λ§ λ””μ‚¬μ΄ν”Œλ¦°κ³Ό μ—μ΄μ „νŠΈ μ˜€μΌ€μŠ€νŠΈλ ˆμ΄μ…˜. 닀쀑 μ—μ΄μ „νŠΈ μ›Œν¬ν”Œλ‘œ, μ‹€ν–‰ κ°€λŠ₯ν•œ κ³„νš, λ§ˆμΌμŠ€ν†€ λΆ„ν•΄.

MIT License

● Why ROACH PI? μ™œ ROACH PIμΈκ°€μš”?

●
Fully Open Source μ™„μ „ν•œ μ˜€ν”ˆμ†ŒμŠ€

Every line is on GitHub. No hidden prompts, no secret system instructions, no obfuscated behavior. Tools, event hooks, skill injections, and agent prompts are all plain TypeScript and Markdown.

λͺ¨λ“  μ½”λ“œκ°€ GitHub에 κ³΅κ°œλ˜μ–΄ μžˆμŠ΅λ‹ˆλ‹€. μˆ¨κ²¨μ§„ ν”„λ‘¬ν”„νŠΈ, λΉ„λ°€ μ‹œμŠ€ν…œ μ§€μ‹œμ–΄, λ‚œλ…ν™”λœ λ™μž‘ μ—†μŒ. 도ꡬ, 이벀트 ν›…, μŠ€ν‚¬ μ£Όμž…, μ—μ΄μ „νŠΈ ν”„λ‘¬ν”„νŠΈκ°€ λͺ¨λ‘ 일반 TypeScript와 Markdown으둜 μž‘μ„±λ˜μ–΄ μžˆμŠ΅λ‹ˆλ‹€.

β˜…
Observable μ˜΅μ €λ²„λΉŒλ¦¬ν‹°

The footer displays real-time metrics every session: prompt cache hit rate, context usage bar, active tools, branch, and model. See exactly how your context is being utilized.

푸터에 λ§€ μ„Έμ…˜ μ‹€μ‹œκ°„ λ©”νŠΈλ¦­μ΄ ν‘œμ‹œλ©λ‹ˆλ‹€: ν”„λ‘¬ν”„νŠΈ μΊμ‹œ 적쀑λ₯ , μ»¨ν…μŠ€νŠΈ μ‚¬μš©λŸ‰ λ°”, ν™œμ„± 도ꡬ, 브랜치, λͺ¨λΈ. μ»¨ν…μŠ€νŠΈκ°€ μ–΄λ–»κ²Œ μ‚¬μš©λ˜λŠ”μ§€ μ •ν™•νžˆ ν™•μΈν•˜μ„Έμš”.

β– 
Transparent by Design λ””μžμΈλΆ€ν„° 투λͺ…ν•˜κ²Œ

No magic. Read the source and see exactly what the agent does. Skill rules are plain Markdown. Agent prompts are plain text. Every event hook is auditable.

λ§ˆλ²• μ—†μŒ. μ†ŒμŠ€λ₯Ό 읽고 μ—μ΄μ „νŠΈκ°€ μ •ν™•νžˆ 무엇을 ν•˜λŠ”μ§€ ν™•μΈν•˜μ„Έμš”. μŠ€ν‚¬ κ·œμΉ™μ€ 일반 Markdown, μ—μ΄μ „νŠΈ ν”„λ‘¬ν”„νŠΈλŠ” 일반 ν…μŠ€νŠΈ, λͺ¨λ“  이벀트 훅이 감사 κ°€λŠ₯ν•©λ‹ˆλ‹€.

β–Ά
Simple Over Clever 화렀함보닀 λ‹¨μˆœν•¨

We prefer straightforward solutions over elaborate orchestration. Fewer moving parts, fewer failure modes. See our contributing guidelines.

λ³΅μž‘ν•œ μ˜€μΌ€μŠ€νŠΈλ ˆμ΄μ…˜λ³΄λ‹€ λ‹¨μˆœν•œ 해결책을 μ„ ν˜Έν•©λ‹ˆλ‹€. μ›€μ§μ΄λŠ” 뢀뢄이 μ μ„μˆ˜λ‘ μ‹€νŒ¨λ„ μ μŠ΅λ‹ˆλ‹€. κΈ°μ—¬ κ°€μ΄λ“œλΌμΈμ„ μ°Έμ‘°ν•˜μ„Έμš”.

β–Ά Quick Start λΉ λ₯Έ μ‹œμž‘

The extension provides a three-phase workflow: clarify β†’ plan β†’ execute. Each phase is driven by a slash command.

μ΅μŠ€ν…μ…˜μ€ 3단계 μ›Œν¬ν”Œλ‘œλ₯Ό μ œκ³΅ν•©λ‹ˆλ‹€: λͺ…ν™•ν™” β†’ κ³„νš β†’ μ‹€ν–‰. 각 λ‹¨κ³„λŠ” μŠ¬λž˜μ‹œ λͺ…λ ΉμœΌλ‘œ κ΅¬λ™λ©λ‹ˆλ‹€.

  1. /setup β€” Configure Settings β€” μ„€μ • ꡬ성

    Run once after installation to apply recommended settings. Sets quietStartup: true so the extension's custom ROACH PI banner replaces the default startup listing.

    μ„€μΉ˜ ν›„ ν•œ 번 μ‹€ν–‰ν•˜μ—¬ ꢌμž₯ 섀정을 μ μš©ν•©λ‹ˆλ‹€. μ΅μŠ€ν…μ…˜μ˜ μ»€μŠ€ν…€ ROACH PI λ°°λ„ˆκ°€ κΈ°λ³Έ μ‹œμž‘ λͺ©λ‘μ„ λŒ€μ²΄ν•˜λ„λ‘ quietStartup: trueλ₯Ό μ„€μ •ν•©λ‹ˆλ‹€.

  2. /clarify β€” Resolve Ambiguity β€” λͺ¨ν˜Έμ„± ν•΄κ²°

    The agent asks dynamic, context-aware questions one at a time while exploring the codebase in parallel. Ends with a structured Context Brief that defines scope, constraints, and success criteria.

    μ—μ΄μ „νŠΈκ°€ μ½”λ“œλ² μ΄μŠ€λ₯Ό λ³‘λ ¬λ‘œ νƒμƒ‰ν•˜λ©΄μ„œ 동적이고 λ¬Έλ§₯에 λ§žλŠ” μ§ˆλ¬Έμ„ ν•˜λ‚˜μ”© λ¬»μŠ΅λ‹ˆλ‹€. λ²”μœ„, μ œμ•½, 성곡 기쀀을 μ •μ˜ν•˜λŠ” κ΅¬μ‘°ν™”λœ μ»¨ν…μŠ€νŠΈ λΈŒλ¦¬ν”„λ‘œ λλ‚©λ‹ˆλ‹€.

  3. /plan β€” Create an Executable Plan β€” μ‹€ν–‰ κ°€λŠ₯ν•œ κ³„νš 생성

    Transforms the Context Brief into an implementation plan with exact file paths, code blocks, commands, and expected outputs. No placeholders β€” every step is immediately executable by a worker agent.

    μ»¨ν…μŠ€νŠΈ λΈŒλ¦¬ν”„λ₯Ό μ •ν™•ν•œ 파일 경둜, μ½”λ“œ 블둝, λͺ…λ Ήμ–΄, μ˜ˆμƒ 좜λ ₯이 ν¬ν•¨λœ κ΅¬ν˜„ κ³„νšμœΌλ‘œ λ³€ν™˜ν•©λ‹ˆλ‹€. ν”Œλ ˆμ΄μŠ€ν™€λ” μ—†μŒ β€” λͺ¨λ“  단계가 μ›Œμ»€ μ—μ΄μ „νŠΈκ°€ μ¦‰μ‹œ μ‹€ν–‰ν•  수 μžˆμŠ΅λ‹ˆλ‹€.

  4. /ultraplan β€” Milestone Decomposition β€” λ§ˆμΌμŠ€ν†€ λΆ„ν•΄

    For complex multi-day tasks. Dispatches 5 independent reviewer agents in parallel (Feasibility, Architecture, Risk, Dependency, User Value), then synthesizes findings into an optimized milestone DAG.

    λ³΅μž‘ν•œ 며칠짜리 μž‘μ—…μ„ μœ„ν•΄. 5개의 독립적인 리뷰어 μ—μ΄μ „νŠΈλ₯Ό λ³‘λ ¬λ‘œ λ””μŠ€νŒ¨μΉ˜(타당성, μ•„ν‚€ν…μ²˜, 리슀크, μ˜μ‘΄μ„±, μ‚¬μš©μž κ°€μΉ˜)ν•œ ν›„, μ΅œμ ν™”λœ λ§ˆμΌμŠ€ν†€ DAG둜 μ’…ν•©ν•©λ‹ˆλ‹€.

  • /loop β€” Session Loop Scheduler β€” μ„Έμ…˜ 루프 μŠ€μΌ€μ€„λŸ¬

    Schedule recurring prompts at fixed intervals. Cron-style timing β€” fires on schedule regardless of execution state. Supports 5s, 10m, 2h, 1d formats. Up to 100 concurrent jobs with per-job error isolation and automatic cleanup on session shutdown.

    κ³ μ • κ°„κ²©μœΌλ‘œ 반볡 ν”„λ‘¬ν”„νŠΈλ₯Ό μŠ€μΌ€μ€„λ§ν•©λ‹ˆλ‹€. Cron μŠ€νƒ€μΌ 타이밍 β€” μ‹€ν–‰ μƒνƒœμ™€ 관계없이 μŠ€μΌ€μ€„λŒ€λ‘œ μ‹€ν–‰λ©λ‹ˆλ‹€. 5s, 10m, 2h, 1d ν˜•μ‹μ„ μ§€μ›ν•©λ‹ˆλ‹€. μ΅œλŒ€ 100개 λ™μ‹œ μž‘μ—…, μž‘μ—…λ³„ μ—λŸ¬ 격리, μ„Έμ…˜ μ’…λ£Œ μ‹œ μžλ™ 정리.

  • πŸ’‘ Tip: The ask_user_question tool is always available β€” the agent will ask clarifying questions autonomously whenever it detects ambiguity, even outside /clarify mode.

    πŸ’‘ 팁: ask_user_question λ„κ΅¬λŠ” 항상 μ‚¬μš© κ°€λŠ₯ν•©λ‹ˆλ‹€ β€” /clarify λͺ¨λ“œ μ™Έμ—μ„œλ„ λͺ¨ν˜Έμ„±μ„ κ°μ§€ν•˜λ©΄ μžλ™μœΌλ‘œ μ§ˆλ¬Έν•©λ‹ˆλ‹€.

    β–  Key Features μ£Όμš” κΈ°λŠ₯

    β˜…
    Multi-Agent Orchestration 닀쀑 μ—μ΄μ „νŠΈ μ˜€μΌ€μŠ€νŠΈλ ˆμ΄μ…˜

    12 bundled agents for exploration, execution, planning, and review. Delegate tasks via single, parallel (max 8 tasks, 4 concurrent), or chain modes with cycle detection and depth guards.

    탐색, μ‹€ν–‰, κ³„νš, 리뷰λ₯Ό μœ„ν•œ 12개 λ²ˆλ“€ μ—μ΄μ „νŠΈ. 사이클 감지 및 깊이 κ°€λ“œμ™€ ν•¨κ»˜ 단일, 병렬(μ΅œλŒ€ 8개 μž‘μ—…, 4개 λ™μ‹œ), 체인 λͺ¨λ“œλ‘œ μž‘μ—…μ„ μœ„μž„ν•˜μ„Έμš”.

    ●
    Engineering Discipline μ—”μ§€λ‹ˆμ–΄λ§ λ””μ‚¬μ΄ν”Œλ¦°

    Karpathy rules auto-injected into code-writing agents. Automatic slop-cleaner spawns after successful runs to remove LLM-specific patterns while preserving behavior.

    Karpathy κ·œμΉ™μ΄ μ½”λ“œ μž‘μ„± μ—μ΄μ „νŠΈμ— μžλ™ μ£Όμž…λ©λ‹ˆλ‹€. 성곡적인 μ‹€ν–‰ ν›„ λ™μž‘μ„ μœ μ§€ν•˜λ©΄μ„œ LLM νŠΉμ • νŒ¨ν„΄μ„ μ œκ±°ν•˜λŠ” μžλ™ 슬둭 ν΄λ¦¬λ„ˆκ°€ μ‹€ν–‰λ©λ‹ˆλ‹€.

    β– 
    Plan Execution Pipeline κ³„νš μ‹€ν–‰ νŒŒμ΄ν”„λΌμΈ

    plan-compliance β†’ plan-worker β†’ plan-validator pipeline with information barriers. Validators never see execution context β€” they judge solely from the plan document and codebase.

    정보 μž₯벽이 μžˆλŠ” plan-compliance β†’ plan-worker β†’ plan-validator νŒŒμ΄ν”„λΌμΈ. κ²€μ¦μžλŠ” μ‹€ν–‰ μ»¨ν…μŠ€νŠΈλ₯Ό 보지 μ•ŠμŠ΅λ‹ˆλ‹€ β€” κ³„νš λ¬Έμ„œμ™€ μ½”λ“œλ² μ΄μŠ€λ§ŒμœΌλ‘œ νŒλ‹¨ν•©λ‹ˆλ‹€.

    β–Ά
    TUI Customization TUI μ»€μŠ€ν„°λ§ˆμ΄μ œμ΄μ…˜

    Custom ROACH PI ASCII banner, branded footer with directory, branch, model, context bar, cache hit rate, and active tools display.

    디렉토리, 브랜치, λͺ¨λΈ, μ»¨ν…μŠ€νŠΈ λ°”, μΊμ‹œ 적쀑λ₯ , ν™œμ„± 도ꡬ ν‘œμ‹œκ°€ μžˆλŠ” μ»€μŠ€ν…€ ROACH PI ASCII λ°°λ„ˆ 및 λΈŒλžœλ””λ“œ ν‘Έν„°.

    ●
    Context Management μ»¨ν…μŠ€νŠΈ 관리

    Microcompaction truncates old tool results. Phase-aware summarization preserves workflow state across compaction. Phase and goal document survive context limits.

    λ§ˆμ΄ν¬λ‘œμ»΄νŒ©μ…˜μ΄ 였래된 도ꡬ κ²°κ³Όλ₯Ό μžλ¦…λ‹ˆλ‹€. 단계 인식 μš”μ•½μ΄ μ»΄νŒ©μ…˜ μ „λ°˜μ— 걸쳐 μ›Œν¬ν”Œλ‘œ μƒνƒœλ₯Ό λ³΄μ‘΄ν•©λ‹ˆλ‹€. 단계 및 λͺ©ν‘œ λ¬Έμ„œκ°€ μ»¨ν…μŠ€νŠΈ ν•œκ³„λ₯Ό μ΄ˆκ³Όν•©λ‹ˆλ‹€.

    β–Ά
    Session Loop Scheduler μ„Έμ…˜ 루프 μŠ€μΌ€μ€„λŸ¬

    Schedule recurring prompts with /loop 5m check status. Cron-style fixed intervals, per-job error isolation, AbortController cancellation, timeout protection, and automatic cleanup on session shutdown. Up to 100 concurrent jobs.

    /loop 5m check status둜 반볡 ν”„λ‘¬ν”„νŠΈλ₯Ό μŠ€μΌ€μ€„λ§ν•©λ‹ˆλ‹€. Cron μŠ€νƒ€μΌ κ³ μ • 간격, μž‘μ—…λ³„ μ—λŸ¬ 격리, AbortController μ·¨μ†Œ, νƒ€μž„μ•„μ›ƒ 보호, μ„Έμ…˜ μ’…λ£Œ μ‹œ μžλ™ 정리. μ΅œλŒ€ 100개 λ™μ‹œ μž‘μ—….

    β˜…
    11 Behavioral Skills 11개 행동 μŠ€ν‚¬

    Bundled LLM skill rulesets: clarification, plan crafting, run plan, review work, simplify, systematic debugging, Karpathy discipline, Rob Pike optimization, and more.

    λ²ˆλ“€ LLM μŠ€ν‚¬ λ£°μ…‹: λͺ…ν™•ν™”, κ³„νš μž‘μ„±, κ³„νš μ‹€ν–‰, μž‘μ—… 리뷰, λ‹¨μˆœν™”, 체계적 디버깅, Karpathy λ””μ‚¬μ΄ν”Œλ¦°, Rob Pike μ΅œμ ν™” λ“±.

    ● FAQ

    What is the pi coding agent? pi μ½”λ”© μ—μ΄μ „νŠΈλž€ λ¬΄μ—‡μΈκ°€μš”?

    pi is a terminal-based AI coding agent that helps you read, write, and debug code directly from your terminal. It supports extensions, custom tools, themes, and multiple AI providers. GitHub β†’

    piλŠ” ν„°λ―Έλ„μ—μ„œ 직접 μ½”λ“œλ₯Ό 읽고, μ“°κ³ , λ””λ²„κΉ…ν•˜λŠ” 터미널 기반 AI μ½”λ”© μ—μ΄μ „νŠΈμž…λ‹ˆλ‹€. μ΅μŠ€ν…μ…˜, μ»€μŠ€ν…€ 도ꡬ, ν…Œλ§ˆ 및 μ—¬λŸ¬ AI ν”„λ‘œλ°”μ΄λ”λ₯Ό μ§€μ›ν•©λ‹ˆλ‹€. GitHub β†’

    Do I need to install the engineering-discipline skills separately? μ—”μ§€λ‹ˆμ–΄λ§ λ””μ‚¬μ΄ν”Œλ¦° μŠ€ν‚¬μ„ λ³„λ„λ‘œ μ„€μΉ˜ν•΄μ•Ό ν•˜λ‚˜μš”?

    No. All engineering discipline skills are bundled with this extension. Just install the extension and run /setup β€” everything is ready out of the box.

    μ•„λ‹ˆμš”. λͺ¨λ“  μ—”μ§€λ‹ˆμ–΄λ§ λ””μ‚¬μ΄ν”Œλ¦° μŠ€ν‚¬μ΄ 이 μ΅μŠ€ν…μ…˜μ— λ²ˆλ“€λ‘œ ν¬ν•¨λ˜μ–΄ μžˆμŠ΅λ‹ˆλ‹€. μ΅μŠ€ν…μ…˜μ„ μ„€μΉ˜ν•˜κ³  /setup을 μ‹€ν–‰ν•˜μ„Έμš” β€” λͺ¨λ“  것이 μ¦‰μ‹œ μ‚¬μš© κ°€λŠ₯ν•©λ‹ˆλ‹€.

    How do the subagent modes work? μ„œλΈŒμ—μ΄μ „νŠΈ λͺ¨λ“œλŠ” μ–΄λ–»κ²Œ μž‘λ™ν•˜λ‚˜μš”?

    Single: One-off tasks sent to a specific agent.
    Parallel: Dispatch multiple agents at once (max 8 tasks, 4 concurrent). Each runs independently.
    Chain: Sequential pipeline where each step receives the previous step's output via {previous} placeholder.
    All modes include cycle detection, depth limits (max 3), and concurrency control.

    단일: νŠΉμ • μ—μ΄μ „νŠΈμ— λ³΄λ‚΄λŠ” μΌνšŒμ„± μž‘μ—….
    병렬: μ—¬λŸ¬ μ—μ΄μ „νŠΈλ₯Ό ν•œ λ²ˆμ— λ””μŠ€νŒ¨μΉ˜(μ΅œλŒ€ 8개 μž‘μ—…, 4개 λ™μ‹œ). 각각 λ…λ¦½μ μœΌλ‘œ μ‹€ν–‰.
    체인: 각 단계가 이전 λ‹¨κ³„μ˜ 좜λ ₯을 {previous} ν”Œλ ˆμ΄μŠ€ν™€λ”λ‘œ λ°›λŠ” 순차 νŒŒμ΄ν”„λΌμΈ.
    λͺ¨λ“  λͺ¨λ“œμ— 사이클 감지, 깊이 μ œν•œ(μ΅œλŒ€ 3), λ™μ‹œμ„± μ œμ–΄κ°€ ν¬ν•¨λ©λ‹ˆλ‹€.

    What does the slop-cleaner do? 슬둭 ν΄λ¦¬λ„ˆλŠ” 무엇을 ν•˜λ‚˜μš”?

    The slop-cleaner automatically runs after successful worker or plan-worker executions. It detects and removes LLM-specific code patterns (unnecessary comments, over-verbose logging, redundant type assertions, etc.) while preserving the original behavior.

    슬둭 ν΄λ¦¬λ„ˆλŠ” worker λ˜λŠ” plan-worker μ‹€ν–‰ 성곡 ν›„ μžλ™μœΌλ‘œ μ‹€ν–‰λ©λ‹ˆλ‹€. μ›λž˜ λ™μž‘μ„ μœ μ§€ν•˜λ©΄μ„œ LLM νŠΉμ • μ½”λ“œ νŒ¨ν„΄(λΆˆν•„μš”ν•œ 주석, κ³Όλ„ν•œ λ‘œκΉ…, 쀑볡 νƒ€μž… 단언 λ“±)을 κ°μ§€ν•˜κ³  μ œκ±°ν•©λ‹ˆλ‹€.

    Can I use this extension without the /ultraplan command? /ultraplan λͺ…λ Ή 없이 이 μ΅μŠ€ν…μ…˜μ„ μ‚¬μš©ν•  수 μžˆλ‚˜μš”?

    Absolutely. The three slash commands are independent. Use /clarify + /plan for single-session tasks. Use /ultraplan only when you need milestone-level decomposition for complex, multi-day projects.

    λ¬Όλ‘ μž…λ‹ˆλ‹€. μ„Έ 개의 μŠ¬λž˜μ‹œ λͺ…령은 λ…λ¦½μ μž…λ‹ˆλ‹€. 단일 μ„Έμ…˜ μž‘μ—…μ—λŠ” /clarify + /plan을 μ‚¬μš©ν•˜μ„Έμš”. λ³΅μž‘ν•œ 며칠짜리 ν”„λ‘œμ νŠΈμ— λ§ˆμΌμŠ€ν†€ μˆ˜μ€€ λΆ„ν•΄κ°€ ν•„μš”ν•  λ•Œλ§Œ /ultraplan을 μ‚¬μš©ν•˜μ„Έμš”.