This document lists every specialist agent available in the /coord pipeline, with its role, activation condition, subagent_type value, and a concrete invocation example.
μ΄ λ¬Έμλ /coord νμ΄νλΌμΈμμ μ¬μ© κ°λ₯ν λͺ¨λ μ λ¬Έκ° μμ΄μ νΈλ₯Ό μν , νμ±ν 쑰건, subagent_type κ°, ꡬ체μ μΈ νΈμΆ μμμ ν¨κ» λμ΄ν©λλ€.
Agents are dispatched using Claude Code's subagent_type parameter, either by the pipeline automatically or by the user directly via the @sub shorthand.
μμ΄μ νΈλ νμ΄νλΌμΈμ΄ μλμΌλ‘ λλ μ¬μ©μκ° @sub λ¨μΆμ΄λ₯Ό ν΅ν΄ μ§μ subagent_type λ§€κ°λ³μλ₯Ό μ¬μ©νμ¬ λ°μ‘λ©λλ€.
# Internal pipeline dispatch (νμ΄νλΌμΈ λ΄λΆ λ°μ‘)
{
"subagent_type": "explore",
"prompt": "Find all authentication-related files in the repository",
"run_in_background": True
}
# User shorthand (μ¬μ©μ λ¨μΆμ΄)
@sub Find all authentication-related files in the repositoryThe pipeline selects the right agent automatically based on task classification from Stage 2. Manual @sub invocations bypass the pipeline and run the agent directly.
νμ΄νλΌμΈμ Stage 2μ μμ
λΆλ₯λ₯Ό κΈ°λ°μΌλ‘ μλμΌλ‘ μ¬λ°λ₯Έ μμ΄μ νΈλ₯Ό μ νν©λλ€. μλ @sub νΈμΆμ νμ΄νλΌμΈμ μ°ννκ³ μμ΄μ νΈλ₯Ό μ§μ μ€νν©λλ€.
Role: Codebase discovery, file structure mapping, and pattern identification. Use when the task requires understanding what already exists before making changes.
μν : μ½λλ² μ΄μ€ νμ, νμΌ κ΅¬μ‘° λ§€ν, ν¨ν΄ μλ³. λ³κ²½νκΈ° μ μ 무μμ΄ μ‘΄μ¬νλμ§ μ΄ν΄ν΄μΌ ν λ μ¬μ©ν©λλ€.
When to use (μ¬μ© μμ ):
- Any task involving 3 or more search queries (3κ° μ΄μμ κ²μ μΏΌλ¦¬κ° ν¬ν¨λ λͺ¨λ μμ )
- "How is X implemented in this codebase?" questions (μ½λλ² μ΄μ€μμ Xκ° μ΄λ»κ² ꡬνλλμ§ μ§λ¬Έ)
- Before refactoring to understand current architecture (νμ¬ μν€ν μ²λ₯Ό μ΄ν΄νκΈ° μν΄ λ¦¬ν©ν λ§ μ )
Example (μμ):
@sub Find all places in the codebase where database connections are opened,
and check whether they are properly closed in finally blocks.
Role: Multi-source web research, synthesis across academic papers, blog posts, and official documentation. Returns a structured report with citations.
μν : λ€μ€ μμ€ μΉ λ¦¬μμΉ, νμ λ Όλ¬Έ, λΈλ‘κ·Έ ν¬μ€νΈ, 곡μ λ¬Έμ κ° ν©μ±. μΈμ©κ³Ό ν¨κ» ꡬ쑰νλ λ³΄κ³ μλ₯Ό λ°νν©λλ€.
When to use (μ¬μ© μμ ):
- Comparing approaches to solve a technical problem (κΈ°μ μ λ¬Έμ ν΄κ²° μ κ·Όλ² λΉκ΅)
- Finding recent (post-knowledge-cutoff) information (μ§μ μ»·μ€ν μ΄ν μ΅μ μ 보 μ°ΎκΈ°)
- Literature survey for a research paper or proposal (λ Όλ¬Έ λλ μ μμλ₯Ό μν λ¬Έν μ‘°μ¬)
Example (μμ):
@sub Research the current state of transformer model compression techniques
for edge deployment. Compare pruning, quantization, and distillation
approaches. Include papers from 2024 onward.
Role: Implementation strategy design. Produces a structured plan with dependency mapping, parallelization annotations, and resource estimates. Should run before any implementation work begins.
μν : ꡬν μ λ΅ μ€κ³. μμ‘΄μ± λ§€ν, λ³λ ¬ν μ£Όμ, 리μμ€ μΆμ μ΄ μλ ꡬ쑰νλ κ³νμ μμ±ν©λλ€. ꡬν μμ μ΄ μμλκΈ° μ μ μ€νν΄μΌ ν©λλ€.
When to use (μ¬μ© μμ ):
- Before implementing any feature with 3+ steps (3λ¨κ³ μ΄μμ κΈ°λ₯ ꡬν μ )
- When requirements are clear but the execution order is not (μꡬμ¬νμ λͺ ννμ§λ§ μ€ν μμκ° λΆλͺ νν λ)
Example (μμ):
@sub Design an implementation plan for adding rate limiting to the FastAPI
service. The plan must identify which components need changes, the order
of changes, and which steps can be parallelized.
Role: Ambiguous request clarification. Converts vague user intents into precise, numbered requirements with acceptance criteria. Prevents scope creep by surfacing implicit assumptions.
μν : λͺ¨νΈν μμ² λͺ μΈν. λͺ¨νΈν μ¬μ©μ μλλ₯Ό μμ© κΈ°μ€μ΄ μλ μ νν λ²νΈ λ§€κ²¨μ§ μꡬμ¬νμΌλ‘ λ³νν©λλ€.
When to use (μ¬μ© μμ ):
- When the user's request has multiple valid interpretations (μ¬μ©μ μμ²μ μ¬λ¬ μ ν¨ν ν΄μμ΄ μμ λ)
- Before starting any project where the scope is unclear (λ²μκ° λΆλͺ νν νλ‘μ νΈ μμ μ )
Example (μμ):
@sub Clarify this requirement: "Make the dashboard faster."
Identify: which dashboard, which metrics define "faster",
what the current baseline is, and what success looks like.
Role: Full-system cross-domain design. Produces architecture diagrams (ASCII), component responsibility assignments, data flow descriptions, and integration contracts.
μν : μ 체 μμ€ν λλ©μΈ κ° μ€κ³. ASCII μν€ν μ² λ€μ΄μ΄κ·Έλ¨, μ»΄ν¬λνΈ μ± μ ν λΉ, λ°μ΄ν° νλ¦ μ€λͺ , ν΅ν© κ³μ½μ μμ±ν©λλ€.
When to use (μ¬μ© μμ ):
- Designing a new system from scratch (μ μμ€ν μ²μλΆν° μ€κ³)
- Evaluating a proposed architecture change (μ μλ μν€ν μ² λ³κ²½ νκ°)
- Cross-cutting concerns spanning frontend + backend + infra (νλ‘ νΈμλ + λ°±μλ + μΈνλΌμ κ±ΈμΉ ν‘λ¨ κ΄μ¬μ¬)
Example (μμ):
@sub Design the architecture for a real-time collaborative document editor.
Include the WebSocket layer, CRDT conflict resolution approach,
persistence strategy, and how the frontend state synchronizes.
Role: Server-side system design. Specializes in API design, database schema, service decomposition, authentication flows, and caching strategies.
μν : μλ² μ¬μ΄λ μμ€ν μ€κ³. API μ€κ³, λ°μ΄ν°λ² μ΄μ€ μ€ν€λ§, μλΉμ€ λΆν΄, μΈμ¦ νλ¦, μΊμ± μ λ΅μ μ λ¬ΈμΌλ‘ ν©λλ€.
When to use (μ¬μ© μμ ):
- Designing REST or GraphQL APIs (REST λλ GraphQL API μ€κ³)
- Planning database schema migrations (λ°μ΄ν°λ² μ΄μ€ μ€ν€λ§ λ§μ΄κ·Έλ μ΄μ κ³ν)
- Service decomposition from monolith (λͺ¨λ리μ€μμ μλΉμ€ λΆν΄)
Example (μμ):
@sub Design the database schema for a multi-tenant SaaS application where
each tenant has isolated data. Choose between row-level security,
schema-per-tenant, and database-per-tenant, and justify the choice.
Role: UI/UX architecture decisions. Handles state management strategy, component hierarchy, routing architecture, rendering strategy (SSR/CSR/SSG), and accessibility compliance.
μν : UI/UX μν€ν μ² κ²°μ . μν κ΄λ¦¬ μ λ΅, μ»΄ν¬λνΈ κ³μΈ΅, λΌμ°ν μν€ν μ², λ λλ§ μ λ΅(SSR/CSR/SSG), μ κ·Όμ± κ·μ μ€μλ₯Ό μ²λ¦¬ν©λλ€.
When to use (μ¬μ© μμ ):
- Choosing a React state management library (React μν κ΄λ¦¬ λΌμ΄λΈλ¬λ¦¬ μ ν)
- Designing component composition patterns (μ»΄ν¬λνΈ ν©μ± ν¨ν΄ μ€κ³)
- Performance optimization for a large SPA (λκ·λͺ¨ SPA μ±λ₯ μ΅μ ν)
Example (μμ):
@sub Evaluate whether to use Zustand, Jotai, or Redux Toolkit for
the state management layer in a Next.js 14 app with server components.
Consider the server/client boundary implications.
Role: Python-specific implementation. Handles async patterns, type hinting, packaging, testing with pytest, and Python performance patterns.
μν : Python νΉν ꡬν. λΉλκΈ° ν¨ν΄, νμ νν , ν¨ν€μ§, pytestλ₯Ό μ¬μ©ν ν μ€νΈ, Python μ±λ₯ ν¨ν΄μ μ²λ¦¬ν©λλ€.
When to use (μ¬μ© μμ ):
- Writing Python code that must follow strict typing conventions (μ격ν νμ΄ν κ·μΉμ λ°λΌμΌ νλ Python μ½λ μμ±)
- Async FastAPI or Django endpoints (λΉλκΈ° FastAPI λλ Django μλν¬μΈνΈ)
- Scientific computing with NumPy, pandas, or PyTorch (NumPy, pandas, PyTorchλ₯Ό μ¬μ©ν κ³Όν κ³μ°)
Example (μμ):
@sub Implement an async context manager in Python 3.12 that manages
a connection pool for PostgreSQL using asyncpg. Include type hints,
docstrings, and a pytest fixture that uses the manager.
Role: Code quality improvement without behavior changes. Identifies dead code, extracts functions, removes duplication, applies SOLID principles, and improves naming.
μν : λμ λ³κ²½ μλ μ½λ νμ§ κ°μ . λ°λ μ½λ μλ³, ν¨μ μΆμΆ, μ€λ³΅ μ κ±°, SOLID μμΉ μ μ©, μ΄λ¦ κ°μ μ μνν©λλ€.
When to use (μ¬μ© μμ ):
- When code works but is difficult to maintain (μ½λλ μλνμ§λ§ μ μ§ κ΄λ¦¬κ° μ΄λ €μΈ λ)
- Before adding new features to legacy code (λ κ±°μ μ½λμ μ κΈ°λ₯μ μΆκ°νκΈ° μ )
- Code review cleanup (μ½λ 리뷰 μ 리)
Example (μμ):
@sub Refactor the authentication module in src/auth/. It currently has a
500-line God class. Extract the token validation, session management,
and OAuth flow into separate classes. Do not change behavior.
Role: Independent review with falsifiability checking. Every claim in the reviewed output must have a "wrong if X" condition. Raises compliance issues, logical gaps, and sycophantic reasoning.
μν : λ°μ¦ κ°λ₯μ± κ²μ¬μ ν¨κ» λ 립μ μΈ κ²ν . κ²ν λ μΆλ ₯μ λͺ¨λ μ£Όμ₯μλ "XμΈ κ²½μ° νλ¦Ό" μ‘°κ±΄μ΄ μμ΄μΌ ν©λλ€.
When to use (μ¬μ© μμ ):
- Automatically after Stage 4 when Opus model is used (Opus λͺ¨λΈ μ¬μ© μ Stage 4 ν μλ)
- Manually via
/challengecommand to critique the last response (λ§μ§λ§ μλ΅μ λΉννκΈ° μν΄/challengeλͺ λ ΉμΌλ‘ μλ) - Before finalizing any proposal, paper, or production deployment (μ μμ, λ Όλ¬Έ, νλ‘λμ λ°°ν¬ νμ μ )
Example (μμ):
@sub Review the proposed caching strategy below and identify:
1. Any claims without evidence
2. Edge cases not handled
3. Performance assumptions that could be wrong
[paste the caching strategy]
Role: Test suite design and coverage analysis. Creates unit tests, integration tests, and property-based tests. Identifies uncovered code paths and writes assertions with meaningful failure messages.
μν : ν μ€νΈ μ€μνΈ μ€κ³ λ° μ»€λ²λ¦¬μ§ λΆμ. λ¨μ ν μ€νΈ, ν΅ν© ν μ€νΈ, μμ± κΈ°λ° ν μ€νΈλ₯Ό μμ±ν©λλ€.
When to use (μ¬μ© μμ ):
- After implementing a new feature (μ κΈ°λ₯ ꡬν ν)
- When test coverage drops below the project threshold (ν μ€νΈ 컀λ²λ¦¬μ§κ° νλ‘μ νΈ μκ³κ° μλλ‘ λ¨μ΄μ§ λ)
- Property-based testing for data transformation functions (λ°μ΄ν° λ³ν ν¨μμ μμ± κΈ°λ° ν μ€νΈ)
Example (μμ):
@sub Write a comprehensive pytest test suite for the payment processing module
in src/payments/. Cover: successful transactions, card decline scenarios,
network timeout handling, and idempotency of repeated calls.
Role: Threat modeling and vulnerability review. Applies OWASP Top 10, checks for injection vulnerabilities, reviews authentication and authorization logic, and assesses data exposure risks.
μν : μν λͺ¨λΈλ§ λ° μ·¨μ½μ κ²ν . OWASP Top 10 μ μ©, μΈμ μ μ·¨μ½μ νμΈ, μΈμ¦ λ° κΆν λΆμ¬ λ‘μ§ κ²ν , λ°μ΄ν° λ ΈμΆ μν νκ°λ₯Ό μνν©λλ€.
When to use (μ¬μ© μμ ):
- Before deploying any user-facing authentication feature (μ¬μ©μ λλ©΄ μΈμ¦ κΈ°λ₯ λ°°ν¬ μ )
- When handling personally identifiable information (κ°μΈ μλ³ μ 보 μ²λ¦¬ μ)
- Annual security audit of a service (μλΉμ€μ μ°κ° 보μ κ°μ¬)
Example (μμ):
@sub Perform a security review of the user registration and login flows
in src/auth/. Check for: SQL injection, timing attacks on password
comparison, missing rate limiting, and insecure password storage.
Role: Profiling and optimization. Identifies algorithmic complexity issues, database N+1 queries, memory leaks, and unnecessary network round trips. Provides before/after benchmarks.
μν : νλ‘νμΌλ§ λ° μ΅μ ν. μκ³ λ¦¬μ¦ λ³΅μ‘λ λ¬Έμ , λ°μ΄ν°λ² μ΄μ€ N+1 쿼리, λ©λͺ¨λ¦¬ λμ, λΆνμν λ€νΈμν¬ λΌμ΄λ νΈλ¦½μ μλ³ν©λλ€.
When to use (μ¬μ© μμ ):
- When profiling shows a hot path (νλ‘νμΌλ§μ΄ ν« ν¨μ€λ₯Ό νμν λ)
- Unexplained latency increase after a deployment (λ°°ν¬ ν μ€λͺ λμ§ μλ μ§μ° μ¦κ°)
- Scaling review before a traffic spike (νΈλν½ κΈμ¦ μ νμ₯μ± κ²ν )
Example (μμ):
@sub Analyze the performance of the report generation endpoint. It currently
takes 8 seconds for a report covering 10,000 rows. Identify the
bottleneck and propose a fix with estimated improvement.
Role: Documentation creation and improvement. Writes API references, user guides, architecture docs, README files, and inline code comments. Focuses on clarity for the target audience and WCAG-compliant structure.
μν : λ¬Έμ μμ± λ° κ°μ . API μ°Έμ‘°, μ¬μ©μ κ°μ΄λ, μν€ν μ² λ¬Έμ, README νμΌ, μΈλΌμΈ μ½λ μ£Όμμ μμ±ν©λλ€.
When to use (μ¬μ© μμ ):
- After implementing a new public API (μ κ³΅κ° API ꡬν ν)
- Onboarding documentation for a new team member (μ νμμ μν μ¨λ³΄λ© λ¬Έμ)
- When existing docs are outdated or missing examples (κΈ°μ‘΄ λ¬Έμκ° μ€λλμκ±°λ μμκ° μμ λ)
Example (μμ):
@sub Write API reference documentation for the /payments endpoint.
Audience: external developers integrating for the first time.
Include: request format, all response codes, error object schema,
and a working cURL example for each scenario.
Role: Scientific hypothesis-evidence-validation workflow using Cynefin triage, Popperian falsifiability, and Bayesian updating. An 8-stage structured reasoning process for any complex problem domain.
μν : Cynefin λΆλ₯, Popperian λ°μ¦ κ°λ₯μ±, Bayesian μ λ°μ΄νΈλ₯Ό μ¬μ©νλ κ³Όνμ κ°μ€-μ¦κ±°-κ²μ¦ μν¬νλ‘μ°. λͺ¨λ 볡μ‘ν λ¬Έμ λλ©μΈμ μν 8λ¨κ³ ꡬ쑰νλ μΆλ‘ κ³Όμ μ λλ€.
When to use (μ¬μ© μμ ):
- Diagnosing a system behavior that is not understood (μ΄ν΄λμ§ μλ μμ€ν λμ μ§λ¨)
- Evaluating competing explanations for an experimental result (μ€ν κ²°κ³Όμ λν κ²½μμ μ€λͺ νκ°)
- Any problem where intuition alone is insufficient (μ§κ΄λ§μΌλ‘λ λΆμΆ©λΆν λͺ¨λ λ¬Έμ )
8-stage workflow (8λ¨κ³ μν¬νλ‘μ°):
- Cynefin domain classification (Cynefin λλ©μΈ λΆλ₯)
- Phenomenon description (νμ μ€λͺ )
- Hypothesis generation (κ°μ€ μμ±)
- Evidence inventory (μ¦κ±° λͺ©λ‘)
- Bayesian likelihood scoring (Bayesian κ°λ₯μ± μ μ λ§€κΈ°κΈ°)
- Decisive experiment design (κ²°μ μ μ€ν μ€κ³)
- Prediction registration (μμΈ‘ λ±λ‘)
- Update and conclusion (μ λ°μ΄νΈ λ° κ²°λ‘ )
Example (μμ):
@sub Apply sci-method to this anomaly: our model achieves 95% AUC on the
validation set but only 61% on the held-out test set. Generate ranked
hypotheses with testable predictions for each.
Role: Systematic failure investigation. Applies the "five whys" method combined with evidence gathering. Distinguishes root causes from symptoms. Produces a causal chain diagram.
μν : 체κ³μ μΈ μ€ν¨ μ‘°μ¬. μ¦κ±° μμ§κ³Ό κ²°ν©λ "5 Why" λ°©λ²μ μ μ©ν©λλ€. κ·Όλ³Έ μμΈκ³Ό μ¦μμ ꡬλΆν©λλ€.
When to use (μ¬μ© μμ ):
- After a production incident (νλ‘λμ μΈμλνΈ ν)
- When intermediate validation in Stage 4 fails (Stage 4μ μ€κ° κ²μ¦μ΄ μ€ν¨ν λ)
- Recurring bugs that have been fixed and returned (μμ λμλ€κ° λ€μ λ°μνλ λ²κ·Έ)
Example (μμ):
@sub The CI pipeline has been failing intermittently for 2 weeks.
Failures occur in the integration test step, always between 02:00β04:00 UTC.
Apply root cause analysis: gather evidence, generate hypotheses, and
recommend the decisive diagnostic step.
| I need toβ¦ (ν μμ ) | Use this agent (μ¬μ©ν μμ΄μ νΈ) |
|---|---|
| Find existing code patterns | explore |
| Research a technical topic online | deep-research-agent |
| Plan implementation before coding | plan |
| Clarify vague requirements | requirements-analyst |
| Design a full system | system-architect |
| Design server APIs and schemas | backend-architect |
| Design UI components and state | frontend-architect |
| Write Python code | python-expert |
| Improve code quality without behavior change | refactoring-expert |
| Independently review output for errors | critic |
| Write or improve tests | quality-engineer |
| Check for security vulnerabilities | security-engineer |
| Find and fix performance bottlenecks | performance-engineer |
| Write documentation or guides | technical-writer |
| Investigate a complex anomaly scientifically | sci-method |
| Find the root cause of a failure | root-cause-analyst |