원클릭으로 Manus에서 모든 스킬 실행

complexity-routing

스타0

포크0

업데이트2026년 3월 28일 22:12

Complexity assessment and agent routing — classifies tasks by complexity tier and selects the optimal agent roster and model tier. Use when routing tasks, assessing complexity, or selecting agents.

설치

Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.

Manus에서 실행

출처

kennedym-ds

kennedym-ds/claudecode-orchestrator

GitHub 저장소 열기 Creator 저장소 보기

다운로드

Manus에서 실행

Complexity Routing

Tier Classification

Assess every incoming request against these signals:

INSTANT (handle directly)

Single factual question
Clarification of existing decision
Status check or summary request
No code changes needed

STANDARD (plan → implement → review)

Single-file or few-file change
Well-understood pattern
Clear requirements, no ambiguity
No security or architectural implications

DEEP (research → plan → implement → review → security)

Multi-file or multi-module change
Requires investigation or research first
Security-sensitive components involved
New integration or API surface
Performance-critical path

ULTRADEEP (research → plan → architect → implement → trilateral review)

Architectural change or new system design
Cross-cutting concerns (auth, data model, API contract)
Compliance or regulatory implications
Irreversible changes (data migration, public API)
Multiple team/role involvement needed

Agent Roster by Tier

Tier	Required Agents	Optional Support
INSTANT	Conductor only	—
STANDARD	Planner → Implementer → Reviewer	Doc-Updater
DEEP	Researcher → Planner → Implementer → Reviewer → Security-Reviewer	Test-Architect, Deploy-Engineer
ULTRADEEP	Researcher → Planner → Architect → Implementer → Reviewer + Red-Team + Security-Reviewer	Threat-Modeler, Test-Architect, Deploy-Engineer

Model Tier Selection

Agent Role	STANDARD	DEEP	ULTRADEEP
Orchestration	sonnet	opus	opus
Planning	sonnet	opus	opus
Implementation	sonnet	sonnet	sonnet
Review	sonnet	opus	opus
Security	—	opus	opus
Support	haiku	haiku	sonnet

Signal Detection Keywords

INSTANT: "what is", "how do I", "explain", "show me", "status"
STANDARD: "fix", "add", "update", "change", "implement"
DEEP: "integrate", "secure", "migrate", "refactor", "optimize"
ULTRADEEP: "redesign", "architect", "rebuild", "replace", "new system"

이 저장소의 다른 Skills

같은 저장소

demo-flow

kennedym-ds/claudecode-orchestrator

Demo pipeline state machine — 7-phase autonomous sequence with delegation context templates, phase transition logic, BLOCKED recovery strategies, and demo-state.json schema. Used exclusively by demo-conductor.

2026-03-310

demo-narration

kennedym-ds/claudecode-orchestrator

Cinematic narration style guide for demo-conductor — ANSI-coloured banner formats, live pipeline scoreboard, audience-facing language, phase summaries, and error narration patterns. Keeps the demo presentation-quality throughout.

2026-03-310

completion-protocol

kennedym-ds/claudecode-orchestrator

Standardized completion and escalation protocol for subagent responses. Ensures the conductor can machine-parse every subagent return. Use when reporting completion status back to the orchestrator.

2026-03-300

learnings-mgmt

kennedym-ds/claudecode-orchestrator

Cross-session learnings lifecycle — schema, storage, retrieval, and pruning of lessons learned during orchestrator sessions. Use when managing learnings via the /learn command.

2026-03-300

team-routing

kennedym-ds/claudecode-orchestrator

Agent Teams assembly and task injection — selects appropriate team, validates prerequisites, estimates cost, injects tasks into the shared task list, and manages team lifecycle.

2026-03-300

budget-gatekeeper

kennedym-ds/claudecode-orchestrator

Token and cost tracking with model tier enforcement

2026-03-290

name	complexity-routing
description	Complexity assessment and agent routing — classifies tasks by complexity tier and selects the optimal agent roster and model tier. Use when routing tasks, assessing complexity, or selecting agents.
user-invocable	false

Complexity Routing

Tier Classification

Assess every incoming request against these signals:

INSTANT (handle directly)

Single factual question
Clarification of existing decision
Status check or summary request
No code changes needed

STANDARD (plan → implement → review)

Single-file or few-file change
Well-understood pattern
Clear requirements, no ambiguity
No security or architectural implications

DEEP (research → plan → implement → review → security)

Multi-file or multi-module change
Requires investigation or research first
Security-sensitive components involved
New integration or API surface
Performance-critical path

ULTRADEEP (research → plan → architect → implement → trilateral review)

Architectural change or new system design
Cross-cutting concerns (auth, data model, API contract)
Compliance or regulatory implications
Irreversible changes (data migration, public API)
Multiple team/role involvement needed

Agent Roster by Tier

Tier	Required Agents	Optional Support
INSTANT	Conductor only	—
STANDARD	Planner → Implementer → Reviewer	Doc-Updater
DEEP	Researcher → Planner → Implementer → Reviewer → Security-Reviewer	Test-Architect, Deploy-Engineer
ULTRADEEP	Researcher → Planner → Architect → Implementer → Reviewer + Red-Team + Security-Reviewer	Threat-Modeler, Test-Architect, Deploy-Engineer

Model Tier Selection

Agent Role	STANDARD	DEEP	ULTRADEEP
Orchestration	sonnet	opus	opus
Planning	sonnet	opus	opus
Implementation	sonnet	sonnet	sonnet
Review	sonnet	opus	opus
Security	—	opus	opus
Support	haiku	haiku	sonnet

Signal Detection Keywords

INSTANT: "what is", "how do I", "explain", "show me", "status"
STANDARD: "fix", "add", "update", "change", "implement"
DEEP: "integrate", "secure", "migrate", "refactor", "optimize"
ULTRADEEP: "redesign", "architect", "rebuild", "replace", "new system"