인터페이스는 현지화되어 있으며, Skill 소개와 설명 문서는 선택한 언어로 AI가 자동 번역합니다. 코드·명령어·경로·전문 용어는 원문을 유지하며 원작자 문서가 우선합니다.
Skills 둘러보기
디자인 및 창작 · SKILL.md

local-vco-roles

local-vco-roles

Codex-local role pack for dialectic multi-agent reviews, designed to be compatible with the local vibe skill.

Independent GitHub snapshot · Automated classification · Original documentation governs

Skill 문서

GitHub

name: local-vco-roles description: Codex-local role pack for dialectic multi-agent reviews, designed to be compatible with the local vibe skill.

local-vco-roles

Codex-local role pack for dialectic multi-agent reviews, designed to be compatible with the local vibe skill.

When to Use

Use this skill when:

  1. You run /vibe and need a stable role set for dialectic review.
  2. You want reproducible multi-role analysis using TeamCreate-style orchestration.
  3. You need standardized outputs across team-lead / bug / architecture / integration / usability roles.

Vibe Compatibility Contract (Must Follow)

  1. Grade handshake:
  • M/L: do NOT force TeamCreate orchestration.
  • XL: TeamCreate workflow is allowed.
  1. Rule-1 boundary:
  • Never mix multiple agent systems for the same task.
  • Follow vibe conflict rules first.
  1. Command priority:
  • User explicit command > vibe routing > this role pack defaults.
  1. Output severity:
  • Always use CRITICAL > HIGH > MEDIUM > LOW.
  1. Staged confirmation:
  • Keep major confirmation points aligned with vibe team protocol.
  1. Memory behavior:
  • Default state tracking via TodoWrite-style task state.
  • Optional enhancements (if available) must not become hard dependencies.

Role Prompt Library

  • Team lead: references/role-prompts/team-lead.md
  • Bug analyst: references/role-prompts/bug-analyst.md
  • Architecture critic: references/role-prompts/arch-critic.md
  • Integration analyst: references/role-prompts/integration-analyst.md
  • Usability analyst: references/role-prompts/usability-analyst.md

Recommended XL Workflow

  1. Read references/vibe-compatibility.md.
  2. Create 5 tasks (lead + 4 analysts).
  3. Assign each role using the prompt files above.
  4. Collect results and aggregate by severity.
  5. Produce one merged decision set (keep / simplify / remove).

Optional Utility

  • Run scaffold helper: scripts/new-run.ps1 -Topic <name>
저장한 Skills