Agent skill
agent-teams
Path D 并行分工模板 — Agent Teams 任务分配、上下文传递、结果合并
Install this agent skill to your Project
npx add-skill https://github.com/WenJunDuan/Rlues/tree/main/vibeCoding/old version/config-agent_v8.3/claude/.claude/skills/agent-teams
SKILL.md
Agent Teams — Path D 并行分工
仅 Path D 使用。前置条件:
CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1
角色模板
标准四人组
Researcher:
focus: "调研 + 技术验证"
tools: [augment-context-engine, mcp-deepwiki, web_search]
output: ".ai_state/research-{topic}.md"
Architect:
focus: "方案设计 + 接口定义"
tools: [augment-context-engine]
output: ".ai_state/design.md"
depends_on: [Researcher]
Implementer:
focus: "编码 + 测试"
tools: [augment-context-engine, context7]
output: "src/ + tests/"
depends_on: [Architect]
Reviewer:
focus: "审查 + 安全 + 经验沉淀"
tools: [code-review, security-guidance]
output: ".ai_state/review.md + .knowledge/"
depends_on: [Implementer]
并行开发组 (大型任务)
Lead:
role: "协调, 不写代码"
分配: "拆解 plan.md 任务 → 分配给 Worker"
监控: "inbox 收汇报, 处理冲突"
Worker_1..N:
role: "独立实现分配的任务"
约束: "只改自己负责的文件"
汇报: "完成后 inbox → Lead"
上下文传递协议
Agent Teams 各 agent 有独立 context。
共享状态通过 .ai_state/ 文件系统:
Lead 写入 → .ai_state/plan.md (任务分配)
Worker 读取 → .ai_state/plan.md (自己的任务)
Worker 写入 → .ai_state/worker-{N}-progress.md
Lead 读取 → .ai_state/worker-*-progress.md (汇总)
启动命令 (Claude Code)
# 在 CLAUDE.md 或交互中触发
"用 Agent Teams 模式开发这个功能"
→ PACE 判定 Path D
→ 读本 Skill
→ Lead 创建任务分配
→ Task tool 启动 Worker subagents
Codex CLI 替代
Codex 无原生 Agent Teams。替代方案:
codex --mode plan拆解任务- 多终端各跑
codex实例 - 手动协调 (通过 .ai_state/ 文件)
- 或: MCP server 模式 + Agents SDK hand-off
Recommended Agent Skills
Expand your agent's capabilities with these related and highly-rated skills.
WenJunDuan/Rlues
agent-teams
Path D 并行编排 — 子代理分工 + Agent Teams 协作。需要 CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1
WenJunDuan/Rlues
tdd
Path 分级 TDD 策略
WenJunDuan/Rlues
context7
E 阶段按需拉取库文档 — 用 ctx7 获取准确 API 信息
WenJunDuan/Rlues
knowledge
跨会话经验管理 — 在 .ai_state 内沉淀和复用项目经验
WenJunDuan/Rlues
code-quality
Rev 阶段 Plugin 编排顺序
WenJunDuan/Rlues
verification
V 阶段 Path 分级验证清单
Didn't find tool you were looking for?