Nelson
Aspegio/nelson
Orchestrates multi-agent task execution using a Royal Navy squadron metaphor — from mission planning through parallel work coordination to stand-down.
Cross-runtime playbook for multi-agent collaboration, agent teams, swarm orchestration, parallel task distribution, capability discovery, and quality gates on Claude Code, Codex, OpenClaw, and Cursor.
Install Agent Teams Playbook
$ npx skills add KimYx0207/Kim_Service --skill agent-teams-playbook -a claude-codeProject install by default; add -g for ~/.claude/skills/.
$ gh skill install KimYx0207/Kim_Service agent-teams-playbook --agent claude-codeProject scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).
$ git clone --depth 1 https://github.com/KimYx0207/Kim_Service.git skills-src && mkdir -p .claude/skills && cp -r skills-src/skills/agent-teams-playbook .claude/skills/agent-teams-playbook && rm -rf skills-srcUse ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.
Claude Code skills documentation · loads skills from .claude/skills/
Install the "agent-teams-playbook" agent skill from https://github.com/KimYx0207/Kim_Service/tree/main/skills/agent-teams-playbook into .claude/skills/agent-teams-playbook/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "agent-teams-playbook", then confirm the skill loads.Claude Code copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$skill-installer install https://github.com/KimYx0207/Kim_Service/tree/main/skills/agent-teams-playbookType this inside Codex. $skill-installer <name> installs a curated skill from openai/skills. The installer writes to $CODEX_HOME/skills (default ~/.codex/skills). Restart Codex if the skill does not show up.
$ npx skills add KimYx0207/Kim_Service --skill agent-teams-playbook -a codexProject install goes to .agents/skills/; add -g for ~/.codex/skills/.
$ gh skill install KimYx0207/Kim_Service agent-teams-playbook --agent codexProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/KimYx0207/Kim_Service.git skills-src && mkdir -p .agents/skills && cp -r skills-src/skills/agent-teams-playbook .agents/skills/agent-teams-playbook && rm -rf skills-srcUse ~/.agents/skills/ instead of .agents/skills for a personal install.
Codex skills documentation · loads skills from .agents/skills/
Install the "agent-teams-playbook" agent skill from https://github.com/KimYx0207/Kim_Service/tree/main/skills/agent-teams-playbook into .agents/skills/agent-teams-playbook/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "agent-teams-playbook", then confirm the skill loads.Codex copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add KimYx0207/Kim_Service --skill agent-teams-playbook -a cursorProject install goes to .agents/skills/; add -g for ~/.cursor/skills/.
$ gh skill install KimYx0207/Kim_Service agent-teams-playbook --agent cursorProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/KimYx0207/Kim_Service.git skills-src && mkdir -p .cursor/skills && cp -r skills-src/skills/agent-teams-playbook .cursor/skills/agent-teams-playbook && rm -rf skills-srcUse ~/.cursor/skills/ instead of .cursor/skills for a personal install.
Cursor skills documentation · loads skills from .cursor/skills/, .agents/skills/, .claude/skills/, .codex/skills/
Install the "agent-teams-playbook" agent skill from https://github.com/KimYx0207/Kim_Service/tree/main/skills/agent-teams-playbook into .cursor/skills/agent-teams-playbook/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "agent-teams-playbook", then confirm the skill loads.Cursor copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ gemini skills install https://github.com/KimYx0207/Kim_Service.git --path skills/agent-teams-playbook--scope user (default) or --scope workspace; --path is the subfolder of the repo that holds the skill; --consent skips the security confirmation prompt.
$ npx skills add KimYx0207/Kim_Service --skill agent-teams-playbook -a gemini-cliProject install goes to .agents/skills/; add -g for ~/.gemini/skills/.
$ gh skill install KimYx0207/Kim_Service agent-teams-playbook --agent gemini-cliProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/KimYx0207/Kim_Service.git skills-src && mkdir -p .gemini/skills && cp -r skills-src/skills/agent-teams-playbook .gemini/skills/agent-teams-playbook && rm -rf skills-srcUse ~/.gemini/skills/ instead of .gemini/skills for a personal install, then run /skills reload.
Gemini CLI skills documentation · loads skills from .gemini/skills/, .agents/skills/
Install the "agent-teams-playbook" agent skill from https://github.com/KimYx0207/Kim_Service/tree/main/skills/agent-teams-playbook into .gemini/skills/agent-teams-playbook/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "agent-teams-playbook", then confirm the skill loads.Gemini CLI copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ gh skill install KimYx0207/Kim_Service agent-teams-playbookInstalls for Copilot at project scope by default; add --scope user for a personal install. Preview a skill first with gh skill preview. Needs GitHub CLI 2.90.0 or later (public preview).
$ npx skills add KimYx0207/Kim_Service --skill agent-teams-playbook -a github-copilotProject install goes to .agents/skills/; add -g for ~/.copilot/skills/.
$ git clone --depth 1 https://github.com/KimYx0207/Kim_Service.git skills-src && mkdir -p .github/skills && cp -r skills-src/skills/agent-teams-playbook .github/skills/agent-teams-playbook && rm -rf skills-srcUse ~/.copilot/skills/ instead of .github/skills for a personal install. Commit .github/skills so cloud agent and code review can use it.
GitHub Copilot skills documentation · loads skills from .github/skills/, .claude/skills/, .agents/skills/
Install the "agent-teams-playbook" agent skill from https://github.com/KimYx0207/Kim_Service/tree/main/skills/agent-teams-playbook into .github/skills/agent-teams-playbook/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "agent-teams-playbook", then confirm the skill loads.GitHub Copilot copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add KimYx0207/Kim_Service --skill agent-teams-playbook -a opencodeOpenCode documents no install command of its own. Project install goes to .agents/skills/; add -g for ~/.config/opencode/skills/.
$ gh skill install KimYx0207/Kim_Service agent-teams-playbook --agent opencodeProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/KimYx0207/Kim_Service.git skills-src && mkdir -p .opencode/skills && cp -r skills-src/skills/agent-teams-playbook .opencode/skills/agent-teams-playbook && rm -rf skills-srcUse ~/.config/opencode/skills/ instead of .opencode/skills for a personal install.
OpenCode skills documentation · loads skills from .opencode/skills/, .claude/skills/, .agents/skills/
Install the "agent-teams-playbook" agent skill from https://github.com/KimYx0207/Kim_Service/tree/main/skills/agent-teams-playbook into .opencode/skills/agent-teams-playbook/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "agent-teams-playbook", then confirm the skill loads.OpenCode copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
Facts
- Skill name
agent-teams-playbook- GitHub stars
- 174
- Token cost
- ~2.7k tokens
- SKILL.md length
- 700 words
- Files
- 16 (incl. scripts)
- Skills in repo
- 8
- Repo updated
- First seen
- Licence
- MIT
At a glance
Cross-runtime playbook for multi-agent collaboration, agent teams, swarm orchestration, parallel task distribution, capability discovery, and quality gates on Claude Code, Codex, OpenClaw, and Cursor.
- Works in 5 steps: 写计划时使用抽象动作名；执行时使用当前平台真实工具名。 → Claude Code 使用宿主当前暴露的 Agent /… → Codex 以当前宿主实际暴露的工具 schema 和权限为准，再选择匹配的… → …
- 多agent、agent协作、agent编排、并行agent、分工协作、拉团队、多代理协作
- SKILL.md covers 跨平台兼容层（新增，不替代原流程）, Codex 宿主 profiles（薄适配）, 适用 vs 不适用 and 用户可见性铁律, plus 6 more sections
- Runs Shell, JavaScript and Python scripts from its folder
What it does
Agent Teams Playbook is an agent skill from KimYx0207/Kim_Service. Cross-runtime playbook for multi-agent collaboration, agent teams, swarm orchestration, parallel task distribution, capability discovery, and quality gates on Claude Code, Codex, OpenClaw, and Cursor. Use for 多agent、agent协作、agent编排、并行agent、分工协作、拉团队、多代理协作 or complex work needing independent roles and review. Map the six-stage workflow to the active host's actual capabilities; report sequential degradation when parallel agents are unavailable.
Its SKILL.md is about 2.7k tokens, which your agent loads only when the skill is triggered. The skill folder holds 19 other files, including scripts (for example `CHANGELOG.md`, `README.md` and `README_EN.md`).
It sits in Agent Workflows, covering Subagents, Quality gates and Multi-agent orchestration. The repository describes itself as: 面向 Claude Code、Codex 等 AI 编码助手的 Hook 与 Agent Skill 开源合集。 The licence is MIT.
When your agent uses it
- 多agent、agent协作、agent编排、并行agent、分工协作、拉团队、多代理协作
- Complex work needing independent roles and review
Example prompts
- “/agent-teams-playbook”
Requirements
- Python 3
- Node.js
- A Bash shell
Workflow steps
5 steps, taken from the first numbered list in SKILL.md.
- 写计划时使用抽象动作名；执行时使用当前平台真实工具名。
- Claude Code 使用宿主当前暴露的 Agent / Task；只有宿主确实暴露 TeamCreate / SendMessage 时才承诺共享团队语义。不要把 Codex 参数复制到 Claude Code。
- Codex 以当前宿主实际暴露的工具 schema 和权限为准，再选择匹配的 profile；不能只凭产品名或版本猜参数。只有工具调用成功且有返回的 Agent 标识，才报告已启动。
- OpenClaw/Cursor 的 team 能力可能来自项目插件、workspace 或 IDE 能力；先探测，再承诺。
- 若平台不支持真正并行或团队通信，明确降级为"主线程分阶段执行"，但仍执行阶段0-5的治理流程。
What it can do on your machine
Read from SKILL.md and the folder at commit e388fd5. It shows what the files ask for, not the result of running them.
Tool permissions
Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.
From allowed-tools in the SKILL.md frontmatter.
Runs code
Ships 1 file in scripts/ (Shell, JavaScript and Python), which the agent can run.
From the folder's file list and the shell code blocks in SKILL.md.
Network
Links to these hosts (documentation or services it may open):
github.comagentskills.io
From URLs in SKILL.md, links to its own repository left out.
Credentials
Names no API keys, tokens, secrets or passwords.
From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.
Context cost
Agent Teams Playbook loads about 2.7k tokens when it runs. Until then it costs about 117 tokens; SKILL.md has 700 words of instructions outside code blocks.
Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.
Safety
Auto-check passed
The automated check found no risky patterns in SKILL.md.
Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); the scripts in this folder are not scanned.
SKILL.md
The full file from KimYx0207/Kim_Service at commit e388fd5, republished under its MIT licence (© KimYx0207). 700 words, ~2,667 tokens.
Download SKILL.mdSave it as
.claude/skills/agent-teams-playbook/SKILL.md (or your agent's skills folder). This skill also uses 15 other files; get the full folder from GitHub.- name
- agent-teams-playbook
- description
- Cross-runtime playbook for multi-agent collaboration, agent teams, swarm orchestration, parallel task distribution, capability discovery, and quality gates on Claude Code, Codex, OpenClaw, and Cursor. Use for 多agent、agent协作、agent编排、并行agent、分工协作、拉团队、多代理协作 or complex work needing independent roles and review. Map the six-stage workflow to the active host's actual capabilities; report sequential degradation when parallel agents are unavailable.
- metadata.version
- 4.8.0
Agent Teams 编排手册
本 Skill 为调用方提供多 Agent 分工、执行和质量检查建议。宿主协调器负责角色边界、过程与最终交付；在 Meta_Kim 中，Router、跨组件状态机和最终验收仍由 Meta_Kim 负责，本包不接管这些职责。
核心理解（铁律）：Agent Teams 是"并行处理 + 结果汇总"模式，不是扩大单个 agent 的上下文窗口。每个 teammate 是独立的执行单元，拥有独立上下文，可以并行处理大量信息，但最终需要将结果汇总压缩后返回主会话。
跨平台兼容层（新增，不替代原流程）
本 Skill 的完整 6 阶段工作流、按需能力解析、Agent → Skill 委派模式、质量把关和故障处理规则仍然保留。四个平台的差异只影响"用什么工具执行"，不影响"是否执行这些阶段"。
先识别当前平台，再把下列抽象动作映射到平台原生能力。不要在非 Claude Code 平台原样承诺 Task、TeamCreate、SendMessage 或 Skill(...) 一定存在；也不要因为平台工具名不同而跳过阶段0-5。
| 抽象动作 | Claude Code | Codex | OpenClaw | Cursor |
|---|---|---|---|---|
| 调用 Skill | Skill(skill="name", args="...") 或 slash skill | 读取并遵循本地 skill 指令；只有宿主暴露 skill 工具时才称为"调用" | 读取并遵循 openclaw/skills 或全局 skill；按 OpenClaw 当前工具执行 | 读取并遵循 .cursor/skills 或全局 skill；按 Cursor 当前 agent 能力执行 |
| 启动独立 Subagent | 使用宿主当前暴露的 Agent / Task，至少携带必填 prompt，并按当前 schema 提供 subagent_type、description/name 和边界 | 使用当前宿主暴露的原生 spawn schema（见下方 profiles） | workspace / agent 调度能力；不可用则主线程分阶段执行 | background agent / agent mode；不可用则主线程分阶段执行 |
| 组建 Agent Team | 宿主暴露时使用 TeamCreate + Agent / Task(team_name) | 同一轮并发调用多个宿主原生 spawn 调用，由主线程汇总；不伪装共享团队总线 | team / workspace 能力存在时使用；否则多个任务或主线程 | background agents / team-like workflow 存在时使用；否则多个任务或主线程 |
| 成员通信/进度 | SendMessage 或 Agent/Task result | 子任务结果回报；只有宿主暴露 agent I/O 时才中途通信 | 平台消息/日志；不可用时用阶段性文本汇报 | IDE/agent 日志；不可用时用阶段性文本汇报 |
| 规划文件 | planning-with-files skill | 若本地 skill/tool 存在则使用；否则使用内联计划或平台计划工具 | 若本地 planning skill 存在则使用；否则维护可见计划记录 | 若本地 planning skill 存在则使用；否则维护可见计划记录 |
平台适配底线：
- 写计划时使用抽象动作名；执行时使用当前平台真实工具名。
- Claude Code 使用宿主当前暴露的
Agent/Task；只有宿主确实暴露TeamCreate/SendMessage时才承诺共享团队语义。不要把 Codex 参数复制到 Claude Code。 - Codex 以当前宿主实际暴露的工具 schema 和权限为准，再选择匹配的 profile；不能只凭产品名或版本猜参数。只有工具调用成功且有返回的 Agent 标识，才报告已启动。
- OpenClaw/Cursor 的 team 能力可能来自项目插件、workspace 或 IDE 能力；先探测，再承诺。
- 若平台不支持真正并行或团队通信，明确降级为"主线程分阶段执行"，但仍执行阶段0-5的治理流程。
Codex 宿主 profiles（薄适配）
当前宿主 schema 优先于以下已知示例。先读取工具名、必填字段、可选字段、返回标识、并发与通信权限；只提交当前 schema 允许的字段。宿主未暴露工具时不要猜名字，也不要为适配自行实现执行引擎。agent_type、model、reasoning_effort 可能受 feature flags 隐藏；即便暴露，也要遵循宿主的选择/继承规则，不自动覆盖。
- codex-v1：已知上游 namespace 为
multi_agent_v1.spawn_agent，任务用message或items，历史继承用布尔fork_context；agent_type仅在当前 schema 暴露并允许选择时可用。 - codex-v2：已知上游顶层
spawn_agent必填task_name和message，历史继承用字符串fork_turns（none、all或正整数字符串）；不能向它传 V1 的fork_context。 - codex-v2-limited：工具名和必填字段同 V2，但宿主可隐藏
agent_type、model、reasoning_effort；不能照搬其他 profile 的可选字段。 - no-capability：没有可用 spawn，或权限不允许并发时，明确采用主线程分阶段执行，仍保留阶段0-5；不得报告真实并行、团队通信或子 Agent 成功。通信工具也必须单独探测。
以下最小调用样例只展示静态协议，不承诺真实宿主已实测。测试夹具验证文档与已知 schema 的一致性；新的宿主 schema 有差异时，以现场发现为准并记录差异。
<!-- contract: codex-v1 -->{"tool":"multi_agent_v1.spawn_agent","arguments":{"message":"Review the bounded task and return evidence.","fork_context":false}}<!-- contract: codex-v2 -->{"tool":"spawn_agent","arguments":{"task_name":"review_task","message":"Review the bounded task and return evidence.","fork_turns":"none"}}<!-- contract: codex-v2-limited -->{"tool":"spawn_agent","arguments":{"task_name":"review_task","message":"Review the bounded task and return evidence."}}<!-- contract: no-capability -->{"tool":null,"arguments":null,"mode":"sequential","parallel":false}参考：OpenAI Codex schema 快照（2026-10-02 核验，V1/V2 与字段开关并存）；Agent Skills 格式规范（自定义版本放在 metadata.version，包事实以 capability.json 为准）。
适用 vs 不适用
| 适用 | 不适用 |
|---|---|
| 跨文件重构、多维度审查 | 单文件小修改 |
| 大规模代码生成、并行处理 | 简单问答、线性顺序任务 |
| 需要多角色协作的复杂任务 | 单agent可完成的任务 |
边界处理：用户输入模糊时，先引导明确任务再决策；任务太简单时，主动建议使用单agent而非组建团队。
用户可见性铁律
- 每个阶段启动前输出计划，完成后输出结果
- 子agent在后台执行，但进度必须汇报给用户
- 任务拆分计划必须经用户确认后再执行；若宿主平台或项目指令要求直接执行，则说明采用的默认假设
- 失败时立即通知：
❌ [角色名] 失败: [原因]，提供重试/跳过/终止选项 - 全部完成后输出汇总报告（见阶段5格式），并说明真实使用的平台工具和任何降级路径
场景决策树
执行顺序：先执行阶段0和阶段1（强制），再根据任务复杂度选择场景（影响阶段2-5）。
| 问题 | 路径 |
|---|---|
| Q0: 阶段1找到可完整解决任务的现有 provider（Agent / Skill / Tool）？ | 是 → 场景2 / 否 → Q1 |
| Q1: 任务复杂度？ | 简单(1-2步) → 场景1 / 中等(3-5步) → 场景3 / 复杂(6+步) → Q2 |
| Q2: 需要明确团队分工？ | 是 → 场景4 / 否 → 场景5 |
- 用户直接指定场景编号时,跳过决策树直接执行
- 未指定场景时，默认用场景3（计划+评审）
- 注意：阶段0（规划）和阶段1（本地能力发现）是所有场景的前置步骤；外部
find-skills只在本地能力确有缺口时触发
5大编排场景
| # | 场景 | 适用条件 | 核心策略 |
|---|---|---|---|
| 1 | 提示增强 | 简单任务，1-2步 | 优化单agent提示词，不拆分不组队 |
| 2 | Provider直接复用 | 任务可由单个现有 Agent / Skill / Tool 完全解决 | 直接绑定已发现 provider，无需外部搜索或组建Agent Teams |
| 3 | 计划+评审 | 中等/复杂任务（默认） | 出计划 → 用户确认 → 并行执行 → Review验收 |
| 4 | Lead-Member | 需要明确团队分工 | Leader协调分配，Member并行执行，通过TaskList协同 |
| 5 | 复合编排 | 复杂任务，无固定模式 | 动态组合上述场景，按阶段切换策略 |
模型分工（所有场景通用）：通过平台支持的模型选择能力按任务复杂度分配；Claude Code 可通过 Task 工具的 model 参数分配——opus处理复杂推理，haiku处理简单任务，sonnet处理常规任务。平台不支持模型选择时，不要写死模型承诺。
协作模式
| 模式 | 通信方式 | 适用场景 | Claude Code 启动方式 | Codex 启动方式 | OpenClaw / Cursor 启动方式 |
|---|---|---|---|---|---|
| Subagent | 子agent → 主协调器单向汇报 | 并行独立任务 | 宿主当前的 Agent / Task | 当前宿主原生 spawn schema | 平台 agent/background/workspace 能力；不可用则主线程分阶段执行 |
| Agent Team | 成员间可双向通信(SendMessage) | 需要协作的复杂任务 | TeamCreate + Agent / Task(team_name)（仅宿主暴露时） | 多个宿主原生 spawn 调用 + 主线程协调；仅在宿主暴露 agent I/O 时中途交互 | 平台 team/workspace/background-agent 能力；没有则降级 |
选择原则：任务间无依赖用Subagent（简单高效），任务间需要协调用Agent Team（功能更强但成本更高）。如果当前平台没有真正的 team bus，只能称为"多个独立 subagent + 主线程汇总"，不能伪装成成员间双向协作。
6阶段工作流（含强制规划和按需能力发现）
重要说明：阶段0和阶段1是所有场景的强制前置步骤，场景选择（1-5）只影响阶段2-5的执行方式。
阶段0：规划准备（Planning Setup）【硬性标准 - 所有场景必经】
优先使用当前平台的 Skill 工具调用 planning-with-files：
Skill(skill="planning-with-files")跨平台适配：
- Claude Code：优先使用
Skill(skill="planning-with-files") - Codex：若本地 skill/tool 可用则使用；否则使用平台计划工具或内联计划，并明确说明没有创建 planning files
- OpenClaw/Cursor：若本地 planning skill 可用则使用；否则维护平台可见计划记录
这将在项目目录创建三个核心文件（当 planning-with-files 可用时）：
task_plan.md- 任务计划和阶段追踪findings.md- 研究发现和知识积累progress.md- 执行日志和进度记录
关键规则（规划文件创建后遵循）：
- 每个阶段开始前读取task_plan.md，完成后更新状态
- 每2次搜索/浏览操作后立即保存发现到findings.md
- 所有错误必须记录到task_plan.md的"Errors Encountered"表格
- 3次失败后升级给用户
铁律：复杂任务不能没有计划就开始执行。若平台没有 planning-with-files，也必须使用等价计划记录或内联计划，不能跳过规划阶段。
Show full SKILL.md (289 more words)Show less
阶段1：任务分析 + 能力发现（Discovery）【硬性标准 - 所有场景必经】
先质疑再执行：
- 需求不合理时主动挑战假设，建议更好的方案
- 区分"现在必须做"和"以后再说"，排除非核心范围
- 任务太大时建议更聪明的起点
- 先描述所需能力，再匹配 agent / skill / tool，避免 name-first 调度
输出任务总览：
| 字段 | 内容 |
|---|---|
| 任务目标 | [一句话描述] |
| 预期结果 | [具体交付物] |
| 验收标准 | [可量化的通过条件] |
| 范围界定 | [must-have vs add-later] |
| 当前平台 | [Claude Code / Codex / OpenClaw / Cursor / Unknown] |
| 预计Agent数 | [按任务 DAG、冲突边界与宿主并发能力确定，不写死数量] |
| 选定场景 | [场景编号+名称] |
| 协作模式 | [Subagent/Agent Team/Degraded] |
能力解析链（命中即停止，不是每次必跑的回退仪式）：
- 本地多类型能力扫描：读取当前运行时可见的 Agents、Skills、Tools、Commands、MCP 和 capability index。只要已有专业 provider 覆盖子任务，就绑定该 provider 并停止搜索。
- 外部能力搜索（仅真实缺口时）：只有本地所有相关 provider 都无法覆盖，而且该能力值得复用或安装时，才调用
find-skills或平台等价搜索。安装仍需用户/宿主授权。 - 运行时原生绑定：Claude Code 将选中的 owner 绑定到
Agent/Task；Codex 将它写入 boundedmessage，再按当前宿主原生 spawn schema 调用。 - 最后降级：只有宿主 Agent surface 缺失、权限阻断，或完整发现后仍无可用 owner，才允许进入显式 degraded mode。通用临时 Subagent 仅在项目策略允许时使用；主线程分阶段执行只用于明确降级，不能冒充正常编排。
铁律：找不到“完全匹配的 Skill”不等于能力失败。已有 Agent、Tool、Command 或 MCP 能完成任务时，不得继续外部搜索，也不得称为回退。
find-skills没有安装结果时，也不能把随后成功的原生 Agent 调用标成 fallback。
阶段2：团队组建
输出团队蓝图：
| 编号 | 角色 | 职责 | 模型 | ownerAgent / subagent_type | Capability Binding | 平台工具/降级 |
|---|---|---|---|---|---|---|
| 1 | [角色名] | [具体职责] | [opus/sonnet/haiku/平台默认] | [已发现 owner] | [Agent / Skill / Tool / Command / MCP] | [Claude Agent/Task、Codex spawn_agent、其他原生表面或显式 degraded] |
说明：最后两列标注已匹配的专业 provider/capability，以及当前平台真实使用的工具或有证据的降级路径。
阶段3：并行执行
- Skill任务：用当前平台的 Skill 调用机制调用本地已安装的skill；Claude Code 示例：
Skill(skill="skill-name", args="任务描述") - Agent任务：使用当前平台原生 subagent 工具；Claude Code 使用宿主暴露的
Agent/Task，Codex 使用当前宿主暴露的原生 spawn schema（见下方 profiles） - 混合编排时skill和subagent可并行运行；平台不支持并行时按阶段顺序执行并说明降级
- 每个agent/skill完成后汇报：
✅ [角色名] 完成: [一句话结果] - 遇到问题时给用户选项，而不是自己默默选一个
Agent → Skill 委派（子agent调用skill的3种模式）：
选中的专业 owner 在 Claude Code 中能否继续调用 Skill，取决于当前 Agent/Task schema 和工具权限；其他平台同样必须先确认宿主是否给子 Agent 开放等价 skill/tool 权限。不要为了获得 Skill 工具而把已有专业 owner 换成 general-purpose。
| 模式 | 流程 | 适用场景 |
|---|---|---|
| 协调器直调 | 协调器 → Skill(skill="name") 或平台等价调用 → 结果 | 单步Skill任务，无需并行 |
| 委派式调用 | 协调器 → subagent prompt="请使用 /skill-name 完成 X" → subagent → Skill/tool 或内联执行 → 汇报 | 并行多个Skill，或Skill耗时较长 |
| 团队成员调用 | team/subagents → 分配任务 → member → Skill/tool 或内联执行 → 汇报 | 需要成员间协调的复杂任务 |
委派式调用关键点：Task/subagent prompt 中写明要调用的Skill名称和参数；只有平台支持 skill 工具时才承诺自动调用，否则按 Skill 文档内联执行并说明。
阶段4：质量把关 & 产品打磨
验收检查：对照阶段1的验收标准逐项检查。
产品打磨（不仅功能完整，更要用户体验优秀）：
- 边界处理：异常输入、空值、极端情况是否覆盖
- 专业度：命名规范、代码风格、错误提示是否友好
- 完整性：文档、配置说明、使用示例是否齐全
- 平台诚实性：是否准确说明了真实调用的工具和降级行为
全部通过 → 进入阶段5。不通过 → 打回修改，最多2轮，仍不通过则通知用户人工介入。
阶段5：结果交付 & 部署移交
输出执行报告：
| 项目 | 内容 |
|---|---|
| 总任务数 | X个，成功Y个，失败Z个 |
| 实际平台工具 | 使用了哪些真实工具 / 是否降级 |
| 各Agent结果 | [角色]: [状态] - [关键产出] |
| 汇总结论 | [综合所有结果的最终结论] |
| 后续建议 | [当前未覆盖但值得做的改进方向] |
部署移交（按需提供）：
- 运行方式：启动命令、环境要求、配置说明
- 验证步骤：用户可自行验证的操作清单
- 已知限制：当前版本的边界和约束
执行底线
【硬性标准】： 0. 强制使用 planning-with-files 或平台等价计划机制：任何复杂任务必须先创建/维护 task_plan.md、findings.md、progress.md，或显式说明使用了平台等价计划记录
- 强制执行能力命中即停止：本地多类型 provider 命中后直接使用；仅真实能力缺口才外部搜索；仅真实宿主/权限/owner 缺口才显式降级
【其他原则】：
2. 先目标，后组织结构——任务不清晰时先澄清，再决定是否组建团队
3. 队伍规模由任务 DAG、冲突边界与宿主并发能力决定，不写死统一数量上限
4. 关键里程碑必须有质量闸门和回滚点
5. 不默认任何外部工具可用；本地多类型能力先验证，find-skills 仅在真实缺口时验证/调用
6. 浏览器多窗口默认互相独立，不共享上下文
7. 成本只是约束，不是固定承诺——不做不切实际的成本预估
8. 危险操作、大规模变更必须先获得用户确认或遵守宿主平台审批规则
9. 不承诺平台没有的能力；尤其不要在 Codex/OpenClaw/Cursor 中直接承诺 Claude Code 的 TeamCreate
故障处理
| 故障类型 | 处理策略 |
|---|---|
| Agent执行失败 | 通知用户，提供重试/跳过/终止/降级选项 |
| Skill不可用 | 先检查现有 Agent/Tool/Command/MCP 是否已覆盖；只有真实能力缺口才外部搜索，只有真实执行表面缺失才显式降级 |
| 平台工具缺失 | 改用该平台可用工具，并明确说明降级路径 |
| 模型超时 | 调整任务复杂度或拆分为更小的子任务 |
| 质量不达标 | 打回修改最多2轮，仍不通过则人工介入 |
| 上下文溢出 | 拆分为更小的子任务，分批执行 |
© KimYx0207, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file
Files
SKILL.md and 15 other files (scripts) in skills/agent-teams-playbook of KimYx0207/Kim_Service.
- SKILL.md
- .gitattributes
- .gitignore
- CHANGELOG.md
- LICENSE
- NOTICE
- README.md
- README_EN.md
- capability.json
- images/二维码基础款.png
- images/微信.jpg
- images/支付宝.jpg
- scripts/install.sh
- tests/fixtures/codex-host-profiles.json
- tests/runtime-contracts.test.mjs
- tests/test_installer.py
Open the folder on GitHubat commit e388fd5
Compare with similar skills
Agent Teams Playbook next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.
| Skill | Stars | Used in | Tokens | Auto-check | Licence | Repo updated |
|---|---|---|---|---|---|---|
| Agent Teams Playbook this skillKimYx0207/Kim_Service | 174 | — | ~2.7k | Automated safety check: Pass | MIT | |
| NelsonAspegio/nelson | 422 | — | ~11k | Automated safety check: Warn | MIT | |
| Team Coordinationbybren-llc/safe-agentic-workflow | 423 | — | ~1.7k | Automated safety check: Pass | MIT | |
| Team Coordinationbybren-llc/safe-agentic-workflow | 423 | — | ~1.9k | Automated safety check: Notes | MIT | |
| Paseo Advisor Second Opiniongetpaseo/paseo | 20k | 1 repos | ~756 | Automated safety check: Pass | Custom licence | |
| O2 Review Loopopenobserve/openobserve | 22k | — | ~3.7k | Automated safety check: Pass | AGPL-3.0 |
Similar skills
Team Coordination
bybren-llc/safe-agentic-workflow
Agent Teams orchestration patterns for multi-agent SAFe workflows.
Team Coordination
bybren-llc/safe-agentic-workflow
Agent Teams orchestration patterns for multi-agent SAFe workflows.
Paseo Advisor Second Opinion
getpaseo/paseo
Launches one separate agent through Paseo to give a second opinion on the current task, with a self-contained briefing and no permission to edit files.
O2 Review Loop
openobserve/openobserve
Splits a change into planner, coder and independent reviewer roles: you confirm a spec, a subagent implements it, and a separate reviewer checks each round's local WIP commit.
Harness Agent Team Designer
revfactory/harness
Designs a project-specific agent harness: defines specialist agents, writes the skills they follow, picks an execution mode and model for each, and keeps the setup maintained.
More from KimYx0207/Kim_Service
All 8 skills in this repoMeta Skill Creator
KimYx0207/Kim_Service
创建、重构或验收可复用技能包。适用于把重复工作流做成跨宿主能力包，并明确触发规则、第一动作、渐进加载、资产模板、脚本校验、触发评测、基线对比、验收证据、闭环治理、公开交付边界和不可伪造的运行证明；也适用于清理冗余参考、拆分人读模板与机器结构数据、补齐首次公开前的验收记录、记录运行反馈、生成写回或不写回决定。当用户提到"做一个 skill / 改 skill / 优化 skill / 评审…
Xiaohongshu Skill
KimYx0207/Kim_Service
生成完整的小红书/Rednote 图文发布包：先自动研究内容机会，再完成标题、正文、封面、6-8 页内页、逐页视觉导演表、Image2 优先图片路线、封面 MVP 确认、批量出图和发布前自检。适用于课程、服务、产品、个人 IP、本地商家和知识分享；单点标题、封面字或改写不触发。
Memory 3layer
KimYx0207/Kim_Service
为长周期、跨会话的 Agent 工作建立平台中立的三层记忆。适用于加载项目记忆、记录可复用事实与每日进展、维护隐性知识、迁移旧版 .claude/memory、检查记忆状态或清理过时条目；Claude Code 与 Codex 可通过各自 Hooks 自动接线，其他 Agent Skills 宿主只使用手动核心。不要用它保存秘密、完整聊天记录、一次性日志或未经确认的推测。
Find Skill
KimYx0207/Kim_Service
Helps users discover and install agent skills when they ask questions like "how do I do X", "find a skill for X", "is there a skill that can...", or express interest in extending capabilities.
Kim Decision
KimYx0207/Kim_Service
A skill your agent uses when the user asks for KIM, Kim, laojin, 老金, 问问老金, 老金怎么看, asks for decision analysis, structured reasoning, product/business/content review, PRD, MVP, user path, growth…
Semgrep Skill
KimYx0207/Kim_Service
Runs the installed local Semgrep CLI through a bounded JSON wrapper with two bundled non-secret rules.
Related
Categories
Questions about Agent Teams Playbook
What does Agent Teams Playbook do?
Cross-runtime playbook for multi-agent collaboration, agent teams, swarm orchestration, parallel task distribution, capability discovery, and quality gates on Claude Code, Codex, OpenClaw, and Cursor. Agent Teams Playbook is an agent skill from KimYx0207/Kim_Service. Cross-runtime playbook for multi-agent collaboration, agent teams, swarm orchestration, parallel task distribution, capability discovery, and quality gates on Claude Code, Codex, OpenClaw, and Cursor.
When should I use Agent Teams Playbook?
Agent Teams Playbook fits situations like: 多agent、agent协作、agent编排、并行agent、分工协作、拉团队、多代理协作; complex work needing independent roles and review.
How do I install Agent Teams Playbook in Claude Code?
Run `npx skills add KimYx0207/Kim_Service --skill agent-teams-playbook -a claude-code`. Or copy the skill folder (skills/agent-teams-playbook in KimYx0207/Kim_Service) into .claude/skills/agent-teams-playbook in your project. Claude Code loads it when a task matches its description.
How do I install Agent Teams Playbook in Codex?
Run `npx skills add KimYx0207/Kim_Service --skill agent-teams-playbook -a codex`. Or copy the skill folder (skills/agent-teams-playbook in KimYx0207/Kim_Service) into .agents/skills/agent-teams-playbook in your project. Codex loads it when a task matches its description.
Can I use Agent Teams Playbook in Cursor, Gemini CLI or GitHub Copilot?
Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add KimYx0207/Kim_Service --skill agent-teams-playbook -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/agent-teams-playbook, .gemini/skills/agent-teams-playbook, .github/skills/agent-teams-playbook and .opencode/skills/agent-teams-playbook in your project.
What does Agent Teams Playbook need to run?
Going by SKILL.md and its folder, Agent Teams Playbook needs a shell, JavaScript and Python for the scripts in its folder. Our summary lists: Python 3; Node.js; A Bash shell.
Does Agent Teams Playbook access the network?
SKILL.md names 2 domains. As links in the text: github.com and agentskills.io. This is read from the text; nothing was executed.
Is Agent Teams Playbook safe to install?
Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. The check reads SKILL.md only: the scripts in the folder are not scanned, so read them before running anything.
What licence does Agent Teams Playbook use?
Agent Teams Playbook is published under the MIT licence (from the LICENSE file in the skill folder). It allows redistribution, so the full SKILL.md is shown on this page.
How many tokens does Agent Teams Playbook use?
About 2.7k tokens (SKILL.md is roughly 11k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.
What are the alternatives to Agent Teams Playbook?
Skills that share tags, products or a category with Agent Teams Playbook: Nelson (Aspegio/nelson, 422 stars), Team Coordination (bybren-llc/safe-agentic-workflow, 423 stars), Team Coordination (bybren-llc/safe-agentic-workflow, 423 stars) and Paseo Advisor Second Opinion (getpaseo/paseo, 20k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.
Who maintains Agent Teams Playbook?
KimYx0207 (a GitHub user) maintains it in KimYx0207/Kim_Service, which has 174 GitHub stars. The repository holds 8 skills in this directory. The repository was last updated on October 10, 2026.
Source: KimYx0207/Kim_Service on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.