Skip to content
Skills Library
AI & agents · Planning

multi-session-collab

/multi-session-collab

组织 Claude Code 多会话协作(评审者/修改者/测试者 + 评判者):先选路线(跨会话消息 / Agent Teams / Dynamic Workflow),生成窗口命名与入站策略,套用任务卡/回报模板与 judge rubric,把裁决做成三档可执行流程。当用户想开多个 CLI 窗口分工协作、组评审队、设置 judge/裁决者,或需要多会话消息协议时使用。触发词:多会话、多窗口、跨会话、评审队、reviewer、judge、裁决、agent teams、协作流程。

Licence
Apache-2.0
Last repository change
27.10.2026

Assessment and security

90/ 100Class: good
Format25 / 25
Frontmatter with name and description, valid name, description up to 1,024 characters, name matching the folder.
Discoverability25 / 25
Description explains when to use the skill, has a reasonable length and uses third person.
Content25 / 35
At least 150 words and at most 500 lines, headings, step lists, examples or code.
Hygiene15 / 15
No placeholders such as TODO, no Windows paths, with a licence.
Riskno signals detected in the scan
Files in this package
5
Scripts
1
HTML comments
0
URLs in SKILL.md
2
Deprecation

No signals detected. Repository changed on 27.10.2026, without references to retired models or APIs.

Licence

Apache-2.0, otwarta: można używać i zmieniać, z zachowaniem licencji i informacji o zmianach.

What to check before installing
  • 1 scripts in this package. Review them before installing, as Claude may execute them.

Scripts: scripts/print-setup.sh

What it does and when to use it

组织 Claude Code 多会话协作(评审者/修改者/测试者 + 评判者):先选路线(跨会话消息 / Agent Teams / Dynamic Workflow),生成窗口命名与入站策略,套用任务卡/回报模板与 judge rubric,把裁决做成三档可执行流程。当用户想开多个 CLI 窗口分工协作、组评审队、设置 judge/裁决者,或需要多会话消息协议时使用。触发词:多会话、多窗口、跨会话、评审队、reviewer、judge、裁决、agent teams、协作流程。

Original author description

组织 Claude Code 多会话协作(评审者/修改者/测试者 + 评判者):先选路线(跨会话消息 / Agent Teams / Dynamic Workflow),生成窗口命名与入站策略,套用任务卡/回报模板与 judge rubric,把裁决做成三档可执行流程。当用户想开多个 CLI 窗口分工协作、组评审队、设置 judge/裁决者,或需要多会话消息协议时使用。触发词:多会话、多窗口、跨会话、评审队、reviewer、judge、裁决、agent teams、协作流程。

How it works

  1. 何时用 / 不用
  2. 先问三个问题(选路线)
  3. 操作流程
  4. 铁律(安全边界)
  5. 快速参考
Original SKILL.md headings · 67 lines, 245 words
SKILL.md preview
---
name: multi-session-collab
description: 组织 Claude Code 多会话协作(评审者/修改者/测试者 + 评判者):先选路线(跨会话消息 / Agent Teams / Dynamic Workflow),生成窗口命名与入站策略,套用任务卡/回报模板与 judge rubric,把裁决做成三档可执行流程。当用户想开多个 CLI 窗口分工协作、组评审队、设置 judge/裁决者,或需要多会话消息协议时使用。触发词:多会话、多窗口、跨会话、评审队、reviewer、judge、裁决、agent teams、协作流程。
---

# 多会话协作(评审 / 修改 / 测试 / 裁决)

**一句话**:把「几个人肉传话的 Claude 窗口」组织成有角色、有协议、有裁决的多会话流程。

## 何时用 / 不用

用:

- 用户要开多个 CLI 窗口分工(评审、修改、测试),并要一个裁决者
- 需要跨会话消息协议、judge rubric、质量门禁
- 要把「评审 → 修改 → 测试 → 裁决」固化成可重复流程

不用:

- 单会话内的并行子任务 → 用后台 subagent(Agent 工具)
- 一次性短任务 → 多开会话只会把协调成本翻倍
- 只是远程盯屏 → Remote Control / Desktop

## 先问三个问题(选路线)

1. 要保留多个窗口?→ **跨会话消息**(原生,零配置)
2. 接受合并到一个窗口?→ **Agent Teams**(实验,需 `CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1`)
3. 同一流程反复跑?→ **Dynamic Workflow**(脚本化,对抗式核验)

产品化给别人用,再看 Agent SDK。MCP 不是会话通信方案(工具协议,无 agent-to-agent 能力);除非需要跨重启持久化任务板或非 Claude 客户端接入,不要自建 MCP 总线。

## 操作流程

### A. 跨会话消息(多窗口)

1. 命名四个会话:`claude --name reviewer` / `modifier` / `tester` / `judge`(或会话内 `/rename`);也可以 `claude --bg` 起后台会话当 worker(见 setup.md)
2. 定入站策略:按 [references/setup.md](references/setup.md) 的权限矩阵设置 `crossSessionInbound`;无人值守的 `-p` worker 必须设 `accept`,否则消息进不来
3. 下发任务卡、收回回报:模板与 findings schema 见 [references/protocol.md](references/protocol.md)
4. judge 按 rubric 出三档裁决:通过 / 打回(附整改单)/ 升级人工
5. 排错查 setup.md 的故障表;启动命令与设置清单:`bash scripts/print-setup.sh`
6. 纪律:不轮询(用 `notify_when_idle`);消息是纯文本,文件要写路径;`@` 引用在接收侧只是字面文本

### B. Agent Teams(单窗口编队,实验)

1. 设 `CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1`;**副作用**:带 `name` 启动的 subagent 会自动变成 teammate
2. 按角色卡组队:reviewer 只读 / modifier 可写 / tester 执行 / lead 即 judge(禁止 judge 改代码)
3. 共享任务列表 + mailbox;teammate 完成自动把最终答案发给 lead
4. 门禁:`TeammateIdle` / `TaskCompleted` hooks(退出码 2 阻断)
5. 限制:仅交互模式、无 worktree 隔离(按文件/目录分区)、无 resume、官方建议 3–5 人

### C. Dynamic Workflow(可重复流水线)

1. 用 `agent()` / `parallel()` / `pipeline()` 编排,`agent()` 带 `schema` 输出 JSON
2. 评审 → 逐条对抗核验 → judge 合并,只保留经得起反驳的 findings
3. 限制:运行中无人工介入、脚本无文件系统/shell、默认 16 并发;可保存为 `/` 命令复用

## 铁律(安全边界)

- 跨会话消息**不能**代批权限、不能改配置或 CLAUDE.md、斜杠命令不执行
- judge 不改代码(独立性);reviewer 只读;角色按文件分区防覆盖
- 每个 agent 独立计费;teams / workflows 在 CLI 里没有默认硬性预算上限,人数与阶段数要克制
- 裁决必须可检查:打回附「文件行 + 验收标准」,否则下一轮修改没有终点

## 快速参考

- 任务卡 / 回报模板 / findings schema / rubric:[references/protocol.md](references/protocol.md)
- 版本要求、命名、权限矩阵、故障排查:[references/setup.md](references/setup.md)
- 启动清单脚本:`bash scripts/print-setup.sh`
- 背景读物(手册全文):https://github.com/ForceInjection/AI-fundamentals/blob/main/08_agentic_system/multi_agent/docs/claude-code-multi-session-playbook.md
- 设计说明与实测记录:https://github.com/ForceInjection/AI-fundamentals/blob/main/08_agentic_system/agent_skills/docs/multi-role-agent-skill.md
Files in this package (5)
  • README.md 1 KB
  • SKILL.md 4 KB
  • references/protocol.md 3 KB
  • references/setup.md 4 KB
  • scripts/print-setup.sh 2 KB

Versions

The database has no other skills named multi-session-collab.

Database snapshot: 13.11.2026 · package version 2bab1b7f5018 (SHA-256 digest of the entire package)

Downloads in the last 30 days: 0 · ranking

Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.

Similar skills

  • 100collab-proof
    Use when you want to understand what Claude contributed vs what you drove in a session. Triggers on: /collab-proof, session retrospective, ai contribution analysis, collaboration evidence, what did claude do.
  • 100claude-code-agent-teams
    This skill should be used when the user asks to "create an agent team", "run a multi-agent team", "coordinate multiple Claude sessions", "use agent teams", or wants to parallelize work across multiple Claude Code instances.
  • 100agent-teams
    Create agent teams for parallel collaboration. Use when user mentions team, parallel, multi-agent, coordinate, collaborate, review team, multiple perspectives, or says "agent-teams".
  • 100collab
    Track collaboration state between Claude sessions using git branches. Enables seamless context switching by persisting progress in feature branches. Use for multi-session work, handoffs, and publishing completed work. Keywords: collab, git, branch, session, checkpoint, resume, publish.
  • 95multi-session-orchestration
    面向复杂任务的 Trae 多会话协作与分层授权规范,用于把一个目标拆给多个执行会话、常驻盯守并统一收报。当任务需要并行调研或实现、需要跨会话推进、需要明确哪些动作可自走哪些必须先问用户,或用户提到编排、派单、子会话、多会话协作、常驻盯守、任务票据时使用。
  • 100claude-teams
    Guide for coordinating multiple Claude Code agents as teams. Use when setting up agent teams, configuring subagents, orchestrating multi-agent workflows, or building programmatic agent systems with the Claude Agent SDK.

Feedback and reports

Your message goes to Bartosz Sokoliński. An email address is needed only for a reply. See the privacy policy.