agent-swe-team

Multi-agent SWE team built on the Workshop model. Full-stack vertical workers, meeting room with @mention notification, private pipes, shared task board. Git worktree isolation, Leader-driven coordination. Mixed Codex/Claude Code engine. Use when a task needs engineering depth beyond a single agent. NOT for simple task parallelism (use agent-task-orchestration) or design discussions (use agent-brainstorm).

5 stars

Best use case

agent-swe-team is best used when you need a repeatable AI agent workflow instead of a one-off prompt.

Multi-agent SWE team built on the Workshop model. Full-stack vertical workers, meeting room with @mention notification, private pipes, shared task board. Git worktree isolation, Leader-driven coordination. Mixed Codex/Claude Code engine. Use when a task needs engineering depth beyond a single agent. NOT for simple task parallelism (use agent-task-orchestration) or design discussions (use agent-brainstorm).

Teams using agent-swe-team should expect a more consistent output, faster repeated execution, less prompt rewriting.

When to use this skill

  • You want a reusable workflow that can be run more than once with consistent structure.

When not to use this skill

  • You only need a quick one-off answer and do not need a reusable workflow.
  • You cannot install or maintain the underlying files, dependencies, or repository context.

Installation

Claude Code / Cursor / Codex

$curl -o ~/.claude/skills/agent-swe-team/SKILL.md --create-dirs "https://raw.githubusercontent.com/northseadl/norix-skills/main/agent-swe-team/SKILL.md"

Manual Installation

  1. Download SKILL.md from GitHub
  2. Place it in .claude/skills/agent-swe-team/SKILL.md inside your project
  3. Restart your AI agent — it will auto-discover the skill

How agent-swe-team Compares

Feature / Agentagent-swe-teamStandard Approach
Platform SupportNot specifiedLimited / Varies
Context Awareness High Baseline
Installation ComplexityUnknownN/A

Frequently Asked Questions

What does this skill do?

Multi-agent SWE team built on the Workshop model. Full-stack vertical workers, meeting room with @mention notification, private pipes, shared task board. Git worktree isolation, Leader-driven coordination. Mixed Codex/Claude Code engine. Use when a task needs engineering depth beyond a single agent. NOT for simple task parallelism (use agent-task-orchestration) or design discussions (use agent-brainstorm).

Where can I find the source code?

You can find the source code on GitHub using the link provided at the top of the page.

Related Guides

SKILL.md Source

# Agent SWE Team — Workshop Engine

## 核心身份 — Supervisor (监工)

你是**用户的持续在线代理**。当用户给出工程目标并触发本技能,你**成为 Workshop Supervisor**。
你不写代码——你启动一支全栈工程团队,监控进度,转发人类意图,保障系统透明。

**三层架构**:

```
用户 ─── 给出目标 ─── 可随时介入
  │
  ▼
你 (Supervisor) ── 启动 Workshop → 监控 → 转发 → 收尾
  │
  ▼
Hub (HTTP Server) ── 纯管道 + @mention 自动唤醒
  │
  ├── Leader    ── 内部协调者:分解目标、分配任务、质检、收工
  ├── Worker×N  ── 全栈工匠:各自独立 worktree,垂直切片
  └── Inspector ── 质检官:基于原始目标整体评估
```

**行为准则**:
- **你是最后防线**。Hub 进程在后台运行,没有你监控就不透明。
- **你不做内部决策**。任务分解、分配、质检由 Leader 处理。
- **你持续轮询**。以退避节奏监控状态。
- **你转发人类意图**。用户说什么 → `ws say "消息"`。

## 全流程

### Phase 1: 启动

```bash
# 1. 确保依赖(只需首次)
cd <skill-dir>/agent-swe-team && npm install

# 2. 后台启动(serve 不返回)
node <skill-dir>/agent-swe-team/scripts/team.mjs serve \
  --cwd <项目目录> --goal "你的目标" &

# 3. 等待就绪 + 获取端口
sleep 3
PORT=$(cat <项目目录>/.workshop/port)
export WORKSHOP_CWD=<项目目录>
```

Hub 启动后自动:创建 `.workshop/`、integration 分支、Worker worktrees、打开 Dashboard、唤醒 Leader。

### Phase 2: 监控循环

使用 `ws.mjs` CLI 工具减少上下文消耗:

```bash
WS="node <skill-dir>/agent-swe-team/scripts/ws.mjs"
```

**退避轮询**:

```
Phase     间隔     命令
启动确认  60s      $WS signal
运行中    120s     $WS signal → 如果 RUNNING → $WS board
尾声      300s     $WS signal → 等 COMPLETED
```

```
$WS signal
├─ "COMPLETED" → 跳到 Phase 3
└─ "RUNNING"   → $WS board
     if leader idle 且有 Worker 完成/空闲:
       $WS wake leader
```

**人类消息**: `$WS say "用户说的内容"`

**@mention 自动唤醒**: 内部 Agent 在会议室 @另一个 Agent 时,Hub 自动唤醒被提及者并注入新消息。Supervisor 无需干预。

### Phase 3: 收尾

```bash
$WS signal   # → "COMPLETED"
cd <项目目录>
RUN_ID=$($WS board | head -1 | awk '{print $NF}')
git diff main..integration/$RUN_ID --stat
git checkout main && git merge integration/$RUN_ID --no-ff
```

### 错误恢复

| 场景 | 信号 | 处理 |
|:---|:---|:---|
| Worker 异常 | 会议室 `"异常终止"` | `$WS wake leader` |
| merge 冲突 | 会议室 `"合并失败"` | `$WS wake leader` |
| Hub 进程挂 | PORT 无响应 | 重新 `serve`(board.json 恢复) |
| Leader 卡 idle | Workers 完成但无动作 | `$WS wake leader` |

## CLI 工具

### ws.mjs — 紧凑子命令

```bash
$WS signal                              # → "RUNNING" 或 "COMPLETED"
$WS board                               # 紧凑面板视图 (~10 行)
$WS wake leader                         # 唤醒 Agent
$WS say "message"                       # 发到会议室
$WS say "@worker-1 检查一下日志"         # @mention → 自动唤醒 worker-1
$WS dm worker-1 "私信"                  # 发 DM
$WS task create "标题" --assign worker-1 # 创建任务
$WS task complete 1 "摘要"              # 完成任务
$WS task progress 1 50 "备注"           # 更新进度
$WS merge worker-1                      # 合并分支
$WS done                                # 结束运行
$WS chat                                # 读会议室
```

### team.mjs — 启动命令

```bash
node scripts/team.mjs serve --goal "目标" [选项]
node scripts/team.mjs status [--cwd <DIR>]
```

| 选项 | 默认值 | 说明 |
|:---|:---|:---|
| `--goal` | 必需 | 目标描述 |
| `--cwd` | cwd | 项目目录 |
| `--roles` | `leader,worker:2,inspector` | 团队组成 |
| `--engine` | `codex` | `codex`(thread resume) / `claude`(new session) |
| `--base` | `HEAD` | 基准 commit |
| `--port` | 自动 | 端口号 |
| `--dry-run` | false | 模拟(不启动 Agent) |

## 参考文档

| 文档 | 用途 | 何时读取 |
|:---|:---|:---|
| [api_reference.md](references/api_reference.md) | Hub 全部 HTTP API + JSON Schema | 需要直接调 HTTP API 时 |
| [internal_roles.md](references/internal_roles.md) | Leader/Worker/Inspector 内部机制 | 理解内部行为 / 调试时 |

## 约定

- **规划/汇报**: 中文
- **代码/命令/文件名**: English
- **Git commit**: 中文 (Conventional Commits)

Related Skills

web-scraper

5
from northseadl/norix-skills

Web scraper with SPA/JavaScript rendering, page interaction, and JS execution. Two-tier engine (HTTP → Playwright browser). Smart discovery, batch fetch, interactive content extraction, OpenAPI parsing. Use when read_url_content fails, SPA rendering needed, or page interaction required.

pm-toolkit

5
from northseadl/norix-skills

自然语言 → Mermaid 图表(16 种)+ 网页原型,本地 Web 面板实时预览/导出。

mobile-testing

5
from northseadl/norix-skills

Android/iOS automated testing: device management, app install, performance profiling, log analysis, screenshot comparison, Maestro E2E orchestration.

llm-agent-dev

5
from northseadl/norix-skills

LLM Agent engineering: pattern selection (12-mode matrix), data simulation, convergence iteration. Covers intent routing, function calling, ReAct, MCP, prompt chaining, guardrails, evaluation.

image-studio

5
from northseadl/norix-skills

AI image generation and editing: e-commerce templates (hero/banner/detail/lifestyle), image refinement (background replace/remove, enhance, retouch, style transfer), icon extraction (bg-removal + detect + smart crop with transparent output).

feishu-integration

5
from northseadl/norix-skills

Feishu (Lark) unified CLI for tasks, documents, wiki, bitable, messaging, approval, and Drive. Supports search/create/edit/publish/export across all modules. Use when reading/writing Feishu docs, searching docs or wiki, managing tasks, sending messages, creating approvals, exporting to Markdown, or any 飞书/Lark interaction.

es-analytics

5
from northseadl/norix-skills

Elasticsearch / SLS 只读数据分析:索引探索、mapping、聚合统计、日志搜索、时间序列、多 Profile。

doc-sentinel

5
from northseadl/norix-skills

Document-code change notification system: traceable doc-code binding via git tree hash, git-diff-driven reconciliation plans with confidence/risk metadata, and idempotent execution. Use when maintaining documentation freshness, detecting stale docs, or binding docs to source code.

coding-net-integration

5
from northseadl/norix-skills

Coding.net DevOps automation: MR lifecycle, CI operations (trigger/logs/stop), artifact registry, cross-project queries, remote file audit.

cnb-cool-integration

5
from northseadl/norix-skills

针对 cnb.cool 的云原生构建(CNB Build):生成/修改 .cnb.yml 流水线、触发规则、 构建环境、runner 资源、缓存、环境变量、手动触发与调试。 在接入、迁移、优化或排查 CNB 构建配置时使用。

agent-task-orchestration

5
from northseadl/norix-skills

Task decomposition and multi-agent orchestration with retry, checkpoint recovery, and real-time monitoring. Mixed Codex/Claude Code engine. Parallel/sequential execution.

agent-front-design

5
from northseadl/norix-skills

Frontend design blueprints with craftsmanship scoring, self-critique loops, and engineering handoff. Aesthetic intelligence against AI homogeneity. Use for: UI/UX design specs, design system creation, visual direction exploration, component design review, design-to-engineering handoff.