claude-devfleet

通过Claude DevFleet协调多智能体编码任务——规划项目、在隔离的工作树中并行调度智能体、监控进度并读取结构化报告。

144,923 stars
Complexity: medium

About this skill

The `claude-devfleet` skill empowers a Claude AI agent to act as a sophisticated project manager for complex software development tasks. It coordinates multiple Claude Code agents, enabling them to work concurrently on different parts of a project within isolated Git worktrees. The process begins with project planning based on a user prompt, which breaks down the request into a Directed Acyclic Graph (DAG) of interdependent tasks. Following user approval, the skill dispatches 'missions' to individual agents, monitors their progress, automatically merges completed work, and orchestrates subsequent dependent tasks. Upon completion, it generates structured reports detailing changes, completed work, encountered errors, and suggested next steps. This skill is ideal for developing robust, production-ready systems by leveraging distributed AI agent capabilities, enhancing efficiency, and ensuring structured project execution.

Best use case

Developing complex software projects that can be broken down into parallel sub-tasks. Building applications with multiple interconnected components (e.g., API, database schema, user interface logic). Creating robust and well-tested systems, such as REST APIs with integrated authentication and comprehensive test suites. Managing development workflows that require iterative improvements, automated integration, and detailed progress reporting. Streamlining team-like development efforts using multiple AI agents.

通过Claude DevFleet协调多智能体编码任务——规划项目、在隔离的工作树中并行调度智能体、监控进度并读取结构化报告。

A completed software project or a significant portion thereof, organized into manageable, executed tasks. Detailed, structured reports on each mission's progress, including file changes, completed work, encountered errors, and suggested next steps. Efficient and potentially faster development cycles for complex coding endeavors through automated task dispatch and merges. Well-organized Git worktrees reflecting each agent's isolated contributions, ensuring clean integration. A clear project plan and execution trace, enhancing transparency and traceability.

Practical example

Example input

plan_project("Build a REST API with authentication and integrated tests for user management.")

Example output

{
  "project_id": "proj_devfleet_123",
  "plan": {
    "description": "REST API for user management with auth and tests",
    "tasks": [
      {"id": "M1", "title": "Setup project boilerplate and dependencies", "dependencies": []},
      {"id": "M2", "title": "Implement user authentication module", "dependencies": ["M1"]},
      {"id": "M3", "title": "Develop user data models and CRUD endpoints", "dependencies": ["M1"]},
      {"id": "M4", "title": "Write unit tests for authentication services", "dependencies": ["M2"]},
      {"id": "M5", "title": "Write integration tests for CRUD operations", "dependencies": ["M3"]}
    ],
    "dag_visualization_url": "[link_to_dag_graph]"
  },
  "status": "Project plan generated. Awaiting user approval to dispatch missions."
}

When to use this skill

  • When a coding task is too large or intricate for a single AI agent to complete efficiently.
  • When you need to distribute coding responsibilities among multiple AI agents for parallel execution.
  • When the project requires structured planning, task dependency management, and automated progress tracking.
  • When developing production-grade software that benefits from isolated development environments (Git worktrees) for each agent's contribution.

When not to use this skill

  • For very simple, single-file, or trivial coding tasks that do not benefit from coordination or parallelization.
  • If you do not have access to or cannot run a Claude DevFleet instance.
  • When the project's nature does not allow for effective task decomposition or parallel execution.
  • When fine-grained, manual control over every line of code is preferred over an orchestrated, agent-driven approach.

Installation

Claude Code / Cursor / Codex

$curl -o ~/.claude/skills/claude-devfleet/SKILL.md --create-dirs "https://raw.githubusercontent.com/affaan-m/everything-claude-code/main/docs/zh-CN/skills/claude-devfleet/SKILL.md"

Manual Installation

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

How claude-devfleet Compares

Feature / Agentclaude-devfleetStandard Approach
Platform SupportClaudeLimited / Varies
Context Awareness High Baseline
Installation ComplexitymediumN/A

Frequently Asked Questions

What does this skill do?

通过Claude DevFleet协调多智能体编码任务——规划项目、在隔离的工作树中并行调度智能体、监控进度并读取结构化报告。

Which AI agents support this skill?

This skill is designed for Claude.

How difficult is it to install?

The installation complexity is rated as medium. You can find the installation instructions above.

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

# Claude DevFleet 多智能体编排

## 使用时机

当需要调度多个 Claude Code 智能体并行处理编码任务时使用此技能。每个智能体在独立的 git worktree 中运行,并配备全套工具。

需要连接一个通过 MCP 运行的 Claude DevFleet 实例:

```bash
claude mcp add devfleet --transport http http://localhost:18801/mcp
```

## 工作原理

```
用户 → "构建一个带有身份验证和测试的 REST API"
  ↓
plan_project(prompt) → 项目ID + 任务DAG
  ↓
向用户展示计划 → 获取批准
  ↓
dispatch_mission(M1) → 代理1在工作树中生成
  ↓
M1完成 → 自动合并 → 自动分发M2 (依赖于M1)
  ↓
M2完成 → 自动合并
  ↓
get_report(M2) → 更改的文件、完成的工作、错误、后续步骤
  ↓
向用户报告
```

### 工具

| 工具 | 用途 |
|------|---------|
| `plan_project(prompt)` | AI 将描述分解为包含链式任务的项目 |
| `create_project(name, path?, description?)` | 手动创建项目,返回 `project_id` |
| `create_mission(project_id, title, prompt, depends_on?, auto_dispatch?)` | 添加任务。`depends_on` 是任务 ID 字符串列表(例如 `["abc-123"]`)。设置 `auto_dispatch=true` 可在依赖满足时自动启动。 |
| `dispatch_mission(mission_id, model?, max_turns?)` | 启动智能体执行任务 |
| `cancel_mission(mission_id)` | 停止正在运行的智能体 |
| `wait_for_mission(mission_id, timeout_seconds?)` | 阻塞直到任务完成(见下方说明) |
| `get_mission_status(mission_id)` | 检查任务进度而不阻塞 |
| `get_report(mission_id)` | 读取结构化报告(更改的文件、测试情况、错误、后续步骤) |
| `get_dashboard()` | 系统概览:运行中的智能体、统计信息、近期活动 |
| `list_projects()` | 浏览所有项目 |
| `list_missions(project_id, status?)` | 列出项目中的任务 |

> **关于 `wait_for_mission` 的说明:** 此操作会阻塞对话,最长 `timeout_seconds` 秒(默认 600 秒)。对于长时间运行的任务,建议改为每 30-60 秒使用 `get_mission_status` 轮询,以便用户能看到进度更新。

### 工作流:规划 → 调度 → 监控 → 报告

1. **规划**:调用 `plan_project(prompt="...")` → 返回 `project_id` 以及带有 `depends_on` 链和 `auto_dispatch=true` 的任务列表。
2. **展示计划**:向用户呈现任务标题、类型和依赖链。
3. **调度**:对根任务(`depends_on` 为空)调用 `dispatch_mission(mission_id=<first_mission_id>)`。剩余任务在其依赖项完成时自动调度(因为 `plan_project` 为它们设置了 `auto_dispatch=true`)。
4. **监控**:调用 `get_mission_status(mission_id=...)` 或 `get_dashboard()` 检查进度。
5. **报告**:任务完成后调用 `get_report(mission_id=...)`。与用户分享亮点。

### 并发性

DevFleet 默认最多同时运行 3 个智能体(可通过 `DEVFLEET_MAX_AGENTS` 配置)。当所有槽位都占满时,设置了 `auto_dispatch=true` 的任务会在任务监视器中排队,并在槽位空闲时自动调度。检查 `get_dashboard()` 了解当前槽位使用情况。

## 示例

### 全自动:规划并启动

1. `plan_project(prompt="...")` → 显示包含任务和依赖关系的计划。
2. 调度第一个任务(`depends_on` 为空的那个)。
3. 剩余任务在依赖关系解决时自动调度(它们具有 `auto_dispatch=true`)。
4. 报告项目 ID 和任务数量,让用户知道启动了哪些内容。
5. 定期使用 `get_mission_status` 或 `get_dashboard()` 轮询,直到所有任务达到终止状态(`completed`、`failed` 或 `cancelled`)。
6. 对每个终止任务执行 `get_report(mission_id=...)`——总结成功之处,并指出失败任务及其错误和后续步骤。

### 手动:逐步控制

1. `create_project(name="My Project")` → 返回 `project_id`。
2. 为第一个(根)任务执行 `create_mission(project_id=project_id, title="...", prompt="...", auto_dispatch=true)` → 捕获 `root_mission_id`。
   为每个后续任务执行 `create_mission(project_id=project_id, title="...", prompt="...", auto_dispatch=true, depends_on=["<root_mission_id>"])`。
3. 在第一个任务上执行 `dispatch_mission(mission_id=...)` 以启动链。
4. 完成后执行 `get_report(mission_id=...)`。

### 带审查的串行执行

1. `create_project(name="...")` → 获取 `project_id`。
2. `create_mission(project_id=project_id, title="Implement feature", prompt="...")` → 获取 `impl_mission_id`。
3. `dispatch_mission(mission_id=impl_mission_id)`,然后使用 `get_mission_status` 轮询直到完成。
4. `get_report(mission_id=impl_mission_id)` 以审查结果。
5. `create_mission(project_id=project_id, title="Review", prompt="...", depends_on=[impl_mission_id], auto_dispatch=true)` —— 由于依赖已满足,自动启动。

## 指南

* 在调度前始终与用户确认计划,除非用户已明确指示继续。
* 报告状态时包含任务标题和 ID。
* 如果任务失败,在重试前读取其报告。
* 批量调度前检查 `get_dashboard()` 了解智能体槽位可用性。
* 任务依赖关系构成一个有向无环图(DAG)——不要创建循环依赖。
* 每个智能体在独立的 git worktree 中运行,并在完成时自动合并。如果发生合并冲突,更改将保留在智能体的 worktree 分支上,以便手动解决。
* 手动创建任务时,如果希望它们在依赖项完成时自动触发,请始终设置 `auto_dispatch=true`。没有此标志,任务将保持 `draft` 状态。

Related Skills

everything-claude-code-conventions

144923
from affaan-m/everything-claude-code

Development conventions and patterns for everything-claude-code. JavaScript project with conventional commits.

DevelopmentClaude

claude-api

144923
from affaan-m/everything-claude-code

Anthropic Claude API patterns for Python and TypeScript. Covers Messages API, streaming, tool use, vision, extended thinking, batches, prompt caching, and Claude Agent SDK. Use when building applications with the Claude API or Anthropic SDKs.

DevelopmentClaude

workspace-surface-audit

144923
from affaan-m/everything-claude-code

Audit the active repo, MCP servers, plugins, connectors, env surfaces, and harness setup, then recommend the highest-value ECC-native skills, hooks, agents, and operator workflows. Use when the user wants help setting up Claude Code or understanding what capabilities are actually available in their environment.

DevelopmentClaude

safety-guard

144923
from affaan-m/everything-claude-code

Use this skill to prevent destructive operations when working on production systems or running agents autonomously.

DevelopmentClaude

repo-scan

144923
from affaan-m/everything-claude-code

Cross-stack source code asset audit — classifies every file, detects embedded third-party libraries, and delivers actionable four-level verdicts per module with interactive HTML reports.

DevelopmentClaude

project-flow-ops

144923
from affaan-m/everything-claude-code

Operate execution flow across GitHub and Linear by triaging issues and pull requests, linking active work, and keeping GitHub public-facing while Linear remains the internal execution layer. Use when the user wants backlog control, PR triage, or GitHub-to-Linear coordination.

DevelopmentClaude

manim-video

144923
from affaan-m/everything-claude-code

Build reusable Manim explainers for technical concepts, graphs, system diagrams, and product walkthroughs, then hand off to the wider ECC video stack if needed. Use when the user wants a clean animated explainer rather than a generic talking-head script.

DevelopmentClaude

laravel-plugin-discovery

144923
from affaan-m/everything-claude-code

Discover and evaluate Laravel packages via LaraPlugins.io MCP. Use when the user wants to find plugins, check package health, or assess Laravel/PHP compatibility.

DevelopmentClaude

design-system

144923
from affaan-m/everything-claude-code

Use this skill to generate or audit design systems, check visual consistency, and review PRs that touch styling.

DevelopmentClaude

click-path-audit

144923
from affaan-m/everything-claude-code

Trace every user-facing button/touchpoint through its full state change sequence to find bugs where functions individually work but cancel each other out, produce wrong final state, or leave the UI in an inconsistent state. Use when: systematic debugging found no bugs but users report broken buttons, or after any major refactor touching shared state stores.

DevelopmentClaude

ck

144923
from affaan-m/everything-claude-code

Persistent per-project memory for Claude Code. Auto-loads project context on session start, tracks sessions with git activity, and writes to native memory. Commands run deterministic Node.js scripts — behavior is consistent across model versions.

DevelopmentClaude

canary-watch

144923
from affaan-m/everything-claude-code

Use this skill to monitor a deployed URL for regressions after deploys, merges, or dependency upgrades.

DevelopmentClaude