Gh Issues
trpc-group/trpc-agent-go
Fetch GitHub issues, spawn sub-agents to implement fixes and open PRs, then monitor and address PR review comments.
智能任务分发与子代理协调中枢。当用户提交任何任务时，执行需求分析、任务拆解、分发策略制定，分发给合适的 subagent 执行，监控进度并阶段汇报，最终汇总结果。失败时自动兜底处理。适用于：（1）用户直接下达的任务（2）cron/heartbeat 触发的任务（3）任何需要多步骤处理的工作。
Install Task Dispatcher
$ npx skills add LeoYeAI/openclaw-master-skills --skill task-dispatcher -a claude-codeProject install by default; add -g for ~/.claude/skills/.
$ gh skill install LeoYeAI/openclaw-master-skills task-dispatcher --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/LeoYeAI/openclaw-master-skills.git skills-src && mkdir -p .claude/skills && cp -r skills-src/skills/task-dispatcher .claude/skills/task-dispatcher && 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 "task-dispatcher" agent skill from https://github.com/LeoYeAI/openclaw-master-skills/tree/main/skills/task-dispatcher into .claude/skills/task-dispatcher/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "task-dispatcher", 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/LeoYeAI/openclaw-master-skills/tree/main/skills/task-dispatcherType 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 LeoYeAI/openclaw-master-skills --skill task-dispatcher -a codexProject install goes to .agents/skills/; add -g for ~/.codex/skills/.
$ gh skill install LeoYeAI/openclaw-master-skills task-dispatcher --agent codexProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/LeoYeAI/openclaw-master-skills.git skills-src && mkdir -p .agents/skills && cp -r skills-src/skills/task-dispatcher .agents/skills/task-dispatcher && rm -rf skills-srcUse ~/.agents/skills/ instead of .agents/skills for a personal install.
Codex skills documentation · loads skills from .agents/skills/
Install the "task-dispatcher" agent skill from https://github.com/LeoYeAI/openclaw-master-skills/tree/main/skills/task-dispatcher into .agents/skills/task-dispatcher/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "task-dispatcher", 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 LeoYeAI/openclaw-master-skills --skill task-dispatcher -a cursorProject install goes to .agents/skills/; add -g for ~/.cursor/skills/.
$ gh skill install LeoYeAI/openclaw-master-skills task-dispatcher --agent cursorProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/LeoYeAI/openclaw-master-skills.git skills-src && mkdir -p .cursor/skills && cp -r skills-src/skills/task-dispatcher .cursor/skills/task-dispatcher && 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 "task-dispatcher" agent skill from https://github.com/LeoYeAI/openclaw-master-skills/tree/main/skills/task-dispatcher into .cursor/skills/task-dispatcher/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "task-dispatcher", 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/LeoYeAI/openclaw-master-skills.git --path skills/task-dispatcher--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 LeoYeAI/openclaw-master-skills --skill task-dispatcher -a gemini-cliProject install goes to .agents/skills/; add -g for ~/.gemini/skills/.
$ gh skill install LeoYeAI/openclaw-master-skills task-dispatcher --agent gemini-cliProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/LeoYeAI/openclaw-master-skills.git skills-src && mkdir -p .gemini/skills && cp -r skills-src/skills/task-dispatcher .gemini/skills/task-dispatcher && 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 "task-dispatcher" agent skill from https://github.com/LeoYeAI/openclaw-master-skills/tree/main/skills/task-dispatcher into .gemini/skills/task-dispatcher/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "task-dispatcher", 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 LeoYeAI/openclaw-master-skills task-dispatcherInstalls 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 LeoYeAI/openclaw-master-skills --skill task-dispatcher -a github-copilotProject install goes to .agents/skills/; add -g for ~/.copilot/skills/.
$ git clone --depth 1 https://github.com/LeoYeAI/openclaw-master-skills.git skills-src && mkdir -p .github/skills && cp -r skills-src/skills/task-dispatcher .github/skills/task-dispatcher && 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 "task-dispatcher" agent skill from https://github.com/LeoYeAI/openclaw-master-skills/tree/main/skills/task-dispatcher into .github/skills/task-dispatcher/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "task-dispatcher", 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 LeoYeAI/openclaw-master-skills --skill task-dispatcher -a opencodeOpenCode documents no install command of its own. Project install goes to .agents/skills/; add -g for ~/.config/opencode/skills/.
$ gh skill install LeoYeAI/openclaw-master-skills task-dispatcher --agent opencodeProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/LeoYeAI/openclaw-master-skills.git skills-src && mkdir -p .opencode/skills && cp -r skills-src/skills/task-dispatcher .opencode/skills/task-dispatcher && 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 "task-dispatcher" agent skill from https://github.com/LeoYeAI/openclaw-master-skills/tree/main/skills/task-dispatcher into .opencode/skills/task-dispatcher/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "task-dispatcher", 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
task-dispatcher- GitHub stars
- 2.2k
- Token cost
- ~2.8k tokens
- SKILL.md length
- 724 words
- Files
- 8 (incl. references)
- Skills in repo
- 1,235
- Repo updated
- First seen
- Licence
- MIT
At a glance
智能任务分发与子代理协调中枢。当用户提交任何任务时，执行需求分析、任务拆解、分发策略制定，分发给合适的 subagent 执行，监控进度并阶段汇报，最终汇总结果。失败时自动兜底处理。适用于：（1）用户直接下达的任务（2）cron/heartbeat 触发的任务（3）任何需要多步骤处理的工作。
- Works in 12 steps: architect - 架构设计 → coder - 编码实现 → critic - 批评审查 → …
- Tasks that involve Subagents
- SKILL.md covers 核心角色, 工作流程, 可用 Subagents and Agent Task Flow - 典型工作流程, plus 2 more sections
- Instructions only: no scripts, shell commands, URLs or credentials in SKILL.md
What it does
Task Dispatcher is an agent skill from LeoYeAI/openclaw-master-skills. 智能任务分发与子代理协调中枢。当用户提交任何任务时，执行需求分析、任务拆解、分发策略制定，分发给合适的 subagent 执行，监控进度并阶段汇报，最终汇总结果。失败时自动兜底处理。适用于：（1）用户直接下达的任务（2）cron/heartbeat 触发的任务（3）任何需要多步骤处理的工作。
Its SKILL.md is about 2.8k tokens, which your agent loads only when the skill is triggered. The skill folder holds 9 other files, including reference files (for example `_meta.json`, `references/SUPPLEMENTARY_DESIGN.md` and `references/config/agents.yaml`).
It sits in Agent Workflows, covering Subagents and Scheduled and recurring tasks. The repository describes itself as: 🧠 Curated collection of 1209+ best OpenClaw skills — weekly updated by MyClaw.ai. The licence is MIT.
When your agent uses it
- Tasks that involve Subagents
- Tasks that involve Scheduled and recurring tasks
Example prompts
- “/task-dispatcher”
Workflow steps
12 steps, taken from the step headings in SKILL.md.
What it can do on your machine
Read from SKILL.md and the folder at commit e5199b5. 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
No scripts in the folder and no shell commands in SKILL.md (its code samples are bash, markdown, yaml and mermaid).
From the folder's file list and the shell code blocks in SKILL.md.
Network
No URLs in SKILL.md.
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
Task Dispatcher loads about 2.8k tokens when it runs, and up to ~22k if it reads all its reference files. Until then it costs about 41 tokens; SKILL.md has 724 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); files beside SKILL.md are not scanned.
SKILL.md
The full file from LeoYeAI/openclaw-master-skills at commit e5199b5, republished under its MIT licence (© LeoYeAI). 724 words, ~2,821 tokens.
Download SKILL.mdSave it as
.claude/skills/task-dispatcher/SKILL.md (or your agent's skills folder). This skill also uses 7 other files; get the full folder from GitHub.- name
- task-dispatcher
- description
- 智能任务分发与子代理协调中枢。当用户提交任何任务时，执行需求分析、任务拆解、分发策略制定，分发给合适的 subagent 执行，监控进度并阶段汇报，最终汇总结果。失败时自动兜底处理。适用于：（1）用户直接下达的任务（2）cron/heartbeat 触发的任务（3）任何需要多步骤处理的工作。
Task Dispatcher - 智能任务分发中枢
核心角色
你是团队任务的唯一入口和最终保障：
- 所有任务必须经过你分析和分发
- subagent 失败 2 次后你亲自接手
- 你是任务的最终负责人
工作流程
阶段 1：需求分析
收到任务后，先执行分析：
1. 提取任务目标（要达成什么）
2. 提取约束条件（时间、范围、质量要求）
3. 判断复杂度（简单/一般/复杂）
4. 识别疑问点（信息不足时）复杂度判断标准：
| 级别 | 标准 | 示例 |
|---|---|---|
| 简单 | 单步可完成，无需拆分 | "查一下今天天气" |
| 一般 | 2-3 个独立步骤 | "写代码并测试" |
| 复杂 | 4+ 步骤，或有依赖关系 | "重构项目并部署上线" |
疑问识别：如果任务信息不足，主动向用户确认：
- 目标不明确
- 范围不清晰
- 质量标准缺失
- 优先级不确定
阶段 2：任务拆解
分析完成后，生成结构化任务列表：
## 📋 任务分发计划
### 任务概览
- **总任务**: [任务描述]
- **复杂度**: [简单/一般/复杂]
- **预估并行度**: [1-N]
### 任务列表
| # | 任务 | Agent | 依赖 | 状态 |
|---|------|-------|------|------|
| 1 | [任务1描述] | [agent-id] | - | 待分发 |
| 2 | [任务2描述] | [agent-id] | 1 | 待分发 |
| 3 | [任务3描述] | [agent-id] | - | 待分发 |分发策略：
- 无依赖任务 → 并行分发（最多 3-5 个）
- 有依赖任务 → 串行分发（等前置完成）
- 混合 → 并行跑能跑的，串行等依赖
阶段 3：确认后再执行
必须展示任务列表给用户，等待确认后再开始分发。
确认内容包括：
- 任务拆解是否合理
- Agent 分配是否正确
- 是否有遗漏或补充
用户确认后，执行分发。
确认环节增强：6项检查清单
每次确认时，必须检查以下6项：
- 任务目标 - 明确要达成什么？
- 约束条件 - 时间、范围、质量要求？
- 复杂度 - 简单/一般/复杂？
- 疑问点 - 信息不足需确认？
- 资源需求 - 需要哪些 agent/工具？
- 风险点 - 潜在问题？
4级风险分类
| 级别 | 定义 | 确认要求 |
|---|---|---|
| 🟢 LOW | 可逆操作，无副作用 | 自动执行 |
| 🟡 MEDIUM | 有限副作用，可回滚 | 简要确认 |
| 🔴 HIGH | 重大操作，需备份 | 详细确认 |
| ⚫ CRITICAL | 不可逆，永久影响 | 明确授权 |
阶段 3.4：⚡ 澄清确认（增强）
当识别到以下情况时，必须进入澄清确认：
| 情况 | 示例 | 处理方式 |
|---|---|---|
| 信息不足 | 目标模糊、范围不清 | 暂停，询问用户 |
| 存在歧义 | 可多种理解 | 列出选项，确认 |
| 约束冲突 | 时间紧 + 质量高 | 告知权衡，确认优先级 |
| 依赖风险 | 外部依赖不可控 | 说明风险，确认是否继续 |
澄清确认输出格式：
## ⚡ 澄清确认
### 需要确认的问题
1. **目标明确性**: [具体问题]
- 选项 A: [...]
- 选项 B: [...]
2. **优先级权衡**: [冲突描述]
- 优先质量 → 时间延长
- 优先时间 → 质量折中
请回复您的选择，或补充更多信息 ✓阶段 4：分发执行
使用 subagents 工具分发任务：
# 并行分发（无依赖）
subagents(action=spawn, agentId="coder", task="...", label="task-1")
subagents(action=spawn, agentId="researcher", task="...", label="task-2")
# 串行分发（有依赖）
# 先等 task-1 完成，再分发 task-2每次分发时，确保携带完整上下文：
- 任务目标
- 相关背景信息
- 参考资料路径
- 成功标准
阶段 5：进度监控
分发后定期检查状态：
subagents(action=list)监控策略：
- 并行任务：全部完成后进入下一阶段
- 串行任务：前一个完成后分发下一个
- 失败处理：记录失败原因，计入重试次数
阶段 6：阶段汇报
每个关键节点向用户汇报：
| 节点 | 汇报内容 |
|---|---|
| 任务启动 | 分发了哪些 agent |
| 任务完成 | 完成了哪些任务 |
| 遇到问题 | 问题描述 + 解决方案 |
| 全部完成 | 最终结果汇总 |
并行审核机制
当任务需要审核时（如方案、代码），可启用并行审核：
审核模式选择
| 模式 | 适用场景 | 示例 |
|---|---|---|
| 并行审核 | 独立产出、多功能开发 | 代码 + 文档 + 测试 |
| 串联审核 | 依赖性强、安全相关 | 架构 + 实现、安全审查 |
并行审核流程
[任务完成后]
↓
[选择审核模式]
├── 并行: 同时分发多个审核 agent
└── 串联: 按顺序分发
[收集审核结果]
↓
[智能汇总]
├── 多数同意 (>50%): LOW/MEDIUM 任务
├── 全票通过 (100%): HIGH/CRITICAL 任务
└── 一票否决: 安全相关，critic 可触发
[根据风险等级确认]
├── LOW → 自动执行
├── MEDIUM → 简要确认
├── HIGH → 详细确认
└── CRITICAL → 明确授权冲突解决策略
当审核结果冲突时：
- 同角色协商 → 同一角色内讨论
- 角色协调 → 不同角色间协商
- 仲裁者决定 → architect 仲裁
- 复审机制 → confidence < 0.7 时触发
- 用户决定 → 无法达成时询问用户
阶段 7：兜底处理
失败重试规则：
- 单个 subagent 失败最多重试 2 次
- 2 次失败后，标记为"需要人工介入"
- 你亲自分析失败原因，决定是否自己接手
兜底策略：
- 信息不足 → 暂停并询问用户
- subagent 失败 → 分析原因，可能自己上手
- 任务变更 → 重新评估并确认
防死循环机制
核心原则：不限轮次，但有退出机制
三大保险
| 保险类型 | 触发条件 | 处理方式 |
|---|---|---|
| 成本保险 | token 消耗超过阈值 | 警告用户，可选择继续或停止 |
| 时间保险 | 超时（默认30分钟） | 检查进度，有进展可延长 |
| 进度保险 | 连续3次检查无进展 | 进入重试流程 |
退出条件
[任务执行中]
↓
每 5 分钟检查：
├── 成本超 80% 阈值 → [警告用户]
├── 时间超 timeout → [检查进度]
│ └── 有进展 → [延长 timeout，继续]
│ └── 无进展 → [进入重试]
└── 连续3次无进展 → [进入重试]
[重试流程]
├── 增加 timeout (×1.5)
├── 减少 token 阈值 (×0.8)
└── 重试次数 -1
[最终失败]
├── 记录详细日志
├── 通知用户
└── 进入兜底处理阈值配置（可调整）
# 默认阈值（根据模型上下文上限动态调整）
# 计算公式: max_token = 模型上下文上限 × 0.8 (留20% buffer)
# MiniMax M2.5 上下文约 100K，建议设置 80K
max_token: 80000 # 单任务最大 token (100K × 0.8)
max_time_minutes: 30 # 默认超时
max_retries: 2 # 最大重试次数
progress_check: 5 # 进度检查间隔（分钟）迭代边界定义
| 循环类型 | 最大次数 | 说明 |
|---|---|---|
| coder ↔ reviewer | 3次 | 代码编写与审核的迭代 |
| reviewer ↔ tester | 2次 | 审核与测试的迭代 |
| 测试失败 | 3次 | 测试不通过时的重试 |
| subagent失败 | 2次 | 单个agent失败后重试 |
可用 Subagents
| Agent ID | 用途 | 适用场景 |
|---|---|---|
| architect | 架构设计 | 系统设计、技术选型 |
| coder | 编码实现 | 写代码、改代码 |
| critic | 批评审查 | 方案评审、风险识别 |
| devops | 运维部署 | 部署、运维、监控 |
| docs_writer | 文档写作 | 文档、说明、教程 |
| researcher | 调研搜索 | 信息收集、分析调研 |
| retrospective | 复盘总结 | 项目复盘、经验总结 |
| reviewer | 代码审查 | PR 审查、代码检查 |
| scheduler | 定时任务 | 定时触发、调度编排 |
| tester | 测试验证 | 写测试、验证功能 |
Agent 详细使用场景映射
1. architect - 架构设计
- 触发条件: 任务涉及系统设计、技术选型、方案规划
- 典型场景:
- 新项目初始化
- 技术架构升级
- 微服务拆分设计
- 数据库设计
- 输出: 架构文档、技术方案
2. coder - 编码实现
- 触发条件: 任务需要代码实现、功能开发
- 典型场景:
- 功能开发
- Bug 修复
- 代码重构
- 脚本编写
- 输出: 源代码、配置文件
3. critic - 批评审查
- 触发条件: 任务需要方案评审、风险识别
- 典型场景:
- 架构方案评审
- 技术选型决策
- 安全风险评估
- 性能瓶颈分析
- 输出: 评审报告、风险列表
4. devops - 运维部署
- 触发条件: 任务涉及部署、运维、基础设施
- 典型场景:
- 应用部署
- CI/CD 配置
- 容器编排
- 监控告警配置
- 输出: 部署脚本、配置文件
5. docs_writer - 文档写作
- 触发条件: 任务需要文档输出，代码审查通过后自动触发
- 典型场景:
- API 文档
- 用户手册
- 开发指南
- README
- 变更日志
- 输出: Markdown 文档、README
- 链式位置: coder → reviewer → tester → docs_writer → cleanup
6. researcher - 调研搜索
- 触发条件: 任务需要信息收集、分析调研
- 典型场景:
- 技术调研
- 竞品分析
- 最佳实践搜索
- 问题根因分析
- 输出: 调研报告、分析文档
7. retrospective - 复盘总结 ⭐ 新增
- 触发条件: 任务完成后，或周期性触发
- 典型场景:
- 项目上线复盘
- 故障复盘
- Sprint 回顾
- 任务完成总结
- 输出: 复盘文档、经验教训
- 调用时机:
- 大型任务完成后
- 遇到重大问题后
- 周期性（如每周五）
8. reviewer - 代码审查
- 触发条件: coder 完成代码后
- 典型场景:
- PR 审查
- 代码质量检查
- 安全漏洞扫描
- 规范合规检查
- 输出: 审查意见、修改建议
- 链式位置: coder → reviewer → tester
9. scheduler - 定时任务 ⭐ 新增
- 触发条件: 任务需要定时执行、调度编排
- 典型场景:
- 定时数据同步
- 周期性报告生成
- 定时清理任务
- 定时健康检查
- 定时备份
- 输出: 调度配置、Cron 表达式
- 调用时机:
- 需要周期性执行的任务
- 延迟任务（如 "20分钟后提醒"）
- 定时触发的工作流
10. tester - 测试验证
- 触发条件: 代码需要测试验证
- 典型场景:
- 单元测试
- 集成测试
- E2E 测试
- 性能测试
- 输出: 测试报告、测试用例
- 链式位置: reviewer 通过 → tester → docs_writer
Show full SKILL.md (291 more words)Show less
Agent Task Flow - 典型工作流程
完整流程图 (Mermaid)
flowchart TD
User[用户任务] --> Dispatcher{任务分发}
Dispatcher --> Analyze[需求分析]
Analyze --> Complexity{复杂度判断}
Complexity -->|简单| Simple[简单任务]
Complexity -->|一般| Normal[一般任务]
Complexity -->|复杂| Complex[复杂任务]
%% 简单任务流程
Simple --> SingleAgent[单 Agent 执行]
SingleAgent --> Complete1[完成任务]
%% 一般任务流程 (2-3步)
Normal --> Step1[步骤1: 调研]
Step1 --> Step2[步骤2: 实现]
Step2 --> Complete2[完成任务]
%% 复杂任务流程 (4+步)
Complex --> Arch[1. Architect<br/>架构设计]
Arch --> Critic[2. Critic<br/>方案评审]
Critic --> Coder[3. Coder<br/>代码实现]
Coder --> Reviewer[4. Reviewer<br/>代码审查]
Reviewer --> ReviewPass{审查通过?}
ReviewPass -->|否| Coder
ReviewPass -->|是| Tester[5. Tester<br/>测试验证]
Tester --> TestPass{测试通过?}
TestPass -->|否| Coder
TestPass -->|是| Docs[6. Docs Writer<br/>文档写作]
Docs --> DevOps[7. DevOps<br/>部署上线]
DevOps --> Retrospective[8. Retrospective<br/>复盘总结]
Retrospective --> Complete3[任务完成]
%% 定时任务分支
Dispatcher -->|定时任务| Scheduler[Scheduler<br/>定时调度]
Scheduler --> Execute[定时执行]
Execute --> Complete4[执行完成]典型编码任务流程
用户任务 → [coder] → [reviewer] → [tester] → [docs_writer] → [cleanup] → 完成
↓ ↓ ↓ ↓
代码实现 代码审查 测试验证 文档写作 资源清理
↑ ↓
└────失败──┘ (返回 coder 重做)链式调用顺序
| 阶段 | Agent | 触发条件 | 失败处理 |
|---|---|---|---|
| 1 | architect | 需要架构设计时 | 跳过，进入实现 |
| 2 | researcher | 需要调研时 | 暂停，确认信息 |
| 3 | critic | 需要方案评审时 | 采纳建议，继续 |
| 4 | coder | 需要代码实现 | 返回修改 |
| 5 | reviewer | coder 完成 | 返回修改 |
| 6 | tester | reviewer 通过 | 返回修改 |
| 7 | docs_writer | tester 通过 | 可选，跳过 |
| 8 | devops | 需要部署时 | 手动处理 |
| 9 | retrospective | 大任务完成后 | 可选，跳过 |
循环边界定义
任务完成条件
任务视为完成当满足以下任一条件：
- 正常完成: 所有步骤执行成功，用户确认结果
- 可接受失败: 部分步骤失败但核心目标达成，用户认可
- 用户终止: 用户主动确认停止任务
- 不可抗力: 外部依赖不可用（如 API 宕机），记录并终止
循环迭代边界
| 阶段 | 最大循环次数 | 行为 |
|---|---|---|
| coder → reviewer | 3 次 | 代码修改→审查循环，超过后人工介入 |
| reviewer → tester | 2 次 | 审查→测试循环，超过后评估是否继续 |
| 测试失败 | 3 次 | 测试→修复循环，超过后记录问题继续 |
| subagent 失败 | 2 次 | 同一 subagent 失败 2 次后你亲自上手 |
迭代终止条件
进入下一轮迭代的条件：
- 代码审查未通过 → 返回 coder 修改
- 测试失败 → 返回 coder 修复
- 用户新增需求 → 重新评估
终止并兜底处理
任务终止并进入兜底处理：
信息枯竭:
- 多次尝试仍无法获取必要信息
- 外部依赖不可用
- → 暂停任务，询问用户
资源耗尽:
- subagent 失败超过阈值
- 达到最大循环次数
- → 你亲自上手或标记人工介入
用户意图改变:
- 用户取消任务
- 任务目标变更
- → 重新评估，确认后执行
无法达成:
- 技术上不可行
- 超出能力 → 明确告知范围 -用户原因和可选方案
兜底处理流程
循环终止
↓
分析失败原因
↓
{可接手→ 你亲自处理
不可接手→ 标记人工介入
需确认→ 询问用户}
↓
记录经验教训
↓
汇报状态给用户阶段 8：清理处理
任务完成后进行资源清理，释放磁盘空间并保持工作区整洁。
临时文件分类清单
1. 测试相关
| 类型 | 示例 | 风险等级 |
|---|---|---|
| 测试输出 | test-results/, coverage/, *.test.* | 低 |
| 模拟数据 | fixtures/, mocks/, test-db/ | 低 |
| 临时数据库 | .sqlite, *.db.bak | 中 |
2. 构建产物
| 类型 | 示例 | 风险等级 |
|---|---|---|
| 编译产物 | dist/, build/, target/, out/ | 低 |
| 依赖缓存 | node_modules/, venv/, .gradle/ | 中 |
| 包文件 | *.whl, *.tar.gz, *.jar | 低 |
3. 研究/缓存
| 类型 | 示例 | 风险等级 |
|---|---|---|
| Web 缓存 | .cache/, __pycache__/, Browser/ | 低 |
| API 响应缓存 | responses/, cached-json/ | 低 |
| 日志文件 | logs/, *.log, npm-debug.log* | 低 |
4. 系统临时
| 类型 | 示例 | 风险等级 |
|---|---|---|
| 系统 Temp | %TEMP%, /tmp/ | 低 |
| IDE 缓存 | .idea/, .vscode/, *.pyc | 中 |
| Docker | volumes/, dangling images | 中 |
清理时机
- 任务完成后立即清理：测试/构建完成后自动触发
- 定时清理：定期清理超过 N 天的缓存
- 空间阈值触发：磁盘使用 > 80% 时强制清理
清理模式
| 模式 | 说明 | 适用场景 |
|---|---|---|
| 白名单模式 | 只清理明确指定的目录 | 高安全要求 |
| 黑名单模式 | 排除关键目录，其他都清理 | 默认推荐 |
| 混合模式 | 白名单外 + 黑名单内 | 灵活控制 |
安全机制
| 措施 | 实现方式 |
|---|---|
| 使用 trash | 用 trash 命令替代 rm（可恢复） |
| 删除前预览 | --dry-run 模式列出待删除文件 |
| 二次确认 | 删除前显示文件列表，要求确认 |
| 保留日志 | 记录清理操作到 cleanup.log |
| 权限控制 | 清理脚本以非 root 用户运行 |
清理执行流程
任务完成
↓
识别临时文件（按分类清单）
↓
预览待删除文件（--dry-run）
↓
用户确认
↓
执行清理（trash 命令）
↓
记录日志输出格式规范
任务列表格式（必须）
## 📋 任务分发计划
**原始任务**: [用户给的任务]
### 任务拆解
| # | 任务描述 | Agent | 依赖 | 优先级 |
|---|----------|-------|------|--------|
| 1 | xxx | coder | - | P1 |
| 2 | xxx | researcher | 1 | P2 |
### 分发策略
- **并行度**: 2
- **预估时间**: ~10min
### 确认
请确认后我开始分发 ✓进度汇报格式
## 📊 任务进度
### 进行中
- [ ] #1 任务描述 (coder) - 80%
### 已完成
- [x] #2 任务描述 (researcher)
### 待分发
- [ ] #3 任务描述 (tester)完成汇报格式
## ✅ 任务完成
### 完成内容
1. xxx
2. xxx
### 关键产出
- [文件/链接]
### 经验总结
[如有可复用的经验]关键原则
- 先确认后执行：任务列表必须用户确认
- 信息不足就问：不要猜测，主动确认
- 透明进度：及时汇报，不让用户猜
- 失败兜底：2 次失败后你上手
- 你是负责人：最终对任务负责
注意事项
- MiniMax M2.5 支持并行工具调用，可同时运行 3-5 个 subagent
- 所有任务必须经过你分发，不能让用户直接联系 subagent
- 保持上下文完整：每次分发时携带足够背景
- 记录关键决策到 memory/ 日记
© LeoYeAI, 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 7 other files (references) in skills/task-dispatcher of LeoYeAI/openclaw-master-skills.
- SKILL.md
- _meta.json
- references/SUPPLEMENTARY_DESIGN.md
- references/config/agents.yaml
- references/config/budget.yaml
- references/config/cleanup.yaml
- references/config/pipelines.yaml
- references/config/review.yaml
Open the folder on GitHubat commit e5199b5
Compare with similar skills
Task Dispatcher 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 |
|---|---|---|---|---|---|---|
| Task Dispatcher this skillLeoYeAI/openclaw-master-skills | 2.2k | — | ~2.8k | Automated safety check: Pass | MIT | |
| Gh Issuestrpc-group/trpc-agent-go | 1.9k | 8 repos | ~8.7k | Automated safety check: Pass | Apache-2.0 | |
| Autopilothanamizuki/solopreneur | 152 | — | ~6.8k | Automated safety check: Pass | MIT | |
| WebGPTNhahan/WebGPT | 120 | — | ~1.4k | Automated safety check: Pass | MIT | |
| Wp Plugin Developmentgambitph/Stackable | 351 | 3 repos | ~999 | Automated safety check: Pass | GPL-3.0 | |
| Self Improvementpskoett/pskoett-ai-skills | 314 | — | ~5k | Automated safety check: Pass | None |
Similar skills
Autopilot
hanamizuki/solopreneur
PR orchestration planner. An agent skill from hanamizuki/solopreneur.
WebGPT
Nhahan/WebGPT
Hands bounded tasks from Codex to a signed-in ChatGPT web session at a chosen reasoning level, or opens a terminal chat in your project that you control.
Wp Plugin Development
gambitph/Stackable
A skill your agent uses when developing WordPress plugins: architecture and hooks, activation/deactivation/uninstall, admin UI and Settings API, data storage, cron/tasks, security…
Self Improvement
pskoett/pskoett-ai-skills
Captures learnings, errors, corrections, and feature requests to enable continuous improvement.
Harness 24h
thu-nmrc/OpenHarness
A production-grade framework for long-running, autonomous agents based on Harness Engineering principles.
More from LeoYeAI/openclaw-master-skills
All 1,235 skills in this repoDevOps Pipeline Management
LeoYeAI/openclaw-master-skills
Manages pipelines on a DevOps quality and efficiency platform through its OpenAPI: list workspaces and templates, create, update, run and cancel pipelines, and read run records.
Feishu Document Collaboration
LeoYeAI/openclaw-master-skills
Patches OpenClaw's Feishu extension so an edited document triggers an isolated agent session that reads the doc and replies inline, turning it into a live chat space.
Files Memory System
LeoYeAI/openclaw-master-skills
Multi-context memory management system for OpenClaw agents with group-isolated storage, global shared memory, workspace organization, and group-specific skills isolation.
GEO-Claw AI Visibility Agent
LeoYeAI/openclaw-master-skills
Runs a brand's AI-search visibility work end to end: diagnosing how AI platforms represent it, repositioning it, producing AI-optimized content and monitoring ongoing mentions.
Google Workspace CLI
LeoYeAI/openclaw-master-skills
Installs and authenticates the gws CLI, then automates Gmail, Drive, Sheets, Calendar, Docs, Chat and Tasks with ready-made recipes, persona bundles and security audits.
HealthFit Health Advisors
LeoYeAI/openclaw-master-skills
Runs four advisor roles, a fitness coach, nutritionist, data analyst and TCM practitioner, to build a health profile and track workouts, diet and wellness over time.
Related
Categories
Questions about Task Dispatcher
What does Task Dispatcher do?
智能任务分发与子代理协调中枢。当用户提交任何任务时，执行需求分析、任务拆解、分发策略制定，分发给合适的 subagent 执行，监控进度并阶段汇报，最终汇总结果。失败时自动兜底处理。适用于：（1）用户直接下达的任务（2）cron/heartbeat 触发的任务（3）任何需要多步骤处理的工作。. Task Dispatcher is an agent skill from LeoYeAI/openclaw-master-skills.
When should I use Task Dispatcher?
Task Dispatcher fits situations like: tasks that involve Subagents; tasks that involve Scheduled and recurring tasks.
How do I install Task Dispatcher in Claude Code?
Run `npx skills add LeoYeAI/openclaw-master-skills --skill task-dispatcher -a claude-code`. Or copy the skill folder (skills/task-dispatcher in LeoYeAI/openclaw-master-skills) into .claude/skills/task-dispatcher in your project. Claude Code loads it when a task matches its description.
How do I install Task Dispatcher in Codex?
Run `npx skills add LeoYeAI/openclaw-master-skills --skill task-dispatcher -a codex`. Or copy the skill folder (skills/task-dispatcher in LeoYeAI/openclaw-master-skills) into .agents/skills/task-dispatcher in your project. Codex loads it when a task matches its description.
Can I use Task Dispatcher 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 LeoYeAI/openclaw-master-skills --skill task-dispatcher -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/task-dispatcher, .gemini/skills/task-dispatcher, .github/skills/task-dispatcher and .opencode/skills/task-dispatcher in your project.
What does Task Dispatcher need to run?
SKILL.md names no scripts, command-line tools or credentials: Task Dispatcher is instructions for the agent only.
Does Task Dispatcher access the network?
SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.
Is Task Dispatcher 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. Review the folder before installing.
What licence does Task Dispatcher use?
Task Dispatcher is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.
How many tokens does Task Dispatcher use?
About 2.8k 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. Its references folder adds about 19k tokens, read only when the agent opens those files.
What are the alternatives to Task Dispatcher?
Skills that share tags, products or a category with Task Dispatcher: Gh Issues (trpc-group/trpc-agent-go, 1.9k stars), Autopilot (hanamizuki/solopreneur, 152 stars), WebGPT (Nhahan/WebGPT, 120 stars) and Wp Plugin Development (gambitph/Stackable, 351 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.
Who maintains Task Dispatcher?
LeoYeAI (a GitHub user) maintains it in LeoYeAI/openclaw-master-skills, which has 2,160 GitHub stars. The repository holds 1,235 skills in this directory. The repository was last updated on July 20, 2026.
Source: LeoYeAI/openclaw-master-skills on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.