← 返回 Skills 市场
13875
总下载
32
收藏
130
当前安装
1
版本数
在 OpenClaw 中安装
/install agent-orchestrator
功能描述
Meta-agent skill for orchestrating complex tasks through autonomous sub-agents. Decomposes macro tasks into subtasks, spawns specialized sub-agents with dynamically generated SKILL.md files, coordinates file-based communication, consolidates results, and dissolves agents upon completion.
MANDATORY TRIGGERS: orchestrate, multi-agent, decompose task, spawn agents, sub-agents, parallel agents, agent coordination, task breakdown, meta-agent, agent factory, delegate tasks
安全使用建议
Install only if you intentionally want a skill that can coordinate multiple autonomous sub-agents. Before using it, set explicit limits for how many agents may run, which tools they may use, which directories they may read or write, whether Bash and edits require approval, and when generated workspaces should be archived or deleted.
功能分析
Type: OpenClaw Skill
Name: agent-orchestrator
Version: 0.1.0
The skill is classified as suspicious due to its powerful capabilities that introduce significant security risks. The `SKILL.md` for the orchestrator describes dynamically generating `SKILL.md` and `inbox/instructions.md` for sub-agents. Crucially, the `references/sub-agent-templates.md` file reveals that several sub-agent types (e.g., 'Code Agent', 'Analysis Agent', 'Integration Agent') are explicitly granted `Bash` shell access. This combination creates a high-risk prompt injection vector, where a maliciously crafted macro task could instruct the orchestrator to generate sub-agent skills that execute arbitrary commands, leading to potential unauthorized system access or data manipulation.
能力评估
Purpose & Capability
The stated purpose coherently matches the artifacts: decompose work, generate sub-agent skill files, spawn agents, coordinate inbox/outbox files, and consolidate results. The concern is that this is inherently high-impact authority, including autonomous sub-agents and templates that can include Bash, web access, and file read/write/edit capabilities.
Instruction Scope
The instructions do not require explicit confirmation before creating workspaces, writing generated SKILL.md and instructions files, spawning agents, archiving, or cleanup. They also lack clear limits for agent count, runtime, allowed paths, allowed tools, cancellation, or review before passing one agent's output to another.
Install Mechanism
The package contains only text skill and reference files, with no executable installer or bundled script. However, the workflow references scripts/create_agent.py and scripts/dissolve_agents.py even though those scripts are not present in the reviewed artifacts.
Credentials
File operations, Task-based sub-agent spawning, web research, and Bash can be purpose-aligned for orchestration, but the templates describe broad capabilities without scoped path boundaries or per-task containment.
Persistence & Privilege
The design creates persistent per-agent workspaces, status files, optional archives, and autonomous agents that run until completion, but it does not define retention, cleanup boundaries, stop controls, or privilege limits.
如何使用
- 确保已安装 OpenClaw（本地或 Docker 部署）
- 在对话框中输入安装命令：
/install agent-orchestrator - 安装完成后，直接呼叫该 Skill 的名称或使用
/agent-orchestrator触发 - 根据 Skill 的参数说明提供必要输入，即可获得结构化输出
版本历史
v0.1.0
Initial release of agent-orchestrator skill.
- Enables orchestration of complex tasks by spawning autonomous sub-agents for parallelized work.
- Dynamically generates SKILL.md files for specialized sub-agents, detailing roles and protocols.
- Implements a file-based communication system (inbox/outbox/status) for coordination and state tracking.
- Provides a step-by-step workflow: task decomposition, agent creation, dispatch, monitoring, consolidation, and dissolution.
- Includes templates and best practices for defining and managing sub-agent workflows.
元数据
常见问题
Agent Orchestrator 是什么？
Meta-agent skill for orchestrating complex tasks through autonomous sub-agents. Decomposes macro tasks into subtasks, spawns specialized sub-agents with dynamically generated SKILL.md files, coordinates file-based communication, consolidates results, and dissolves agents upon completion. MANDATORY TRIGGERS: orchestrate, multi-agent, decompose task, spawn agents, sub-agents, parallel agents, agent coordination, task breakdown, meta-agent, agent factory, delegate tasks. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 13875 次。
如何安装 Agent Orchestrator？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install agent-orchestrator」即可一键安装，无需额外配置。
Agent Orchestrator 是免费的吗？
是的，Agent Orchestrator 完全免费（开源免费），可自由下载、安装和使用。
Agent Orchestrator 支持哪些平台？
Agent Orchestrator 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（cross-platform）。
谁开发了 Agent Orchestrator？
由 aatmaan1（@aatmaan1）开发并维护，当前版本 v0.1.0。
推荐 Skills