← Back to Skills Marketplace
Agent Orchestrator
by
yuyonghao-123
· GitHub ↗
· v0.1.0
· MIT-0
144
Downloads
0
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install yuyonghao-agent-orchestrator
Description
生产级 Agent 编排器,实现Agent生命周期管理、任务调度、资源限制、服务发现及多节点集群管理功能。
Usage Guidance
This package appears to implement the orchestrator described, but treat it as untrusted code until you verify and fix issues. Recommended steps before installing or running in production:
- Run npm test and node src/index.js inside a sandbox or VM to observe runtime errors. The scheduler currently imports uuid incorrectly (scheduler.js expects a v4 export but utils.js provides uuidv4), which will cause failures — fix the import or export (e.g., change scheduler to const { uuidv4 } = require('./utils') or export v4 from utils).
- Search the remaining (truncated) parts of orchestrator.js for networking (HTTP/RPC) or code that opens sockets or calls external endpoints; cluster features often add network behavior and should be audited.
- Check for any uses of child_process, fs reads of system config, or process.env access in the omitted parts; these would raise additional concerns.
- If you plan to enable cluster mode, review the leader election and event broadcast implementations to confirm they don't leak data or open unexpected ports.
- Run the code with limited privileges, in an isolated environment, and inspect logs/events before trusting it in production.
- If you are not comfortable auditing or fixing the code, ask the author for a corrected release (fixing the uuid import/export and any missing implementations) or prefer an established orchestrator project with active maintainers.
Capability Assessment
Purpose & Capability
The skill name/description (Agent Orchestrator) matches the included code and SKILL.md examples: lifecycle management, scheduling, health checks, load balancing, resource management and cluster features are implemented across files. Required env vars / binaries / config paths are not requested and none appear in the code, which is proportionate to the stated purpose.
Instruction Scope
The SKILL.md instructions are limited and appropriate (npm install, example usage, npm test). However, there are internal code inconsistencies that will cause runtime failures if you follow the instructions as-is: scheduler.js imports { v4: uuidv4 } from './utils' but utils.js exports uuidv4 (not a v4 property), so uuid generation will be undefined and scheduler.submit will throw at runtime. Also part of orchestrator.js shown is truncated and references private helpers like _broadcastEvent; the provided snippet references several internal functions/events whose implementations should be verified (no network endpoints are obvious in the visible code, but cluster features may introduce networking in omitted sections).
Install Mechanism
There is no install spec in the registry metadata; SKILL.md suggests running npm install and package.json is included. package.json has no third-party dependencies listed and scripts are basic (start/test). No external download URLs, tarballs, or untrusted installers are used — low install risk. Still run npm install in a sandbox before production use.
Credentials
The skill declares no required environment variables or credentials and the code does not read process.env or other system credential/config paths in the visible files. Requested privileges are proportional to the stated functionality.
Persistence & Privilege
The skill does not request 'always: true' or elevated persistence. It does not modify other skills or system settings in the visible code. Cluster mode exists as an option but enabling it appears to be an explicit runtime configuration, not automatic persistence.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install yuyonghao-agent-orchestrator - After installation, invoke the skill by name or use
/yuyonghao-agent-orchestrator - Provide required inputs per the skill's parameter spec and get structured output
Version History
v0.1.0
Initial release of Agent Orchestrator.
- Provides agent lifecycle management with health checks, graceful shutdown, and auto-recovery.
- Supports advanced task scheduling with multiple strategies and priority levels.
- Includes resource management: CPU/memory limits, concurrency control, quotas, and monitoring.
- Features dynamic service discovery and ability-based routing.
- Offers cluster management: multi-node, leader election, state sync, and configuration tooling.
- Easy configuration and usage via simple JavaScript API.
Metadata
Frequently Asked Questions
What is Agent Orchestrator?
生产级 Agent 编排器,实现Agent生命周期管理、任务调度、资源限制、服务发现及多节点集群管理功能。 It is an AI Agent Skill for Claude Code / OpenClaw, with 144 downloads so far.
How do I install Agent Orchestrator?
Run "/install yuyonghao-agent-orchestrator" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is Agent Orchestrator free?
Yes, Agent Orchestrator is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does Agent Orchestrator support?
Agent Orchestrator is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created Agent Orchestrator?
It is built and maintained by yuyonghao-123 (@yuyonghao-123); the current version is v0.1.0.
More Skills