Openclaw Multi Agent Orchestrator

OpenClaw 多 Agent 协同工具 - 管理和协调多个 AI Agent。 支持 Agent 注册、任务分发、负载均衡、性能监控、Agent 间通信。 适用于复杂任务分解、多角色协作、企业级应用。 关键词：openclaw, multi-agent, orchestration, automation,...

This package appears to implement a local multi-agent orchestrator and does not request credentials or perform network I/O. Before installing or running it: 1) Note the documentation wrongly claims 'read-only' access — the scripts will create and modify JSON files (agents, messages, collaborations) in the skill's config directory (typically under ~/.openclaw/workspace/skills/...). 2) Inspect the two large scripts (scripts/orchestrator.py and scripts/communication.py) yourself — they contain the state persistence and messaging logic; the smaller scripts are mostly stubs. 3) Run it as a non-privileged user and in a sandbox or test environment first; back up any existing ~/.openclaw workspace you care about. 4) If you need strict auditability, verify there are no hidden network calls by running the scripts in an offline environment or using network monitoring. 5) Do not provide secrets to these scripts; they do not need API keys. If you want higher assurance, ask the author for a provenance statement or run the code through a local security review.

Type: OpenClaw Skill Name: openclaw-multi-agent-orchestrator Version: 0.1.0 The skill bundle provides a functional framework for managing multiple AI agents and their communications locally. The core logic in scripts/orchestrator.py and scripts/communication.py implements task routing, load balancing, and messaging using local JSON files for state persistence. No network activity, unauthorized file access, or malicious execution patterns were detected, and the instructions in SKILL.md are consistent with the provided code.