Solo CEO

一人公司主 Agent 技能。当用户希望 AI 作为"主控 CEO Agent"协调多个预建的长期员工 Agent 完成任务时触发。适用于：任务拆分与分发、多 Agent 协作对话、模拟真实公司沟通流程（最多5轮对话）、最终汇总报告给用户。核心能力：CEO Agent 理解任务、将任务分发给员工 Agents、收...

Before installing or running this skill: - Understand that SKILL.md expects the agent to read and write files under ~/.openclaw/workspace-<agentId>/ and to require agentToAgent=true in openclaw.json. The registry metadata does not declare those paths — treat that as an omission. - If you will store any sensitive data in tasks, avoid letting employees write it into MEMORY.md; treat MEMORY.md as persistent and review its contents regularly. - Only enable agentToAgent if you trust the agents and runtime environment; enabling it opens inter-agent messaging and persistent workspace I/O. - Consider running the skill in a sandboxed account or limited environment so workspace files are isolated, and inspect any files it creates (SOUL.md, MEMORY.md, plan.md). - Ask the publisher (or require a metadata update) to explicitly declare required config paths and file access in the registry entry, and to add guidance on secret handling and sanitization. - If you need a lower-risk alternative, use a coordinator that does not persist data to the host filesystem or that documents exactly what it will read/write.

Type: OpenClaw Skill Name: solo-ceo Version: 2.1.0 The skill bundle defines a coordination framework for a 'CEO' agent to manage sub-agents (coder, operator) within the OpenClaw environment. It utilizes standard framework tools like `sessions_spawn` and `sessions_yield` for task distribution and parallel execution. The logic is focused on task decomposition, planning, and reporting, with no evidence of malicious intent, data exfiltration, or unauthorized system access across the provided files (SKILL.md, _meta.json, and templates).