subagent-orchestrator

Orchestrate OpenClaw subagents for continuous autopilot workflows. Use when managing UI/代码/验证/档案/侦察 task lines, continuing multi-round work without waiting f...

What to consider before installing: - This skill will autonomously read and update AGENT_TASK_BOARD.md and write durable records into a memory/ folder; it will also delete temporary artifacts and spawn further subagents until a stop condition. Make sure you are comfortable with an agent making those workspace file changes automatically. - The metadata does not declare required config paths even though the instructions reference specific files; ensure the task-board and memory locations are where you expect and back them up before use. - Test in a sandbox or with a small non-critical project first to verify behavior and avoid runaway loops. Establish user-facing pause/stop controls and monitor the first runs. - Because it makes persistent file changes, limit the skill to projects where automated edits are acceptable and avoid giving unrelated system credentials (none are requested by the skill).

Type: OpenClaw Skill Name: subagent-orchestrator Version: 1.0.0 The skill bundle is a legitimate orchestration framework for managing multi-agent 'autopilot' workflows. It defines a structured communication protocol (SUBAGENT_RECORD_PROTOCOL.md) and state management rules (SKILL.md) using local markdown files like AGENT_TASK_BOARD.md and memory logs. No evidence of data exfiltration, malicious execution, or harmful prompt injection was found; the instructions focus entirely on coordinating task handoffs and maintaining workflow continuity.