← ClaudeAtlas

manus-crewlisted

CrewAI-like multi-agent workflows — sequential, hierarchical, and parallel execution modes. Agent roles/personas, human-in-the-loop checkpoints, Autogen SelectorGroupChat swarms, and inter-agent message bus.
jarbitechture/claude-skills · ★ 0 · AI & Automation · score 58
Install: claude install-skill jarbitechture/claude-skills
# Manus Crew Workflows CrewAI-like agent orchestration in `~/manus-chatbot/agents/`. ## Execution Modes | Mode | Description | Use Case | |------|-------------|----------| | Sequential | Agents run one after another | Pipeline processing | | Hierarchical | Manager delegates to workers | Complex projects | | Parallel | Agents run simultaneously | Independent tasks | ## Key Modules | Module | Purpose | |--------|---------| | `crew_features.py` | Roles, personas, execution modes, human-in-the-loop | | `smart_swarm.py` | Autogen SelectorGroupChat with multi-model routing | | `swarm_team.py` | Collaborative agent swarm with dynamic task routing | | `multi_agent_coordinator.py` | Inter-agent messaging, handoffs, intent-first routing | | `agent_message_bus.py` | Priority-based pub/sub for agent-to-agent communication | ## Human-in-the-Loop ```python from agents.crew_features import CrewWorkflow workflow = CrewWorkflow( agents=[researcher, writer, reviewer], mode="sequential", human_checkpoints=["after_research", "before_publish"] ) ``` ## Message Bus ```python from agents.agent_message_bus import AgentMessageBus bus = AgentMessageBus() bus.publish("research_complete", data=results, priority=1) bus.subscribe("research_complete", handler=writer_agent.process) ```