agent-smith

To architect a multi-agent system, you provide a complete blueprint specifying hierarchical, parallel, or sequential topologies. The skill coordinates this design using frontmatter-defined configuration to enforce architecture patterns for production-grade agentOS workflows.

Semantic routing in multi-agent orchestration directs tasks between agents based on meaning rather than fixed rules. This skill coordinates the design of a semantic routing layer alongside tool schemas to enable modular communication across engineering domains.

Planning memory architectures for multi-agent systems involves configuring short-term, long-term, and episodic memory layers. This skill coordinates memory architecture planning through frontmatter-defined configuration to structure how agents retain and retrieve context.

Yes, you can set up evaluation metrics for multi-agent coordination. The skill coordinates the evaluation of multi-agent systems by enforcing an evaluation framework within its frontmatter-defined configuration to measure production-grade agentOS performance.

The best way to design tool schemas for a multi-agent system is to define them within a complete architectural blueprint. This skill coordinates tool schema creation alongside semantic routing layer design to ensure proper agent interaction.

You should use hierarchical topologies when coordinating multi-agent systems requires layered control, and sequential topologies for linear step-by-step processing. This skill coordinates architectural planning across both patterns to match your engineering domain requirements.