agent-orchestrator

Multi-agent orchestration coordinates diverse AI agents by automatically discovering available skills, evaluating capabilities, and generating execution plans to operate cohesively. It scans a central registry to match skills to your request.

Automatic skill discovery scans registries and folders to identify new skills. It evaluates their capabilities by parsing SKILL.md metadata, enabling the orchestrator to match relevant skills to user requests for multi-agent workflows.

Yes, you can compose multi-skill plans using flexible orchestration patterns including sequential, parallel, and primary-support. These patterns coordinate cross-skill workflows to execute complex automation tasks end-to-end.

Yes, pyyaml is required as a dependency. The system relies on it to parse SKILL.md metadata stored in YAML format within the central registry, which drives the auto-discovery and capability matching pipeline.

Provide a user query and the system will scan the registry, match relevant skills based on capabilities, and output an orchestration plan. This plan coordinates the selected agents using sequential, parallel, or primary-support patterns.

The orchestration system depends on a central registry of SKILL.md metadata for capability matching. If skills are not registered or metadata is missing, the auto-discovery pipeline cannot evaluate or include them in the execution plan.