beam-agent-manager

Beam agent graph lifecycle operations are automated by routing creation, modification, optimization, and publishing tasks through specialized sub-skills. This orchestration enforces consistency across graph state changes while reducing manual coordination for YAML spec deploys and structural updates.

Yes, you can deploy a Beam agent graph using a YAML spec or a simple POST payload. The orchestration processes the spec to create, update, or publish the agent graph, handling structural changes and prompt, model, and parameter updates automatically.

APE optimization for agent graphs is an automated process that improves agent performance. The orchestrator routes the optimization task to a specialized sub-skill, applying the optimization while enforcing end-to-end lifecycle routing and ensuring consistency across graph state changes.

Updating prompts, models, and parameters in a deployed agent graph is handled by routing the modification request to a specialized sub-skill. This ensures structural changes and parameter updates maintain consistency across the entire agent graph state.

No, manual routing is not required when modifying a Beam agent graph. The orchestration automatically routes operations into appropriate sub-skills for creation, updates, optimization, and publishing, ensuring consistent end-to-end lifecycle management.

Spec-based deploy for agent graph orchestration should be used when building new graphs or applying structural changes. Providing a YAML spec or POST payload allows the orchestrator to automate end-to-end lifecycle routing, from creation through APE optimization to publishing.