agent-factory

Dynamically create specialized worker agents by applying template-driven selection and placeholder substitution to categorize tasks by type (code, ui, integration, test, docs, config) for parallel execution. This approach enables on-demand worker dispatch tailored to specific workloads.

Orchestrate ephemeral worker agents in a DAG-based environment by establishing a complete lifecycle management process: create, execute, and cleanup. This minimizes long-lived side effects while safely coordinating parallel dispatch and recovery via a message bus.

Message bus monitoring tracks dispatched worker agents by facilitating safe parallel coordination and recovery during task execution. The message bus serves as the central channel for observing worker status, enabling dynamic cleanup of ephemeral artifacts upon completion.

Yes, template-based specialization supports diverse workloads like code, ui, integration, test, docs, and config. The system selects a tailored template for the specific task type and applies placeholder substitution to fit the exact requirements before dispatching the worker.

No, you do not need to manually clean up ephemeral skill artifacts after parallel task execution. The lifecycle management process automates the cleanup of these ephemeral artifacts to minimize long-lived side effects and maintain resource efficiency.