agent-factory

To spawn task-specific worker agents for parallel execution, you use a template-driven system that reads worker templates, substitutes placeholders with task data, and generates ephemeral SKILL.md files. This creates specialized workers across code, ui, integration, test, docs, and config tracks.

Template-driven worker agent orchestration is a mechanism that dynamically creates specialized ephemeral agents from templates. It substitutes placeholders with contextual task data and generates dispatch prompts to drive autonomous execution and coordination across multiple parallel tracks.

You create specialized worker agents from templates by providing a task id, name, type, and contextual details. The system reads your template, substitutes placeholders with the provided data, and writes an ephemeral SKILL.md file to define the worker's specific behavior.

Yes, worker templates support parallel task execution across code, ui, integration, test, docs, and config tracks. The orchestration workflow applies placeholder substitution for each track type and generates a dispatch prompt to coordinate autonomous execution.

The best way to orchestrate multiple autonomous worker agents is through an integrated base protocol that generates dispatch prompts. This approach reads worker templates, substitutes task data into placeholders, and writes ephemeral SKILL.md files to drive coordinated parallel execution.

Yes, you need predefined templates to dynamically spawn worker agents. The system relies on reading these templates and substituting placeholders with your task's id, name, type, and contextual details to generate the ephemeral SKILL.md files required for execution.