agent-creator

You can generate a sub-agent configuration by providing a user description of the desired agent. The system outputs a ready-to-save Markdown file containing YAML frontmatter and a structured body detailing its purpose and instructions.

A Markdown sub-agent configuration includes YAML frontmatter with standardized fields like name, description, tools, model, and color, followed by a descriptive body detailing the agent's purpose and workflow instructions.

You define tools and models through a guided requirements gathering workflow. By providing your agent's purpose and constraints, the system automatically maps them to a standardized frontmatter schema for your configuration.

Yes, product teams can use this workflow to rapidly prototype sub-agents for automation pipelines. For example, you can quickly generate a data-cleaner agent definition to handle data wrangling tasks without manual drafting.

The best way to ensure consistent agent definitions is automating creation with a standardized frontmatter schema. This prevents manual drafting errors and outputs structured Markdown files for every sub-agent.

The generated sub-agent configuration files are saved as Markdown files in the ai/shared/agents/ directory. The system outputs a ready-to-save file named after your specified agent.