create-subagent

To create custom AI subagents for Cursor workflows, you add a markdown file with YAML frontmatter defining the agent's name and description, then use the body as the system prompt. This organizes task-specific automation with reusable configurations.

A Cursor subagent is a focused AI assistant running in an isolated context with a tailored system prompt. You need one when delegating specialized tasks like code review, debugging, or domain-specific data processing across projects.

Yes, a Cursor subagent requires YAML frontmatter to satisfy configuration requirements. The frontmatter must include specific fields like name and description to define the agent before the markdown body acts as its system prompt.

Yes, you can load external resources into Cursor subagents. The configuration supports optional directories for scripts, references, and assets, allowing the agent to access reusable components and loadable resources during task execution.

The best way to configure a reusable code reviewer in Cursor is to create a project-level or user-level subagent. You define its code review workflow using a markdown body as the system prompt with YAML frontmatter.

No, Cursor subagents are not limited to project-level configurations. You can create either project-level or user-level subagents, enabling domain-focused assistants with reusable configurations that operate proactively across different environments.