What problem does it solve?
Subagents enable you to isolate specialized AI tasks in separate, reusable agents with dedicated prompts, improving modularity and cross-project consistency.
Core Features & Use Cases
- Frontmatter-driven identity: Each subagent is defined by a YAML frontmatter containing a unique name and description, and stores its system prompt in the Markdown body.
- Location-based scoping: Supports project-level subagents under .cursor/agents/ and user-level subagents under ~/.cursor/agents/ for personal reuse.
- Extensible formats: Works with a Markdown file per subagent and can be extended with example configurations for common tasks (e.g., code review, debugging, data processing).
- Version control friendly: Subagents can be checked into version control to share with teams.
- Testable and reusable: Designed to be easily tested and reused across workflows.
Quick Start
Create a new subagent by placing a Markdown file with a frontmatter name and description in the appropriate agents directory and write the system prompt in the body.