Subagents have arrived in Gemini CLI (developers.googleblog.com)

Gemini CLI has introduced a powerful feature called subagents, which enables the platform to delegate complex or repetitive tasks to specialized expert agents. Each subagent operates within its own context window, complete with custom system instructions and tailored tools, ensuring the main session remains efficient and focused on overarching goals. This delegation allows for the orchestration of tasks, with Gemini CLI acting as the strategic manager, significantly enhancing productivity and streamlining workflows. The significance of this update lies in empowering users to create custom subagents for specific projects or workflows using simple Markdown files with YAML frontmatter. Users can define their own expert agents, enhancing automation in areas like coding standards, project-specific personas, or even bundling them as extensions. Gemini CLI supports parallel processing, allowing users to run multiple subagents simultaneously, which can drastically reduce task completion time. However, caution is advised with parallel tasks that involve significant code changes to prevent conflicts. With built-in subagents and easy configuration, Gemini CLI is poised to make substantial improvements to productivity and collaboration within the AI/ML community.