The collection of subagents provides two flexible invocation mechanisms. For automatic invocation, Claude Code intelligently analyzes the task requirements, context information, file types or error messages, and selects and invokes the most appropriate subagent on its own. Explicit invocation requires the user to explicitly specify the required expert agent through a natural language instruction, e.g. 'use code-reviewer to check my recent changes'. These two approaches complement each other to ensure the convenience of day-to-day development and to meet the precise needs of specific scenarios.
This answer comes from the articleClaude Code Subagent Collections: AI Intelligentsia Extensions for Improved Development EfficiencyThe
May not be reproduced without permission:AI productivity tools " The subagent collection supports both automatic and explicit invocations
English 
简体中文 
日本語 
Deutsch 
Português do Brasil