Unverified50% confidenceFactExact time
Each sub-agent in Claude Code runs in its own context window with independent system prompts, tool access permissions, and permission settings, returning only a summary to the parent agent; built-in sub-agents include Explore, Plan, and General Purpose.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/9/2026
First Seen
Valid until: 1/7/2027
Sources
Related Entities
Related Claims
UnverifiedThe sub-agent mechanism in Claude Code can bypass the context window limit of a single conversation, allowing different sub-agents to focus on their respective responsibilities.76% similarUnverifiedClaude Code 在底层维护一套系统提示词定义 Agent 行为准则和工具使用规范，并通过权限沙箱机制限制 Agent 对文件系统的访问范围73% similarUnverifiedClaude Code's permission mechanism consists of three mutually independent yet combinable parts: the mode determines how actions are approved, rules specify which tool calls are allowed/asked about/denied, and the sandbox restricts the files and network that commands can actually access72% similarUnverifiedClaude Code's sub-agent implementation injects a dedicated knowledge base, behavioral norms, and tool-calling permissions into each agent through the CLAUDE.md configuration file.71% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/991588API
curl https://kongchang.com/api/v1/knowledge/claims/991588MCP
get_claim(id=991588)