Claude Code v2.1.172 Enables Subagents to Spawn Their Own Subagents
Anthropic's Claude Code platform introduced nested subagent spawning in version 2.1.172, allowing subagents to launch their own subagents up to five levels deep. Previously, only the main Claude thread could orchestrate subagents, requiring all pipeline coordination to happen at the top level. The feature enables more modular AI pipelines, though version 2.1.217 disabled nested spawning by default, requiring developers to explicitly set the CLAUDE_CODE_MAX_SUBAGENT_SPAWN_DEPTH environment variable to re-enable it. Additional guardrails introduced across subsequent versions include a per-session cap of 200 subagents and a concurrency limit of 20. One key limitation remains: nested subagents cannot interact directly with users, meaning any approval gates or human-in-the-loop steps must still be handled at the main Claude thread level.
This is an AI-generated summary. ShortSingh links to the original source for the complete article.
Discussion (0)
Log in to join the discussion and vote.
Log in